From 61269e6ea593d05666fe23b217bb1d7284c2e0e5fc8b72d2e83f6855768ff402 Mon Sep 17 00:00:00 2001 From: Marcus Meissner Date: Wed, 23 Nov 2022 12:58:03 +0000 Subject: [PATCH] OBS-URL: https://build.opensuse.org/package/show/Emulators/ppsspp?expand=0&rev=362 --- _servicedata | 2 +- ppsspp-1.12.3~git20211114.tar.xz | 3 - ppsspp-1.13.2~git20221119.tar.xz | 3 + ppsspp.changes | 5177 ++++++++++++++++++++++++++++++ ppsspp.spec | 4 +- 5 files changed, 5183 insertions(+), 6 deletions(-) delete mode 100644 ppsspp-1.12.3~git20211114.tar.xz create mode 100644 ppsspp-1.13.2~git20221119.tar.xz diff --git a/_servicedata b/_servicedata index ee99589..72e6153 100644 --- a/_servicedata +++ b/_servicedata @@ -1,4 +1,4 @@ https://github.com/hrydgard/ppsspp.git - 42bc9066eeae5f805ff24377607679930fed1b61 \ No newline at end of file + 6796d3de8982518307f4b7bd3677bd809f61129f \ No newline at end of file diff --git a/ppsspp-1.12.3~git20211114.tar.xz b/ppsspp-1.12.3~git20211114.tar.xz deleted file mode 100644 index 3174d1a..0000000 --- a/ppsspp-1.12.3~git20211114.tar.xz +++ /dev/null @@ -1,3 +0,0 @@ -version https://git-lfs.github.com/spec/v1 -oid sha256:7614a611ba3936909231cfbdc70c9d3e02aea435481f4ab8f51f2459c54b36f0 -size 229712468 diff --git a/ppsspp-1.13.2~git20221119.tar.xz b/ppsspp-1.13.2~git20221119.tar.xz new file mode 100644 index 0000000..50832b3 --- /dev/null +++ b/ppsspp-1.13.2~git20221119.tar.xz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2c6c065fcf1f3653d0588577e9f19e567fd3c8aa1465b9baefb69f0a64f20acf +size 233300276 diff --git a/ppsspp.changes b/ppsspp.changes index ebabb4c..f6a7342 100644 --- a/ppsspp.changes +++ b/ppsspp.changes @@ -1,3 +1,5180 @@ +------------------------------------------------------------------- +Wed Nov 23 12:51:43 UTC 2022 - meissner@suse.com + +- Update to version 1.13.2~git20221123: + * Test fix + * Cleanup + * Wording tweaks, add scroll view + * Move post-processing settings to DisplayLayoutScreen + * Enable game backgrounds on more screens + * OpenXR - Aspect ratio of menu background fixed + * Rework the display layout editor to ditch the preview thing. + * OpenXR - 2D stereo condition fixed + * Add transparent background to the existing Display Layout Editor. + * Pause menu background: Remove complicated transparency tricks, instead draw game as part of background. + * Kernel: Respect partition param in heap funcs. + * D3D11: Remove unused hack USE flag. + * Move AsyncImageFileView to Common, which required some more shuffling around of stuff. + * Extract the Vulkan descriptor binding cleanup from #16345 + * Test crashfix + * Centralize ClearCacheNextFrame + * Libretro fixes + * Only check for display resized in emuscreen-under-pause-screen, not the other changes. + * Some minor "centralization" + * Split up the GPU config change notifications + * One more refactoring step + * Refactor: Split up updating the display and render size. + * Forgot to commit this in PR #16404 + * softgpu: Fix compile hazard while running. + * softgpu: Correct WX-exclusive platform hazards. + * jit: Add more reasonable estimates for RX protect. + * jit: Protect against write over-estimates. + * Fix resizing issue. Took some refactoring. + * Make the pause screen "transparent" + * OpenXR - 3rd party controller support fixed + * softgpu: Throw away all inverted polygons. + * GE Debugger: Correct texture address corruption. + * Orient Save State / Load State buttons horizontally + * Debugger: Combine mem info on duplicates. + * irjit: Consistently check vec4 safety. + * irjit: Allow VV2Op SIMD with exact overlap. + * irjit: Correct VV2Op SIMD check. + * GPU: Prefer raster for depth blit if supported. + * Typo fixed + * OpenXR - Cursor aspect ratio + * OpenXR - Accept scrolling from both controllers + * OpenXR - Camera adjust issues resolved + * OpenXR - Force PSP menu aspect ratio + * OpenXR - Better dialog support + * OpenXR - Allow higher 2D canvas distance + * OpenXR - Use axis integration for camera adjust only + * OpenXR - Use mouse control only + * OpenXR - Support axis + * OpenXR - Support camera adjust using joystick + * Update zh_TW.ini + * Update zh_TW.ini + * Update zh_TW.ini + * interp: Support memory breakpoints too. + * interp: Centralize memory size handling. + * interp: Allow breakpoints in release mode. + * interp: Allow resume from breakpoint. + * Debugger: Keep flag for any breakpoints. + * Twinbee Portable: Add config flag to avoid the system languages the game doesn't work with + * softgpu: Use threads on self-render if safe. + * UI: Show standard back button on touch layout. + * softgpu: Fix off-by-one rendering after half-pixel. + * softgpu: Avoid splitting rectangles for fog. + * softgpu: Correct tex on fogged rectangles. + * GPU: Automatically reduce depth range on == test. + * GPU: Remove unused GPU_USE flag. + * GPU: Centralize more GPU_USE flags, like depth. + * GPU: Add a small error-compensation to depth clip. + * Vulkan: Mark Adreno bug resolved on recent driver. + * Vulkan: Correct alpha in Adreno workaround. + * GPU: Convert alphamask to FF/00 for 5551. + * UI: Avoid enum/float comparisons. + * VR: Remove unnecessary deprecated header. + * SaveState: Switch from deprecated is_pod. + * GPU: Correct Marvel copy hook size check. + * GPU: Hook US version of Marvel Alliance upload. + * OpenXR - Freezing on Pico 4 fixed + * OpenXR - Test mirroring over 30 various games + * OpenXR - Get all mirroring variants + * Revert "OpenXR - Disable range culling properly" + * GPU: Correct equal depth checks. + * OpenXR - View matrix identity check breaking several games fixed + * OpenXR - Don't get mirroring from identity matrix + * OpenXR - Do not apply head rotation on identity matrix + * GPU: Correct vertex decoder in software transform. + * Mpeg: Correct YUV order from decode. + * Mpeg: Fix ConvertToYuv420 return value. + * GPU: Avoid enabling depth test pointlessly. + * GPU: Prefer scaling depth to 16-bit if using 24. + * GPU: Avoid clears for non-simple depth values. + * TexCache: Cleanup BGRA flag dirtying. + * Fix GLSL compilation issues on really old Adreno drivers. + * GPU: Avoid self memcpy(). + * OpenGL ES: Crash as early as possible if things are bad + * Update zh_TW.ini + * Space added + * TexCache: Correct 16->32 for CLUT4 with shift. + * TexCache: Fix 16->32 colors with CLUT start pos. + * TexCache: Align expandClut buffer, cleanup sizes. + * Restore parallel GLSL builds + * Lift length limit on Android logging + * Log source code for failed pipeline creations + * Avoid drawing with failed-to-create pipelines + * Make Mailbox handle null values. Fixes hang. + * Output shader descriptions directly in the source code + * Improve VkResult logging + * Warning fix in VR code + * OpenXR - Disable range culling properly + * OpenXR - Allow more extreme camera adjusts + * OpenXR - GTA Liberty city stereo fixed + * Properly buildfix GL/VR again + * OpenXR - Camera sliders removed from settings + * Fix sky plane fix (broke in my last VR refactor) + * OpenXR - Camera adjustment using PSP keys + * OpenXR - Keep status of PSP keys + * Restore assert + * Experiment (Vulkan): Temporarily disable GLSL parallel compilation on Android. For #16341 investigation. + * Disable remaining false positive. + * Turn off boundSecondaryIsInputAttachment_ when we don't have one. + * Vulkan: Use the very same view as input attachment and color attachment, not just the same image + * Update zh_TW.ini + * Update zh_CN.ini + * Remove unnecessary assert + * OpenXR - Move configs out of VRRenderer + * OpenXR - Move matrix calculation out of VRRenderer + * OpenXR - Simplify projection matrix creation + * OpenXR - Matrix indexing fixed + * OpenXR - Get rid of separate projection matrices + * OpenXR - Get rid of ovrMatrix4f structure + * Partially rescue translation by copying from Hack Settings + * Setting should be reported + * Blind libretro fix + * Android fix + * Blind Qt fix + * lang: Remove obsolete keys, add new untranslated ones :( + * Change "Simulate block transfer" to "Skip GPU Readbacks". Group the speed hacks together + * Change "Rendering Mode" to just a "Skip buffer effects" checkbox. Reuse translations. + * libretro fix, more games + * GPU: Correct softgpu curve issue. + * GPU: Always skin in decode for software transform. + * softgpu: Always use software skinning. + * GPU: Purify vertTypeIsSkinningEnabled(). + * Windows ARM/ARM64 buildfix + * Remove now-unused lang keys + * Change the "Retain changed textures" option into a compat.ini option. + * GPU: Use skinned position always in bounding check. + * GPU: Cleanup GetVertTypeID() usage. + * Linking fix for normal Android devices + * UWP build fix. Had to reshuffle some stuff and compile some Vulkan files. + * Add a dynamic loader for OpenXR so the VR code can link. Make it all build on Windows. + * Typo fix + * Make a lot more code VR build for all platforms + * Fix type of predictedDisplayTime + * 6dof fix + * Rename 6DOF_PRECISE to HAS_UNIT_SCALE + * Remove ugly float conversion hacks from VR config + * Finish previous buildfix... + * Upgrade gradle, specify NDK version in android/build.gradle + * Yet another gradle upgrade + * Comment fixes (feedback) + * Fix parsing ampersands in server listings for streaming + * Drop some D3D11 screen rotation support (was only used on Windows Phone) + * VK/D3D11: Move the rarely used "u_rotation" uniform to the frame uniform buffer. + * Switch RenderPassType to be a "proper" bitfield enum. + * Support Android file intents with up to 3 dots in path + * Code formatting fixed + * OpenXR - Get platform flags in runtime + * OpenXR - Introduce platform flags + * OpenXR - Fix ifdef ANDROID + * OpenXR - Add an option to disable 72Hz update + * OpenXR - Enable camera movement and stereo in flatscreen mode + * [Libretro] Update core options to v2 + add a few new ones + * Build: Use Makefile for libretro. + * Build: Enable libzip on libretro. + * GE Debugger: Respect unchanged VRAM each frame. + * GE Debugger: Try harder to identify unchanged VRAM. + * GE Debugger: Dump textures only if possibly used. + * GE Debugger: Restrict marking of dirty VRAM better. + * Vulkan: Fix descriptor set binding in geometry shader generator. Should fix #16315 + * OpenXR - Settings code cleanup + * Replacement: Avoid issues with c++14 data(). + * UI: Install textures as a zip if supported. + * OpenXR - Enable stereo for Split/Second + * OpenXR - Workaround for not working inputbox + * Memory initialization fixed + * OpenXR - Let users set the movement length + * OpenXR - Option to disable controller mapping + * OpenXR - Map controller motions on keys + * irjit: Correct prefix validation. + * interp: Correct dprefix accuracy for vrot. + * irjit: Fix unordered float compares. + * Replacement: Don't double count memory usage. + * Replacement: Allow use of textures.zip for Android. + * Replacement: Lock around shared data access. + * Replacement: Cache file data across cache keys. + * Replacement: Log only missing explicit filenames. + * Replacement: Decimate aggressively with high usage. + * Replacement: Reduce IO checks on startup. + * Replacement: Avoid needless exists check per tex. + * interp: Correct vscl/vmscl t prefix handling. + * irjit: Handle vrot overlap more correctly. + * SaveState: Restore replacements in only one place. + * Debugger: Allow currently-invalid memory reference. + * Debugger: Fix input.analog.send validation. + * UWP: Avoid some float conversion warnings. + * Windows: Avoid compile param conflict warning. + * Draw: Correct ambiguous comparison overload. + * typo fixed + * OpenXR - Stereo support for Outrun 2006 added + * OpenXR - Add new virtual key for camera adjust + * OpenXR - Camera side adjust added + * OpenXR - Camera height adjust added + * OpenXR - Do not process key actions during camera adjust + * Vulkan: Log additional queue command. + * armips: Update to latest. + * headless: Update tests. + * Vulkan: Fixes validation errors with "null" textures + * Address feedback + * OpenXR - Allow updating camera in realtime + * OpenXR - Make VR camera parameters floating numbers + * Need to request the proper type of uniform buffer descriptors. + * Add some extra checking for valid stereo shader. + * Actually bind a global uniform buffer, too. Not yet used. + * Vulkan: Reserve descriptor set 0 for frame-global data, move everything else to set 1 + * jit: Reduce some include pollution. + * Update zh_TW.ini + * Update zh_TW.ini + * Fix UB + * Format string fix + * Shadergen fix for OpenGL + * Crashfix in settings menu (Android only, for whatever reason) + * Fix validation issue with clears, remove redundant code + * Clean up GPU_Vulkan::BuildReportingInfo a bit + * Improve compatibility checks for stereo rendering. + * Fix for stereo with textureProj (emulated projection). + * Fix depal texture binding bug + * Name more image views. Very useful with a little patch to the Vulkan debug layers. + * Fix rendering in non-stereo mode + * Pass shadergen tests + * Address feedback + * Stereo now works through shader depal too + * Fix more cases, GTA works now. + * Use arrays more consistently. + * Always use array textures for framebuffers in Vulkan for simplicity. + * wip + * Minor shaderwriter refactor + * Stereo on mobile fixes + * Some refactoring of framebuffer views, layer issues, more work. + * More work. Things are starting to work now. + * Add a couple of stereo display shaders, add infrastructure, compile stereo pipelines + * Non-stereo bugfix + * Start stereo postshader work + * More multiview work + * Misc multiview hackery + * thin3d: Replace hint at future MRT support with basic multi layer support + * Update zh_TW.ini + * OpenXR - Lego Star Wars III stereo support added + * OpenXR - 6DoF and stereo in Lego Star Wars II fixed + * Build: Fix Android/macOS ccache. + * Build: Update deprecated actions. + * headless: Update tests. + * GE Debugger: Correct UV display with prescale. + * Minor ShaderWriter refactor, prep for later changes + * GE Debugger: Allow reading tex on D3D11. + * Vulkan: Use v2 feature checks, and add check for multiview features. + * Testfix + * Minor refactor of ShaderWriter flags. Extracted from the stereo PR + * softgpu: Better approximate slope mip level mode (#16276) + * x86jit: Correct vh2f NAN handling (#16275) + * headless: Update tests. + * VR: Always pass the bounding box check. + * GPU: Account for scissor/viewport in bound test. + * GPU: Respect depth clamp in bounding box check. + * GPU: Allow/use indices in bounding box check. + * GPU: Correct large morph vertex advance. + * GPU: Consume cycles in bounding box check. + * GPU: Correct bounding box for larger counts. + * GE Debugger: List verts on bounding box command. + * Make it possible to specify extensions for Vulkan in the ShaderWriter + * Remove duplicate line in libretro makefile. + * OpenXR - Wrong condition removed (#16271) + * OpenXR - Refactor compatvr.ini file (#16270) + * Apply Kyria's updates to tr_TR (Turkish) translation + * Minor refactoring in PresentationCommon + * PresentationCommon: Use refcounts to simplify shader module management + * Vulkan: Fix frame ordering issue with postprocessing shaders + * GPU: Respect matrix and reverse flag w/o normals. + * softgpu: Multiply prev normal by world matrix. + * softgpu: Respect negate normal flag without norm. + * Update zh_TW.ini + * Update zh_CN.ini + * OpenXR - Camera distance adjust added + * OpenXR - Rearrange VR settings + * Update zh_CN.ini + * Some more codestyle cleanup + * Change wrap_mode to clamp for bicubic upscaler + * Libretro buildfix attempt again + * Remove "attachment" parameter from BindFramebufferAsTexture everywhere. + * Libretro buildfix attempt + * Vulkan: Correct some enum switch warnings. + * Kernel: Fix reported StopThread error. + * headless: Update passing tests. + * headless: Fix crash running some tests on Vulkan. + * headless: Allow screenshot to not be 512x272. + * SDL: fix glew on wayland by ignore glx + * OpenXR - Use only the first projection matrix in frame for mirroing detection + * OpenXR - Ensure scene analyze is called the same way as before + * Build/warning fix + * Remove newly added dependencies on PPSSPP + * Don't use PPSSPP-specific stuff in Common/VR for now. + * Buildfixes + * Android non-gradle buildfix + * CMakeLists.txt fix + * Make as much as possible of the VR code build on all platforms + * Use GPU "use" flags to replace IsVRBuild in the renderer. It remains elsewhere. + * Reorder the GPU USE flags a bit + * OpenXR - Move VR passes count calculation + * OpenXR - Reduce uniform calls + * OpenXR - Simplify projection matrix analyze + * OpenXR - Get rid of VR tweaks + * OpenXR - Move matrix composing out of VR tweaks + * Two more renames to make things read better + * Rename GPU_SUPPORTS_ to GPU_USE_ + * Rename gstate_c.Supports to gstate_c.Use + * GPU: Replace logic ops with blend for simple cases. + * softgpu: Correct linear interp for uneven positions. + * softgpu: Correct drawing outside TL of rectangle. + * softgpu: Make triangle fan rect detection generic. + * OpenXR - Add VR mode for stereo flat screen rendering + * Savedata: Update filelist on file erasure. + * Kernel: Stop reporting invalid semaphore names. + * Kernel: Stop reporting invalid mutex names. + * Code cleanup, optional param renamed: "recreate" -> "remove" + * Fix save states + * OOP + * know unknown in sceMp4Create + * Add draft mp4msv module + * Kernel: Add reason to reported stop error. + * Windows: Improve reporting of WASAPI errors. + * Io: Fail open earlier from ms0: w/ no access flags. + * jit: Run invalidates immediately. + * jit: Consistently check range on invalidate. + * jit: Ignore zero byte icache invalidates. + * D3D9: Correct scissor state cache in Draw. + * headless: Update tests. + * softgpu: Cull a triangle with all negative w. + * softgpu: Fix crash on screenshot w/o display. + * Removed the bugged part of hle.func.scan + new comments + * Vulkan: fix wayland swapchain size 1x1 + * typo + * OpenXR - Ensure the app gets unloaded when killed by OS + * OpenXR - Disable VUlkan for now + * Revert "Vulkan: Create depth/stencil buffers on demand" + * Add ShaderId utility function to be used for some sanity checking. + * Add simple way to add debug annotation in the middle of the command stream. Vulkan-only. + * OpenXR - Create VR framebuffer for Vulkan + * OpenXR - Hookup VR API into Vulkan + * Comments updated, default value for parameter "recreate" set to false + * Added optional parameter to hle.func.scan, improved code of hle.func.removeRange + * GPU: Respect stencil state in shader blend. + * Followup to #16205, fix one more instance of the problem. + * Removes the option, autodetect instead - only enable if the GPU doesn't support bitwise ops. + * ArmJit: Save/restore downcount where needed, we missed a few cases. + * Revert "Disables "ForceCheck" on jit invalidation on ARM32, introduced in #16194" + * Touchpad/mousewheel scrolling acted very weirdly on Mac due to 0 deltas. Make it work better. + * Disables "ForceCheck" on jit invalidation on ARM32, introduced in #16194 + * Warning fixes + * GE Debugger: Display if tex is framebuf. + * GE Debugger: Use a modern GL context. + * GE Debugger: Remove ancient unecessary GL calls. + * GE Debugger: Normalize framebuffer texture preview. + * GE Debugger: Centralize current fb tex fetch code. + * D3D9: Remove block transfer code overrides. + * D3D9: Implement CopyFramebufferToMemorySync(). + * Vulkan: Fix initial depth copy w/o existing depth. + * MIPS: Fix non standard layout offsets. + * GLES: Add debug readback of stencil data. + * GLES: Refactor depth pipeline create. + * GPU: Rename stencil write pipeline. + * GLES: Allow stencil texturing if supported. + * it_IT - Italian localization: Apply giona888's latest changes + * Experimentally revert "Delete some misguided FPU control code, that we definitely don't want" + * Comment about VR uniforms + * Don't use fragmentShaderInt32Support as a replacement for checking for bitwiseOps + * Don't expand alphaColorRef to 128 bytes on backends where we don't need to. + * glslang: Set a missing resource limit. Avoids spurious errors in shader unit test + * Revert "Revert "Loader: Validate offsets and truncation in ELF."" + * Update elfSize to uncompressed elf size when needed. Fixes infinite loading in Wipeout. + * Revert "Loader: Validate offsets and truncation in ELF." + * Track min and max GPU time, init time, render CPU time. + * Delete some misguided FPU control code, that we definitely don't want + * Vulkan (trivial): Fix numDraws stat when merging render passes. Shorten a name. + * GLES: Simplify, enable debug depth readback. + * GLES: Use Draw for depth readback shader. + * GLES: Enable EXT_depth_clamp where supported. + * GLES: Refactor depth shader download. + * GLES: Depth download cleanup. + * Windows: Retain log console on graphics restart. + * jit: Defer invalidations made while running. + * GE Debugger: Reset on shutdown. + * Loader: Get rid of overalloc on module exec. + * Loader: Validate offsets and truncation in ELF. + * D3D11: Correct depth readback. + * Savedata: Cache SFO data during processing. + * Savedata: Skip and defer texture exist check. + * Savedata: Combine file info and dir listing calls. + * Io: Provide directory existence with listing. + * Android: Differentiate empty dir from failure. + * Savedata: Reduce some duplicate file info lookups. + * Font: Reduce internal font loading IO. + * Io: Defer GetFileInfo() until it's needed. + * Io: Reduce operations inside GetFileInfo(). + * Module: Cleanup file read pattern. + * GPU: Rename readback and buffer write operations. + * D3D9: Correct depth readback. + * Depal from dynamic CLUT: When detecting bounds, be more conservative. + * Vulkan: Create depth/stencil buffers on demand + * Fix depal bounds with dynamic CLUT. Fixes lens flare glitches in Ridge Racer + * Remote API: hle.func.removeRange added + * GLES: Hook up depth download. + * Vulkan: Disable geometry shaders for Mali <= 18. + * GPU: Correct sizing account on block transfer. + * GPU: Hook Gods Eater Burst avatar read. + * GPU: Consider depth buffers in block transfer. + * Jpeg: Cleanup debug dump file access. + * Jpeg: Encode and decode YCbCr as 2x2. + * Jpeg: Correct some logging of return types. + * Debugger: Fix off-by-one in func scanning. + * GPU: Refuse to set fb_address == z_address. + * GPU: Ignore depth when masked and ALWAYS. + * GLES: Use GL_LUMINANCE on GLES for indexed tex. + * D3D9: Add simple rendered CLUT handling. + * D3D9: Correct sampler states for binding 1+. + * Jpeg: Correct colorspace conversion funcs. + * Jpeg: Decode buffers using configured stride. + * Jpeg: Improve error handling for YCbCr decode. + * Jpeg: Retain previous width/height. + * OpenXR - Revision of VR rendering + * Jpeg: Improve timing/errors for GetOutputInfo. + * Psmf: Fix save states. + * Jpeg: Play it safe with load/unload handling. + * Jpeg: Validate CreateMJpeg() parameters. + * Jpeg: Cleanup Init/Finish error checks. + * Fix as @hrydgard suggest + * Typo fix, thanks iota97 + * Shader uniforms (VK/D3D11): Fix issue where we could overwrite the fourth component padding. + * GPU: Clip against neg Z even w/o cull support. + * GPU: Verify generated shader buffer length. + * GPU: Optimize clip distances needed. + * Vulkan: Correct clamped Z clip when clipping neg Z. + * Vulkan: Avoid max_vertices=12 if unnecessary. + * Vulkan: Clip clamped depth in geometry shader. + * Forget remove the include + * Store crc change to u32 + * Reporting: Always report hleReportDebug. + * Reporting: Avoid accessing ticks before start. + * Vulkan: Generate indices in clipping. + * Compat: Remove DisableRangeCulling. + * Vulkan: Clip to neg z in the geometry shader. + * Improve Freedesktop integration * Move scalable icon install from `${prefix}/share/pixmaps` (`pixmaps` has been deprecated long time ago) to `${prefix}/share/icons/hicolor/scalable/apps` * Add `Keywords`, `Categories`, and `MimeType` keys to `.desktop` file * Add shared MIME-info on [`application/x-compressed-iso`](https://en.wikipedia.org/wiki/.CSO "Compressed ISO") MIME type + * Try to build fix + * Fix Kurohyou again. See #9576 + * GPU: Skip matching a framebuf for RAM. + * GPU: Use flags to fix triggered upload/download. + * Add crc record to psmfplayer + * Add crc record into mpeg module + * Add crc calculation in loading module + * Vertex decoder: Add fallback for non-SSE4.1 + * Fix another crash with non-buffered rendering + * More libretro buildfix, thanks stuken + * Attempt to fix libretro build + * Fix validation error with the GPU profiler + * Improve Vulkan profiler (show framebuffer tag for render passes) + * Fixed syntax accessing method reviewed by Unknown W. Brackets + * softgpu: Minor opt, ignore unused z_stride. + * GPU: Normalize framebuf addresses. + * GE Debugger: Ignore mirrors for target in record. + * GE Debugger: Correct mask in target breakpoints. + * GPU: Mask away unused bits in framebuf/zbuf ptr. + * Vulkan: Cleanup unused geometry shader vars. + * Vulkan: Allow configuring geometry shaders on/off. + * Vulkan: Block geometry shaders on older Mali. + * Vulkan: Use geo clip distance only where supported. + * Vulkan: Enable geo shader for culling. + * Vulkan: Correct geometry shader culling. + * Vulkan: Cull in geoshader, hack to on for now. + * Vulkan: Basic geoshader code generation. + * Vulkan: Update shader cache format for geo shaders. + * Vulkan: Add geometry shader ID tracking. + * Vulkan: Clean up shader module tag. + * GPU: Add dirtying for geo shader state. + * UI: Prevent rendering during reboot. + * assets/lang: translate some key phrases in pl_PL.ini + * headless: Update tests. + * GE Debugger: Fix memcpy/memset recording. + * GE Debugger: Record the Edram translation value. + * Ge: Report and save Edram translation value. + * Debugger: Fix game.reset by adding a reboot status. + * Build: Ignore no symbols link warning. + * Build: Fix debug build on Windows 32-bit. + * Tighten up some color format checks with displays and copies + * GPU: Add geometry path for shader writer. + * unittest: Cleanup stage/glsl version. + * Draw: Add COLOR1 semantic. + * ES2 crash fix: Don't draw depth if lacking fragment shader depth write. + * Build: Raise macOS/iOS min version for c++17. + * armips: Use native std::filesystem on Windows/UWP. + * armips: Also update UWP and ARM64, etc. + * armips: Update to UTF-8/c++17 armips. + * Core: Remove disallowed register. + * Build: Update zstd to latest dev. + * armips: Update to latest. + * Better readability and optimization insertion into container by replacing 'insert' -> 'emplace', 'push_back' -> 'emplace_back' + * Fixed bug and memleaks + * Code readability, vec reserve() and remove excess c_str() + * Correct cleaning string and remove unused vars + * GE Debugger: Correct rounded coords in vertex list. + * softjit: Fix dst blend shift. + * softgpu: Avoid projecting textures in common case. + * GPU: Fix missed dirtying when fast loading tgen. + * GPU: Restore matrices with dirtying. + * Use compat.ini to avoid readbacks in Silent Hill: Shattered Memories as well. + * Remove redundant call to SetMenu + * Add hidden ini file setting that you can use to hide the menu bar. + * Typo fix + * Update README.md + * OpenXR - Add README.md about the SDK + * Update the Vulkan debug names when reassigning depth buffers. + * Add Silent Hill: Origins to compatibility setting BlockTransferAllowCreateFB. + * If a framebuffer starts using a different depth buffer than before, re-point. + * Quiet the Vulkan miniprofiler (for texture uploads etc) a bit + * softgpu: Correct matrix value update wrapping. + * Ge: Move matrix reading into GPU. + * GLES: Fix colortest/logicop uint/int conversion. + * Debugger: Add API to scan memory for funcs. + * Debugger: Fix doc typo for disasm/assemble names. + * OpenXR - Fix FBO colors on Pico + * OpenXR - Fix the build on Quest + * OpenXR - Support for Pico controllers + * Vulkan: Better debug names for RENDER passes. + * Remove alpha ignore in xbrz texture shaders. + * OpenXR - Initialization on Pico fixed + * OpenXR - Code formatting + * Forget remove unused include + * Change to mpegLibVersion + * Better bit scrambling when computing draw call IDs for vertex cache. + * softgpu: Handle rectangle texture projection. + * softgpu: Implement triangle texture projection. + * softgpu: Expand texture coords to include q. + * softgpu: Split clippos out of rasterization vert. + * GPU: Correct flat normal projection mapping. + * GPU: Correct normalized zero normal proj map. + * softgpu: Save last tc/normal in vertex reading. + * Fix color test. + * Add compatibility flag for loading pixels on framebuffer create using nearest filtering + * Mpeg: Correct Au update size. + * GPU: Apply color test mask as a uint. + * GPU: Consistently use uvec3 for colortest. + * ivec->uvec, comment fix + * ShaderUniforms: cleanup, put every "4-float" on a line for clarity + * Shuffle constants around, squeezing them into gaps. Saves another 16 bytes. + * Fragment shader uniforms: Pack color mask in 32 bits instead of expand to 128 bits. + * Fold the "materialUpdate" flag into the light ubershader part. + * Add some recent flags to FragmentShaderDesc + * Remove suspicious dirty flag + * Prepare for dynamic mat update + * Minor debugger fix + * Always do the vertex shader part of the fog computation. + * GPU: Fix SSE4 Vec3f normalize. + * Oops! Also, testfix + * Fix light ubershader for D3D11 and OpenGL, GLES unsigned/signed stuff + * Experiment: Generate "Ubershaders" that can handle all lighting configurations + * Replace faulty layout transition deduplication algorithm. + * Vulkan: Add debug names to our command buffers, so you can see which one errored in validation + * Minor cleanup in TinySet + * Trying to improve sceMpegAvcDecodeYCbCr + * (UWP) Remove expandedResources + * Remove debug log. Add some new debug log though, unrelated to this PR, for fb clut + fb texture. Plus a couple asserts. + * OpenXR - Pico specific code refactor + * softgpu: Avoid waiting for a thread to drain. + * softgpu: Increase queued prims. + * softgpu: Fix triangle strip with partial rects. + * softgpu: Simply 5551 blending fast path. + * softpu: Refactor out 5551 fast path checks. + * softgpu: Avoid fast path in another wrong case. + * D3D9: Allow INTZ depth buffers more correctly. + * Vulkan: Avoid allocating depth images for stuff like temp copies, depal buffers etc. + * Bump the shader cache version + * Vulkan: Don't compile pipeline variants that don't make sense given their flags. + * OpenXR - Initialize Pico subsystems + * OpenXR - Disable performance extension on Pico + * OpenXR - Support multiple loaders + * GE Debugger: Add option to track pixel in preview. + * OpenXR - Add manifest for Pico + * OpenXR - Rename Quest build variant + * GE Debugger: Copy images optionally with alpha. + * GE Debugger: Add a direct copy for preview images. + * OpenXR - PreprocessStep function implemented + * OpenXR - Simplify skyplane detection in GTA + * OpenXR - Get rid of an ugly if condition + * GPU: Cleanup unused CheckAlpha() funcs. + * softjit: Skip reading dst pixel where blended out. + * softgpu: Optimize rectangle sampling/blending. + * softgpu: Ignore a needless color test case. + * GE Debugger: Tag frame dump replay VRAM writes. + * GE Debugger: Fix small tex/clut recopying. + * HLE: Fix bad logging without latestSyscall. + * Vulkan: Avoid crash on two backbuffer steps. + * Remove ill-formed asserts + * More tweaks, fix crash on exit (double-join thread) + * Small tweak + * Can't seem to get around this condition variable... Oh well. + * Readback fix + * Greatly simplify the thread synchronization. + * Flatten the Run function, to make it easier to reorganize. + * Make functions private that should be private + * Fix assertion issue when playing Cars over public adhoc server. + * Best practices layer warning: Don't pass in unused clear values. (probably zero impact, heh). + * Handle depth buffer write-after-write hazards in the queue runner + * Fix leak of shader modules from thin3d. + * Another uninitialized variable (VAI minihash/hash) + * Make valgrind happy + * Fix memory leak (of Promise objects) in VK pipeline wrappers + * Add Burnout Dominator to list of games that require buffered rendering + * GPU: Upload depth only on first usage. + * softgpu: Fix self-render detect in Ridge Racer. + * softgpu: Cache reused indexed verts. + * softgpu: Skip matrix multiply for fog factor calc. + * softgpu: Remove unnecessary state param. + * Texture/Framebuffer match: Ignore stride if texHeight == 1. Fixes Ridge Racer lens flares. + * Use Unknown's idea from #16081 instead of the UploadDepthForCLUTTexture compat.ini flag. + * Fix Vulkan regression of #16075 due to silly typo. + * Fix corruption of Ridge Racer speedometers with AutoMaxQuality enabled. + * Cleanup logic + * Fix for render pass merge + * If depth or stencil are cleared in a renderpass, set the pipeline flag. + * Vulkan: Don't have renderpasses store/load depth buffers when we don't use them + * Preserve depth buffer on framebuffer resize, if has been used. + * More lenient heuristic for uploading depth buffers. Still behind compat flag. See #11100 + * Minor cleanups around dirtying of render state + * softgpu: Avoid unnecessary flushing for curves. + * GLES: Avoid resizing tessellation data textures. + * GPU: Correct neg normal for hwtess texgen. + * GE Debugger: Default auto flush enabled. + * GPU: Use hardware tess params for texgen. + * Dirty more state after depal + * Don't try to replace or scale CLUT8-on-GPU textures. + * Shouldn't 'floor' there + * CLUTs can be loaded from small rectangular textures. Need to linearize. + * OpenXR - Stereo mirroring fixed, disable stereo when no world scale defined + * Vulkan: Avoid crash in headless on finish. + * Kernel: Match index lookup behavior for tls. + * GPU: Simplify depth clamped clip planes. + * GPU: Clip clamped depth accounting for perspective. + * GPU: Account for w properly in lines, fixing width. + * softgpu: Narrow blend check for fast path further. + * headless: Use display buf for compare screenshot. + * GPU: Correct display framebuffer reading. + * softgpu: Fix display framebuffer read. + * Fix BGRA issue on D3D with GPU CLUT textures + * softgpu: Correct clear rect off by one issues. + * OpenXR - Compile fix + * UI: Fix controller autoconf list result. + * headless: Update tests. + * Kernel: Disallow partitions 8/10 to match tests. + * OpenXR - Support for all headset's IPDs fixed + * Kernel: Allow volatile for MsgPipe buffers. + * Kernel: Allow volatile for Tlspl objects. + * Throw in assorted warning fixes + * Kernel: Allow kernel/volatile VPLs as well. + * Submit main command buffer before acquire. Fix sync bug. + * When taking screenshots, make sure that the readback command happens on the right cmdbuf. + * Kernel: Allow allocating FPLs in volatile memory. + * Minor code cleanups (hasBegun is now redundant, for example) + * Rewrite some present logic for slightly more clarity. Fixes bad logic and a minor race condition. + * Various refactoring + * Unify depth texture and framebuffer fetch checks + * Unify range culling detection + * Start unifying setting of the GPU feature flags, now that thin3d has feature detection. + * Kernel: Load volatile memblocks from save states. + * Kernel: Check sceKernelAllocPartitionMemory args. + * Kernel: Fix save states for volatile memory alloc. + * GE Debugger: Include rendered CLUTs in frame dumps. + * Address feedback + * Vulkan: Remove the useThread bool entirely. + * Start moving stuff out to VulkanFrameData.cpp + * softgpu: Skip flushing on no change. + * softgpu: Skip more triangles with equal verts. + * Windows: Normalize paths for OpenDirectory(). + * GE Debugger: Save current clut in frame dumps. + * HLE: sceKernelAllocPartitionMemory volatile memory support (partition 5) + * Gamedata: Cleanup install request args/validate. + * Fix the UI setting "Buffer graphics commands". Was off by 1. + * GPU: Clip depth properly when also clamping. + * GLES: Support more clip distances. + * softgpu: Correct line early z checks. + * GE Debugger: Allow search by state name. + * GE Debugger: Show time spent stepping. + * softgpu: Fix s8 primitives in throughmode. + * headless: Add some new and passing tests. + * GPU: Fix crash on imm vert triangles. + * softgpu: Correct secondary color on imm verts. + * softgpu: Fix state/continuation for imm prims. + * softgpu: Support fog and color1 on imm verts. + * softgpu: Refactor imm draws to bypass vert read. + * softgpu: Simplify vertex reading. + * softgpu: Simplify vertex range culling. + * softgpu: Explicitly flush on sync and output. + * GPU: Minor cleanup of unnecessary virtual. + * Quick GE debugger stepping fix, after the recent vulkan swapchain acquire change + * Update pspautotest submodule + * Update ffmpeg submodule (no big news, just readme updates and a gcc buildfix) + * Consider the Adreno and Mali stencil-discard bugs the same. Use the better check from the Mali bug. + * Show GPU device ID in system information if available. + * headless: Benchmark a test or frame dump. + * headless: Add option for MSE pass threshold. + * headless: Switch screenshot error to MSE. + * Headless: Small cleanup of options handling. + * softgpu: Run early Z tests in fast rect path. + * UI: Fix some logic dependent on screen tags. + * Remove assert (this happens when playing framedumps) + * Finish the switch to late acquire, moving the swapchain responsibility to the queue runner. + * RunSteps cleanup + * Move FrameData out of VulkanRenderManager + * Move back buffer rendering to a separate "present" command buffer + * Remove splitSubmit setting. Now we submit init commands before the acquire. + * Flag shader blending as broken on Adreno for now. + * Need to have input attachments in the descriptor pool. + * Remove logspam + * Show render pass types in GPU profiler + * Use subpass dependencies to implement shader framebuffer read in Vulkan. + * Add two more extensions to device extension lookup + * Careful with min filtering of framebuffer textures only if auto max quality is on. + * Increase log level of ppsspp version + * Ah, need to enable the Mali workaround for Outrun's water, too. + * Logic op fix, min filter fix. Also remove some unnecessary indentation. Fixes the water on Adreno (no logic) + * Log the screen tag when logging scissor errors. + * Add tags to all our UI screens. Useful to debug UI problems. + * Fix const pointer conversion error for RISC-V + * D3D9 crashfix (flare doesn't render correctly though) + * Copy the texels directly in LoadCLUT + * Small simplification + * Address feedback, change roles of the two dynamic clut fbos to eliminate the need for state (see next commit) + * Seems Need for Speed: Shift has exactly the same lens flare, enable the same compat settings. + * OpenXR - Better variable/methods naming + * OpenGL and D3D11 fixes + * OpenXR - Add code comments + * Log improvements + * Ridge Racer fix + * Get depal-from-dynamic-CLUT working + * Set texcache entry status flag if CLUT needs to be dynamic + * Add way to bind cached textures to a DrawContext + * Allow binding "native" texture objects to thin3d + * Start implementing ApplyTextureDepal + * Cleanup in ge_constants.h, add virtual CLUT8 buffer format + * OpenGL fix + * Comment + * Force low resolution for effects in Burnout Dominator. Fixes lens flare at high res + * Enable depth uploads on render-to-clut-buffer. Esoteric but needed for #11100. Compat flag for now. + * Enable DownloadCLUT and ShaderBitmask for Burnout Dominator + * Detect the CLUT8/8888 situation + * Buffer copy matching workaround + * OpenXR - Ensure the head pose is locked during frame rendering + * OpenXR - Vulkan support in progress + * Switch multiple bools as parameters to DecodeTextureLevel to a flags enum. + * Fix regression in Tantalus games. See #15914 + * GE Debugger: Allow fb copies while stepping. + * GPU: Fix simulating logicop with blend and shader. + * OpenXR - Move skyplane compat flag into compatvr.ini + * Unify CLUT texture size at 512 to reduce complexity. Worth the extra bytes. + * Extract function GetReinterpretPipeline + * Savedata: Return errors more correctly on load. + * softgpu: Force correct alignment on 32-bit. + * Common: Fix some type compare warnings on 32-bit. + * softgpu: Remove mostly unused screen offset state. + * softgpu: Cache texture bufws at 16 bit. + * softgpu: Reduce size of VertexData texture coords. + * OpenXR - Initial work on Vulkan support + * Non-buffered rendering crashfix + * OpenXR - Refactor variable name + * OpenXR - Crash when using 3D stereo fixed + * OpenXR - Requested header change in compatvr.ini + * OpenXR - Use per game stereo separation file + * Address feedback + * Better shader compile logging + * Don't start blending in the shader just because there's bitmasks, if blend is disabled. + * Debug-name shader modules with their shader desc strings + * softgpu: Ignore stencil test mask in ALWAYS mode. + * softgpu: Include early Z in func description. + * softgpu: Detect longer strips of rectangles. + * softgpu: Detect full triangles as rectangles. + * softgpu: Refactor triangle cull processing. + * softgpu: Store vertex colors as packed RGBA8. + * softgpu: Enable early Z tests a bit more often. + * softgpu: Reduce some minor bin item field sizes. + * README: Move some older releases to history. + * GE Debugger: Add some constants to expressions. + * GE Debugger: Add fields to register expressions. + * Add Rewrite to DrawSyncEatCycles hack, softlocks without it + * GE Debugger: Eval breakpoints as if register set. + * GE Debugger: Provide addrs in breakpoint conds. + * Remove CLUT matching case that shouldn't be there yet. + * softgpu: Switch vert continue buffer to members. + * softjit: Implement accurate fog color blending. + * softgpu: Correct accuracy of fog calculation. + * softgpu: Restrict sprite fast path to throughmode. + * Shadergen issues found by the test + * Switch more texture matching calculations to use bytes for matching. + * Texture/framebuffer matching: Do margin check in bytes. + * Add ShaderDepalMode enum, use in shader ID. Replaces smoothed bit. + * Implement shader depal for D3D11. + * DownloadFramebufferForCLUT: Apply the fix mentioned in #8590 + * CLUT download: Replace the broken block transfer check with a new compat flag. + * vertexjit: Use consistent skinInDecode. + * softgpu: Check depth test early on simple stencil. + * softgpu: Clamp/wrap textures at 512 pixels. + * GE Debugger: Correct confusing block transfer size. + * D3D11: Prevent crash on larger textures. + * TexCache: Allow valid bufw sizes. + * Vulkan: Avoid crash if large texture attempted. + * GE Debugger: Fix crash on bp before list running. + * GE Debugger: Fix crash stepping with large tex. + * softgpu: Avoid calling unordered coords tl/br. + * softgpu: Avoid over-aggressive rect conversion. + * GE Debugger: Fix logging of texture size. + * softgpu: Allow almost flat rectangles to go fast. + * softgpu: Refactor duplicate rectangle compat check. + * Add logging for LoadCLUT from framebuffer. See #15923 . Logs nicely in Burnout Dominator. + * GE Debugger: Correct cond break removal warnings. + * Vulkan: Wake compile thread when enqueued. + * Update version to 1.13.2 + * Update README.md for v1.13.2 + * CI: make indentation in tarball.yml consistent + * CI: also upload release source tarball on tag push + * GPU: Dirty params when converting viewport state. + * softgpu: Correct morph weight updates. + * Add Emulator to Categories in ppsspp.desktop.in + * Vulkan: Declare gl_Position as invariant. See #11583 + * readme link + * D3D9: Correct viewport offset sign. + * D3D11: Correct viewport offset sign handling. + * Remove a per-audio-frame allocation in sceAudio + * Initialize a bunch of Dialog structs + * Font: Correct uninitialized values. + * Atrac: Avoid accessing uninitialized stream status. + * Atrac: Avoid an uninitialized value. + * softgpu: Correct bad munmap() size. + * atrac3: Apply the overallocation workaround in a couple more places. Fixes #15788 . + * Atrac3+: Allocate some extra bytes for the data buffer to workaround decoder bug. + * atrac3: Apply the overallocation workaround in a couple more places. Fixes #15788 . + * Warning fix + * atrac3: Allocate some extra bytes for the data buffer to workaround decoder bug. + * OpenXR - Compile fix + * OpenXR - Do not mashup Common/GPU + * OpenXR - Refactor + * Match FragmentShaderFlags definition with the other PR. + * Some more debug names to vulkan objects + * UI: Resize after visiting developer settings. + * GPU: Make NotifyConfigChanged() calls consistent. + * Build: Another attempt to fix libretro. + * Fix compiler warnings + * Assorted cleanup + * Fix for desc struct + * Vulkan: Improve tagging of pipelines for debugging purposes + * Extend lifetime of ShaderModule objects since we might need to recreate pipelines for different render passes + * Fix pipeline lifetime issue, misc. + * Correct merging of render passes. However, we have lifetime issues. + * Lifetime fixes, cleanups + * Fix shader caching. + * Also create renderpass variants on demand + * Create framebuffer variants on demand + * Split each renderpass/framebuffer into multiple "compatibility classes" (RenderPassType). + * Another attempt at fixing libretro, this time the win build failed + * OpenXR - Use fog color to replace the sky + * OpenXR - Refactor VR compat integration + * GE Debugger: Show imm prim flag detail in disasm. + * GPU: Handle more flags on imm prim command. + * softgpu: Use vertType prim override for flags. + * GPU: Allow usage of texturing in immediate verts. + * GPU: Handle immediate prims more accurately. + * GE Debugger: Count imm prims as prims. + * softgpu: Fix immediate rendering crash. + * GPU: Fix crash on overlap copy w/o cur target. + * GPU: Fix CanDepalettize() missing false paths. + * Build: Add additional zstd file for libretro. + * d3d9: Correct decode of 16-bit textures. + * Add Lubos to credits list + * GE Debugger: Add more missing GE disasm. + * softgpu: Implement matrix data wrap-around. + * Ge: Read matrix num/data back like hardware. + * Build: Include all zstd files in libretro. + * GE Debugger: Add conditions to cmd breakpoints. + * GE Debugger: Add address breakpoint conditions. + * GE Debugger: Allow expressions for goto address. + * Remove impossible check + * Buildfix, warn fixes while at it + * NFS speed effect: Improve framebuffer choice logging, tighten up stride checks. + * Correct libretro buildfix (hopefully) + * OpenXR - Config for stereo separation added + * GE Debugger: Add a table of register metadata. + * OpenXR - Stereo matrices fixed + * libretro build fix, see #15968 + * Remove reference to non-existent file from Common.vcxproj + * The Darkstalkers subpixel check should be == 0. Fixes #14953 again + * OpenXR - Fixing FOV better + * Fix for ARM 32-bit. + * Fix case problem in libretro make file, see #15964 + * Remove the assembly implementation of 4x4 matrix mul, in favor of intrinsics. + * Disable BlueToAlpha mode when logic op is present. Re-enable for Outrun and DiRT. + * Vulkan: Update to latest VMA. + * Build: Update xxhash to v0.8.1. + * Build: Update to zstd v1.5.3. + * Build: Split ppsspp_common out for Android bins. + * Water effect fix #2: If ForceLowerResolutionForEffects, don't upscale textures participating in small-resolution passes. + * Water effect fix #1: Force "Lower resolution for effects" on in Outrun and DiRT 2 + * Fix "Lower resolution for effects" in Ratchet & Clank. Bad option for this game though. + * Fix the new logic-op-in-shader on OpenGL and D3D11 + * OpenXR - Mismatch in stereo and normal uniform fixed + * OpenXR - Do not include lens projection in nonVR builds + * OpenXR - Split projection matrix for color and depth buffer + * Shader unit test fix + * Preserve alpha in shader logic op emulation. Include logic op in check for packUnorm4x8. + * OpenXR - Mismatched call order fixed + * OpenXR - Multipass rendering fixed + * OpenXR - Stereo without multiview added + * Turn on logic ops in shader when shader bitmasking is used. + * Shader bit setup and code generation for logic-ops-in-shader + * UI: Only force ASCII messages if they're ASCII. + * headless: Fix startup on latest Android versions. + * arm64jit: Handle branch/jump in branch delay slots. + * x86jit: Refactor and fix jump in branch delay slot. + * irjit: Handle branch/jump in branch delay slots. + * Fix dialog rendering (PPGe). Fixes #15954 + * spaces->tabs + * Fix GLES check + * Additional cleanup + * Add log functions to state structs + * Move the computation of simulate-logic shader type earlier + * Some more renaming and refactoring, extracted from a future PR + * Channel mask refactor + * Correct LogicOp API support detection + * Minor refactor with pipelines in QueueRunner (makes more information available for easier debugging) + * Don't parallelize shader compiles in debug mode, sync in debug allocator kills performance. + * interp: Handle jumps in branch delay slots better. + * softgpu: Correct texture check in fastpath. + * softgpu: Validate alpha test properly for fastpath. + * softgpu: Avoid rect fast path with subpixel offset. + * Kernel: Use PSPPointer for status objects. + * Io: Use PSPPointer instead of struct temps. + * Font: Remove needless ReadStruct(). + * Ge: Switch from ReadStruct to PSPPointer. + * Psmf: Use PSPPointer instead of WriteStruct. + * Net: Remove usage of Memory::WriteStruct(). + * HLE: Remove misc usage of WriteStruct(). + * PPGe: Cleanup memory writes for clarity. + * Fix lying comment, log level reduction + * Promise: Add SpawnEmpty (couldn't come up with a better name), fix bug with multiple waiters + * Switch Vulkan pipelines to use promises for synchronization + * VS 2022 warning fix + * More buildfixing + * Buildfix + * Buildfix + * Fix silly logic, fixes the shader leak. + * Work towards finding the leak + * Don't block on shader creation until it's time to create the pipeline. + * Use shader module promises as keys in pipeline cache instead of the actual shader modules. + * Create shader modules (and compile GLSL) on worker threads + * minor sv translation fix + * Use promises to hold shader modules, for later asyncification + * Remove confusing resetFramebufferRead flag from secondary framebuffer binding + * Remove setting "Disable slower framebuffer effects" + * Remove unused function + * langtool: Run fmt, remove wrong comments + * langtool: Add easy command to remove a key. + * unitest: Correct wrong free, initialize data. + * vertexjit: Update unit test, oops. + * Avoid recomputing the shaderblend setup in ComputeFragmentShaderID. + * Small steps + * Some reordering + * More renaming + * Plumb the computed pipeline state into ComputeFragmentShaderID. + * Some renaming. Store the BlendState/MaskState. + * OpenXR - FBO intialization better + * vertexjit: Remove unused ReadUV() cases. + * vertexjit: Remove unused ReadPos() cases. + * vertexjit: Clamp through float pos during decode. + * Replacement: Fix reload of replaced textures. + * Replacement: Detect OOM even with replacements. + * Vulkan: Prevent saving tex during replace load. + * Revert "Merge pull request #15930 from sum2012/kernel-minor" + * compile fix + * OpenXR - Viewport init better + * OpenXR - Emulate mouse cursor + * Fix image format when dumping textures. + * Fix texture saving for Vulkan + * Display paths with backslashes on Windows. Show texture path when saving. + * Shrink VkRenderData from 88 to 64 bytes. + * Another buildfix attempt + * libretro buildfix + * Remove allocation from the heaviest MemBlockInfo path (BlockTransfer). Required some refactoring. + * Remove a per-audio-frame allocation in sceAudio + * Shrink the DeferredDrawcall struct, because why not. Assorted cleanup + * Shrink the TransitionRequest struct, use TinySet + * TinySet bugfix, add "append" + * fbtex matching: Shrink the matching struct, use TinySet instead of vector to avoid an allocation. + * Stricter checks for negative framebuffer offsets. Fixes #15937 + * Merge GetFramebufferCandidates and GetBestCandidateIndex into one function. + * Much faster framebuffer matching (no more ToString every time there's more than 1 candidate, etc). + * Rename Slow Framebuf effects to Shader Blending (might remove). Reassign readbacks to BlockTransferGPU. + * KillZone: Fix margin detection on the title screen, fixing some glitches. Need more fixes though to fix the shimmer effect. + * Extract the split code to a function + * Quicker way to update the through projection matrix with the split framebuffer offset + * Cleanup (more to do, though) + * Allow signed texture offsets on framebuffers + * Offset the draws properly. Now just some texel clamping left to fix + * Check X coordinate of through-mode drawcalls to figure out which of the splits to render to + * Force clearing framebuffer for GTA games + * Tweak ForceMax60Fps to work for Killzone + * Split the framebuffer in Killzone, with a compatibility flag. + * Small tweak in texture matching. Fixes #15927 + * Fix sceKernelUnlockMutex timing + * Remove "DisableReadbacks" compat flag. MotoGP no longer needs it after BlockTransferAllowCreateFB was added. + * Fix race condition in replacement texture loading + * Implement shader blending for D3D9 + * Clean up mip size calculations + * GE Debugger: Improve display list disasm. + * Address feedback + * Add support for reading depth buffers to the PackFramebufferSync function. + * OpenXR - Multiview refactor to make it optional + * Add another stat, for self-tex + * Remove the reinterpret compat flag, now always on + * Fix OpenGL bug when framebuffer fetch is available. Add two new stats. + * Better framebuffer checks, remove all ways that framebuffer formats can change. + * Use better pattern for enum switches + * Some Vulkan renderpass load/store optimizations. + * Depal: Set scissor tightly around processed area. + * OpenXR - Compile fix + * OpenXR - Swapchain acquiring fixed + * UI: Validate UI scissor inside bounds. + * GPU: Make Spongebob hack checks more readable. + * GPU: Recheck depth/stencil after Spongebob hack. + * GPU: Add a way to defer dirty rechecks. + * GPU: Fix possible shift truncation. + * Update the Spongebob workaround check to work with MX vs ATV: Reflex. See #15907 + * riscv: Add a simple unit test. + * OpenXR - Flag to easily disable multiview shaders added + * OpenXR - Keep using GLES2 headers for now + * More spongebob check tweaks, assert fix + * Remove irrelevant stencil state checks + * Address additional feedback, thanks! + * Safer spongebob depth inverse check, includes color check and depth func + * Unify the spongebob depth inverse check + * riscv: Emit compressed instructions. + * riscv: Improve sign reduce/immediate readability. + * riscv: Emit ADD/SUB/etc. for ADDW/SUBW/etc. on R32. + * riscv: Include an LI helper in the emitter. + * Comments and fixes + * Block transfer bug fix + * OpenGL graphics fixes + * Comment fixes + * Reinterpret is the proper fix for God of War's shadows. + * OpenXR - Dummy integration compile fix + * OpenXR - OPENXR ifdef removal + * Fix UV offsets in 16->32 reinterpret. Better, but something is still not quite right. + * GL bugfix, log cleanup + * Enable reinterprets and create-buffer-on-copy for the Tantalus games + * Fix problem matching framebuffers for reinterpret + * Logging improvements + * Remove some old logging that's not very interesting anymore + * OpenXR - Enable performance mode + * Log improvement + * When cloning framebuffers for reinterpret, adjust width according to bpp + * Framebuffer texture matching: Remove heuristics other than seqCount + * Accept color textures as reinterpretable between 32 and 16-bit formats + * Implement the stencil/alpha reverse trick for all backends + * Fix bug in reinterpret float impl + * Comment fix, cleanup + * Minor D3D11 refactoring + * Fix Katamari + * Bugfix, better logging + * Quick hack to widen the depal texture if 0x200000 swizzle is needed + * More successful swizzle impl. Not perfect + * Unsuccessful swizzle implementation + * wip + * Initial implementation of 32/16-bit color reinterpret blits. + * Refactor reinterpret to prepare for 16/32-bit + * Prepare for 16/32 reinterpret + * Detect need for 16<->32-bit reinterprets + * Work in bytes in the framebuffer matching algorithm for block copies, avoid margins + * Finish the depal refactoring. + * Fix shader compilation error in depth->565 + * riscv: Emit CSR manipulation instructions. + * riscv: Validate FixupBranch usage better. + * riscv: Emit float instructions. + * riscv: Emit atomic operations. + * riscv: Emit mul/div instructions. + * riscv: Emit fence and 64-bit instructions. + * riscv: Emit 32-bit 3-op instructions, validate GPR. + * riscv: Emit 32-bit load/store and imm ops. + * riscv: Emit standard B/U/J type ops. + * riscv: Add initial emitter shell. + * Remove support for framebuffers changing stride (already unreachable). + * Revert minimum stride + * Add missing terminator to VR settings array. Better logging. + * More comment fixes, cleanups + * Comments + * Comment updates + * Merge the two find FindTransferFramebuffer* methods + * Additional renaming + * Rename variables to make the two functions match + * Split up FindTransferFramebuffers + * Assert removed + * Tighten the check a bit more + * Reorder some parameters. Detect buffers in stride gaps. + * Remove unused function decl + * Fix smoothed CLUT lookups. Shouldn't go beyond the mask's reach. + * Add floating point version of reinterpret shaders + * Fix BlueToAlpha together with reinterpret. Add back fake reinterpret for now. + * Reimplement texture format reinterpretation + * Fix smoothed depal on GLES. Don't enable filtering if not using. + * Fix checks for depth texture support, GLES syntax errors + * OpenGL: Fix regular depal (shader depal still worked) that broke recently + * Remove duplicate uniform queries + * Fix wrong width/height values being stored in thin3d textures + * Drive-by texture slot management cleanup + * OSD: Scale large multiline messages to fit better on screen (like shader logging) + * GPU: More consistently clear ptrs on DeviceLost(). + * GPU: Correct Draw2D::DeviceRestore(). + * GE Debugger: Fix bad read on step at start of VRAM. + * GE Debugger: Add separate step based on vsync. + * GPU: Centralize SetDisplayFramebuffer(). + * GE Debugger: Record 1 flip if no display calls. + * Vulkan: Very minor chnages after checking Best Practices with new validation layer + * Use Draw2D for depal shaders (except the actual blit, for now) + * Break out Draw2D in a class + * Switch reinterpret shaders over to the Draw2D framework. + * More Draw2D refactoring + * Small Draw2D refactor + * Rename VirtualFramebuffer->format to fb_format to line up with fb_address and fb_stride (fundamental parameters) + * Rename fmt->fb_format + * D3D9: Correct viewport offset sign. + * D3D11: Correct viewport offset sign handling. + * Copy color from overlapping framebuffers on bind, under certain conditions. + * Implement smoothed depal for the "old" depal path as well. + * Better checks for smoothed depal + * Remove SmoothedDepal compat setting, instead detect the ramp directly. + * Special case depal lookups for Test Drive's strange usage. + * More renaming. Fix shader view for Vulkan + * DepalettizeCommon -> TextureShaderCommon. Simplifications. + * Convert TextureShaderApplier to a member function in DepalCache. + * Simplify depal shader apply code a bit. + * Replacement: Read files only within time budget. + * Replacement: Be more consistent about base level. + * Vulkan: Cleanup minor logging. + * Config: Default PSP model to SLIM on all platforms. + * Linux: Consider Vulkan if OpenGL init fails. + * GLES: Correct preview of mip levels > 0. + * GPU: Share CLUTs if no extended CLUT. + * Headless: Set softgpu config flag. + * Headless: Fix comparison for oversized buffers. + * UI: Show return address for exec crashes. + * Core: Show exception on misaligned jump. + * irjit: Validate alignment in slow memory mode. + * irjit: Add constructs for validing mem access. + * More D3D9 work on depth textures. Something still missing. + * Address more feedback, cleanup scale parameter + * Remove leftover resolutionScale depal config parameter + * Compatibility check + * Use a DX9-compatible method to pass texwidth and height. the effect doesn't work on DX9 yet though. + * Fixes for D3D11 + * Remove debug code, cleanup + * Fix glitch when changing render resolution + * Really not sure how we can detect whether a depth deswizzle is needed :( Hopefully temporary flag... + * Texture bind channel fix. Still need to figure out a way to decide to swizzle or not. + * Fix graphics in Ratchet & Clank. WIP + * Fix naming of temp fbos to be in PSP pixels + * Allow binding depth as 565 by going through depal. + * softgpu: Correct morph weight updates. + * softgpu: Flush on transfer to pending tex read. + * softgpu: Restrict CLUT to proper size. + * GE Debugger: Include extended CLUT in frame dumps. + * GPU: Restrict mip CLUT enhancement a bit. + * GE Debugger: Correct 16-bit CLUT display. + * GPU: Correct depal CLUT texture for 5551/565. + * GPU: Purify ConvertViewportAndScissor(). + * GPU: Dirty params when converting viewport state. + * Address feedback to PR #15858 + * Restore the clearing optimization, avoiding unnecessary depth copies + * Fix deferred-depth for bezier/spline. Move updating of last_frame_depth_render to GPUCommon. + * Minor fixes. Everything seems fine now. + * Implement copying color to depth on load. However, weird stuff is happening.. + * Extract depth copies to CopyToDepthFromOverlappingFramebuffers + * Remove the color-to-depth mode + * Defer depth copies until depth buffer is actually used. + * Texture bind channel fix + * More consistent handling of lack of depth texture support. Small log improvement. + * Fix bug in framebuffer matching where the wrong format was checked for depth matching. Fixes #15867 + * Update ko_KR.ini + * OpenXR - Step closer to identify multiview FBO issue + * OpenXR - Use OpenGL ES 3 and ensure FBO pair calls + * OpenXR - Initialization errors fixed + * Update ko_KR.ini + * Update ko_KR.ini + * Update ko_KR.ini + * OpenXR - Multiview in progress + * Initialize a bunch of Dialog structs + * Fix dependency of RISC-V targets on common JIT code + * Update ko_KR.ini + * Bias towards the most recently written framebuffer + * Remove restrictions on VRAM addresses for depth and color + * Always attempt to match both depth and color + * Pre-reject DXT formats, Misc + * Show seqCount in candidate list + * Stride fix in matching + * Framebuffer matching simplification + * OpenXR - Multiview uniforms working + * OpenXR - Refactor display framerate + * Fix the raster depth blit again, which I recently broke. + * Typo fix + * Silverfall: Add the US game ID too + * Add Silverfall to [AtracLoopHack]. At least we're gathering a good list of test cases once we have an idea for a real fix... + * Use sequence numbers instead of a tracking array for depth buffers + * Revise comments according to feedback + * Oops, remove wrong comments + * Comment updates + * Clean up and comment framebuffer struct better, add bind sequence numbers + * After recording a GE dump, open an explorer window pointing at the file + * Remove the DX9 namespace, for consistency with the other backends + * Vulkan: Restore a minor old forgotten optimization + * More removing redundant stuff + * OpenXR - Disable multiview for now + * Add a newline to "Leaving main" message. + * Refactor 2D pipeline creation + * OpenXR - GL and OXR logging added + * OpenXR - Resuming renderer fixed, haptics support added + * Minor cleanup in reinterpret + * Rely on refcounting for stencil upload shader modules + * thin3d: Properly AddRef and Release shader modules from the pipeline in backends that need it + * add bin and elf to picker as they now work + * Remove some unused code in FramebufferManagerD3D11.cpp/h + * remove storage fileloader and storage folderbrowser + * Remove deprecated function (mostly) + * Remove some unnecessary bool returns. + * Move out the reinterpret stuff to ReinterpretFramebuffer.cpp. Draw2D refactor. + * Assorted minor cleanup + * Revert "Pause the debugger on TRANSFERSTART (treat that as a prim)" + * remove unneeded switch case as un broken cases fall through + * OpenXR - Recenter changes pitch of the 2D screens + * OpenXR - Ensure PSP aspect ratio for 2D content + * OpenXR - Option to set canvas distance added + * OpenXR - 2D mode hack mappable (using screen keycode) + * GPU: Avoid using depth buffers for blending. + * GE Debugger: Fix crash on 0 count prim. + * GE Debugger: Correct base accounting in playback. + * GPU: Skip fb create upload when clearing. + * GE Debugger: Save tab configuration. + * GE Debugger: Cleanup duplicate dlist redraws. + * GE Debugger: Correct tab updating for right panel. + * GE Debugger: Actually allow swapping tabs. + * GE Debugger: Add context menu to control tabs. + * OpenXR - Integrate multiview into system + * GE Debugger: Show/hide tabs if empty. + * GE Debugger: Track tabs per panel. + * OpenXR - Config for stereoscopic vision added + * OpenXR - Default config values fixed + * OpenXR - Unused parameter removed + * OpenXR - VR renderer adjustments + * OpenXR - Framebuffer multiview used + * Move the display list view to its own column on the right + * Add plain Create method to CtrlDisplayListView + * Pause the debugger on TRANSFERSTART (treat that as a prim) + * Add a second set of tabs to the Windows GE debugger. Also reorder stuff. + * oop + * oop + * oop + * Use compat.ini instead + * Add atrac hack setting + * Net: Cleanup initialization syntax. + * GLES: Correct some variable shadowing. + * libpng: Correct func pointer type warning. + * Global: Cleanup some unreferenced warnings. + * Font: Fix bad memory read on invalid font data. + * Assign value to version where appropriate + * Font: Correct uninitialized values. + * Atrac: Avoid accessing uninitialized stream status. + * OpenXR - Support for multiple projection matrices in the scene + * OpenXR - Rotation mirroring fixed in a few games + * Update + * A better way to get property + * Build: Correct ccache keys. + * Atrac: Avoid an uninitialized value. + * softgpu: Ensure TCs are initialized. + * softgpu: Correct bad munmap() size. + * x86jit: Avoid an instruction form Valgrind hates. + * Headless: Correct bad pointer without gfx context. + * Core: Correct shutdown on late init failure. + * Try to fix shader compilation error on Adreno reported by Lubos + * OpenXR - Remove OPENXR ifdef from UI + * OpenXR - Better detecting of 2D menus + * Debugger: Bump GE dump version to ignore bad flag. + * Debugger: Correct GE frame dump VRAM dirty flag. + * Debugger: Fix crash on bad request. + * GPU: Avoid stencil force pass when writing depth. + * GPU: Write stencil fail to alpha is RGB masked. + * Fix feature checks + * Enable framebuffer depth blits through raster. Fixes games like Jeanne D'Arc on D3D9. + * Some enum renaming, move RasterChannel to GPU.h. + * Careful with mipmaps on old OpenGL ES. Lost this check in the refactor, made it clearer than before at least. + * Texcoord attrs must be called a_texcoord0, this should be fixed. Oh well. + * Convert depal shaders to use the ShaderWriter. Add tests for depal shaders. + * Cleanup + * OpenXR - Detect 3D content better + * Warning fix + * Remove BGRA color swap in depal for D3D, no longer needed (and breaks Simpsons) + * Update zh_CN.ini + * Headless: Cleanup unused/leak warnings. + * OpenXR - Detecting UI mode improved + * Change indentation to tabs. + * UI: Show better warning on MAC reset. + * Lang: Fix some mismatched lang keys. + * Build: Remove no-format. + * Common: Avoid format warning with log timestamp. + * GLES: Correct alpha check for depal framebuffers. + * Add -Wformat (see issue #15792), no idea if ok. + * Oops, this iOS check was still required. + * More framebuffer blit fixes. + * Improve the NPOT check + * Forgot these... + * Make GLQueueRunner/RenderManager aware of device caps, use it to check for NPOT and some other stuff + * Remove optimization in BlitUsingRaster that shouldn't have been there (redundant, lacked feature check) + * Reimplement bicubic upscaling. + * GLES: Correct semantic name for Draw2D. + * GPU: Clarify new framebuf stencil upload. + * Dialog: Add missing header for MSVC 2017. + * GPU: Skip alpha stencil upload when safe. + * GPU: Cleanup remnants of GLES-only stencil upload. + * Fixed infinite loop and crash which could occur when using "Import from cheat.db". + * Fix feature check for wrapping of non-pow-2 textures + * Feature detection fix on old GL, more compiler warning fixing + * D3D9: Fix a lot of stuff + * More cleanup, fixes. + * More cleanup + * Buildfixes + * Lock out shaderDepal from D3D for now, will implement later. + * Feature detection fixes + * Delete all the duplicate implementations of ApplyTextureFramebuffer + * Remove Vulkan2D + * Move the depal cache instance ownership to TextureCacheCommon + * Compiles on Vulkan, but doesn't yet work. + * Switch over D3D9 as well. Buggy, will debug after unification + * Works with D3D11, except the depth texturing case + * Simplify + * Fixy fix + * D3D11 prep + * Move the TextureShaderApplier out to DepalettizeCommon + * Move more stuff to Common + * OpenGL: Fix shader depal too. + * Move to Common + * Mostly works for OpenGL. + * Port the OpenGL depal path to thin3d + * Fix some compiler warnings + * ShaderWriter: Fix GLSL 1.0 output + * Fix corrupted colors when using software upscaling with OpenGL. + * Use safe_string when printing %s with Memory::GetCharPointer. + * Prefer using "%04x" over "%04hx". + * State handling reordering in D3D11 and D3D9 backends. + * Fix logging warning issue. + * Update /lang/ru_RU.ini + * OpenXR - 72Hz support added + * OpenXR - Unused variable removed + * OpenXR - HUD is enabled based on gstate + * Fix include issue. + * Updated copyright's year. + * Replace the dummy PSN name with Nickname on sceNpInit. + * Moving sceNpMatching2-stuff to sceNp2.cpp/h. PS: Register_sceNpMatching2 not being added to HLETables because currently the callback didn't work properly yet and could crash the game. + * Implement NP Signin Dialog (faked signin) + * Updated NP Library PS: Some of registered modules should probably moved to a separate files (ie. sceNp2.cpp). + * Added a FunctionWrapper used by sceNpService + * UWP: Fix missing font file causing missing text fields. Fixes #15771 + * Delete more unused code + * Fixes, state invalidation + * Delete unused stuff in all backends + * Remove the D3D9 and Vulkan BlitFramebuffer functions + * Replace D3D11's BlitFramebuffer with the common one + * Convert FramebufferBlit from the GL backend to use thin3d + * D3D11: Remove more custom drawing + * Fix upscaling in Vulkan. Fixes #15779 + * OpenXR - Forgotten file added + * OpenXR - Introduce VR tweaks + * UWP buildfix + * Reimplement filtering in DrawActiveTexture + * Delete the OpenGL implementation of DrawActiveTexture. + * Delete the Vulkan implementation of DrawActiveTexture, fixups to shader gen + * Delete the D3D9 impl of DrawActiveTexture + * Move to common (todo: merge with above) + * Convert the D3D11 DrawActiveTexture to thin3d + * Draw2D + * Need a separate cap bit for stencil blits to prevent misuse on D3D9 + * Cleanup and testfixes + * Unit test, comment + * Rebind the framebuffer only if we're in the blit path. + * Bufferless path for D3D11/Vulkan + * Get D3D shaders to build (though not working correctly) + * Switch to ShaderWriter. Works in OpenGL + * Unify stencil buffer upload using Draw. Only OpenGL tested yet (shaders need adaptation). + * Implement DrawUP for D3D11 + * OpenXR - HUD support added + * GLES: FramebufferBlit: Don't blit if the target is the current framebuffer. + * Clean up blit/copy feature detection. Don't need fast GPU flags for these. + * Fix crash in shader upscaling. This variable shadowed the one that now lives in Common. + * OpenXR - Estimate world unit scale + * OpenXR - Head orientation in some games fixed + * OpenXR - Refactor the common math methods + * Use BlueToAlpha in more games instead of ReinterpretFramebuffers/ShaderColorBitmask + * Fix Vulkan as well + * Fix for D3D11 and GL, but not Vulkan for some weird reason + * Avoid upscaling textures used for color-to-depth. + * Disable texture filtering when drawing color into depth + * D3D9 fix + * OpenGL fix + * D3D11 fix + * Implement color-to-depth for Vulkan, start implementing for D3D11 + * Add logging for some non-traditional uses of framebuffers. + * Fix small breakage in Vulkan texture decoder + * Fix Macross glitches. + * Vulkan: Don't try to auto-mipmap 3D textures in max quality filter mode + * OpenXR - 6DoF head movement support added + * Fixes UWP building issue after Retargeting Solution to a different SDK version. PS: This won't work when selecting the "latest installed version" tho, so a specific version need to be selected when retargeting. + * OpenXR - Renderer support for axis mirror added + * OpenXR - Use a build variant instead of a module + * CI: fix tag regex to match also shorter tags like v1.13 + * Address feedback + * Check for color/Z clashes + * Track depth buffers separately from framebuffers to track previous use for copies + * Give the mip bias its own uniform flag. + * Possible flicker fix for Macross + * Revert mistaken change in D3D9 fragment shader gen + * Support 3D textures in OpenGL ES, add feature check + * Testfix, cleanup + * Implement 3D texturing in the OpenGL backend too. Assorted fixes. + * Remove some dead code + * Implement 3D textures for D3D9 as well + * Implement 3D textures for D3D11 + * Some safety checks to avoid bad combinations, probably not really needed in practice. + * Quick attempt at fixing the Macross glitch + * Fix the mip bias to take the texture depth and sampling offset into account. + * Implement new discovery of a larger CLUT than we thought, plus non-shared CLUTs for 8-bit indices. + * Initial implementation of 3D texturing through equal-size mips (see #6357) + * CI: make indentation in tarball.yml consistent + * Reporting: Disable when debugger methods used. + * GPU: Small scaling variable cleanup. + * CI: also upload release source tarball on tag push + * Revert debug change + * Unify the GLES LoadTextureLevel with the D3D ones. + * Prepare to convert the GLES functions too + * Share LoadTextureLevel between the two D3D backends. + * Cleanup + * Part 2 of the above + * D3D: Part 1 of breaking out the API stuff + * libretro buildfix + * UWP buildfix + * Simplify CPU upscaling code + * Vulkan stencil fixes + * Make the D3D9 backend handle stencil state in more similar ways to the others. + * More thin3d d3d11 fixes + * thin3d: Make writeMask and compareMask dynamic in all backends. D3D11 needs emulation. + * thin3d: Remove dual sided stencil, not supported by all backends anyway + * Remove some indentation + * Use PrepareBuildTexture from the Vulkan backend too + * Change from maxLevel to levelsToLoad, for better readability. Cleanup. + * Define texture loading plan semantics better, pre-port some from Vulkan + * Vulkan: One step closer to using PrepareBuildTexture + * Update PrepareBuildTexture + * Use PrepareBuildTexture from DX9 backend + * Use PrepareBuildTexture from D3D11 backend + * Break out the preparation part of BuildTexture from GLES + * Additional cleanup + * Get the GLES BuildTexture slightly closer + * Make the BuildTexture of DX9 and D3D11 backends more similar + * Remove indentation, useless code + * Remove the DX API access from LoadTextureLevel + * Remove fake-mipmap logic from LoadTextureLevel + * Minor code simplification + * Remove unused parameters to some conversion functions + * Remove unused useBGRA parameter + * Remove the ReplacedTextureFormat enum + * Remove support for 16-bit replacement texture formats + * Update ru_RU.ini + * readme link + * Windows: Add a simple window message to get the base pointer. + * Update assets/lang/ru_RU.ini + * OpenXR - Switching between 2D/VR fixed + * Update version for 1.13.1 + * Update README.md for 1.13.1 + * Replacement: Cleanup leak on replacement reload. + * Replacement: Clear cache on disable. + * irjit: Correct another PurgeTemps case. + * irjit: Clarify PurgeTemps, guard a couple ops. + * Add confirmation dialog when generating a new Mac address + * OpenXR - VR settings added, better UI detection + * better pt-pt translation + * Fix bug in blue-to-alpha - alpha blending could be on when it shouldn't be. + * Reporting: Avoid crash if socket connect fails. + * Some zero init, add a null check in TextureReplacer. See #15730 + * OpenXR - Move VR folder under Common + * OpenXR - Static added to global variables + * OpenXR - Version info refactor + * OpenXR - Integrate SYSPROP_HAS_FILE_BROWSER + * OpenXR - Automatically switch between fullscreen/VR + * Update version for 1.13 release + * README.md update for the 1.13 release + * Simple hack to hide the search button in settings on small screens. Needs a better solution. + * OpenXR - FoV fix for Quest 1 + * OpenXR - Basic look around implementation + * Build: Update to NDK r21e. + * Switch to using the D3D9 state cache in thin3d, to avoid confusion + * Clarity improvement in dx state cache stencil func + * Remove unnecessary complexity from the D3D9 state cache + * Remove bool that was always true in state mapping. + * irjit: Correct initialization warning. + * irjit: Add test for double clobber in #15713. + * Headless: Ignore __testcompare.png in tree. + * irjit: Simplify more arithmetic to Movs. + * irjit: Add tests for IR passes. + * OpenXR - Enable 6DoF mode when in game + * Revert the x-offset bpp fix temporarily. + * Add GPU stat for number of depth copies per frame + * OpenXR - Code formatting fixed + * OpenXR - Disable HW scaler on VR headsets + * OpenXR - Default device config added + * OpenXR - Hide unsupported menus + * OpenXR - VR code formatting fixed + * OpenXR - Finalize VR renderer interface + * Libretro fix + * Windows fixes + * Remove unnecessary parameters from MakePixelsTexture + * Delete some unused code. + * Separate logging for GetPointer/GetPointerWrite + * Split GetPointer into two versions, to help with const correctness + * OpenXR - Pragma once instead of ifdefs + * Allows "merging" render targets that overlap on the Y access. Fixes #7295 (Juiced 2) + * OpenXR - Code formatting header removal + * OpenXR - Code formatting global variables + * OpenXR - Code formatting { + * OpenXR - Button repeating implemented + * OpenXR - Controller mapping better + * irjit: Prevent corruption on purge temps pass. + * Add UWP Platform to Issue Templates + * (UWP) Update Dummy Manifest + * (UWP) Remove Reference to Deleted File. + * (UWP) Delete Old Key + * (UWP) Building Improvements + * Remove redundant file from packages. + * Prevent Dummy file from appearing in packages. + * Revert "(UWP) Small Building Tweaks." + * Remove a redundant mutex lock in logging. + * Remove redundant file from package. + * Prevent dummy file from being in the package. + * Main screen: Display .ppdmp filenames on buttons in grid mode + * Remove log noise + * Mali driver bug workaround for driver 32+. + * OpenXR - Hacky mapping to make the games playable + * OpenXR - UI controls integrated + * OpenXR - UI scale/resolution fixed + * OpenXR - Input class added + * Osk: Allow upper/lower for all keyboards. + * Debugger: Update to latest web debugger. + * Debugger: Fix SVG content-type for proper display. + * Use the recent Clone Wars fix for Star Wars: Force Unleashed too + * OpenXR - Rendering UI fixed + * OpenXR - Rendering into VR framebuffer works + * OpenXR - Dummy rendering working + * Update zh_CN.ini + * Add a simple compat flag to workaround the Clone Wars issue, #12949 + * Apply leoxxx's spacing changes in zh_TW.ini + * Update zh_CN.ini + * Update zh_CN.ini + * Add Zettai Zetsumei Toshi 3 bypass for softlock on character select screen + * zh_CN translation: Apply spacing rules according to leoxx. See #15660 + * Io: Correct access field on dir listings. + * Config: Keep recentIsos lock owned by Config. + * langtool: Add command move-key to move a line globally from one section to another. Update deps + * Update zh_TW.ini + * Update zh_CN.ini + * Add files via upload + * compat.ini for Mali stencil bug: Add the Suikoden game too + * OpenXR - Initial version of VR renderer added + * Mali workaround - add compat.ini entry for Mahou Shoujo Nanoha A's Portable. + * Config: Cleanup recent on a thread. + * Config: Move recent ISOs access to accessors. + * Io: Show message if files are outside valid ISO. + * irjit: Allow unaligned loads by default. + * UI: Reduce IO for texture ini on dev settings. + * Improve comment, variable name + * Vulkan fragment shaders: Don't opportunistically emit "layout (early_fragment_tests)" + * Slightly improve logging of failed block device reads + * Fixes Kenka Bancho too. + * Tweak the Mali stencil-test workaround to work for Surf's Up. See #15016 + * Dynamic update + * Remove unmotivated check preventing texture ini creation on mobile. (#15647) + * Common: add explicit for BSDs after 3ed74350129c + * OpenXR - VR mode initialization added + * Fix comment + * Double quote + * Be explicit on the possibility of failure + * Use hashtag for RGBA format + * OpenXR - VR framebuffer class added + * OpenXR - Hide options which are not relevant for VR + * OpenXR - SDK added + * Change theme endianess + * Update zh_CN.ini (#15657) + * Add check that makes PPSSPP work on Xbox through the debugger + * Replace Win32 file IO with UWP safe variants and add support for getting drives to UWP build (#15652) + * OpenXR - Quest target into Android Studio project added + * compat.ini: Add R-Type games to IntraVRAMBlockTransferAllowCreateFB. + * Compat: Fix backwards check. + * UI: Add small deadzone for analog speed. + * UI: Clamp analog speed to 0 on opposite value. + * Update zh_CN.ini + * Update zh_TW.ini + * UI: Reset views when analog speed mapped. + * UI: Allow cancel from analog speed mapping. + * Change comment based on new findings. + * UWP: Skip pad name lookup. + * UI: Add analog speed limit mapping. + * Core: Track names of connected pad devices. + * interp: Fix constant typo. + * Compat: Add setting to skip compat flags. + * Reporting: Log when vertex count truncated. + * Updated wiki page key in all translation INIs. + * Remove left over + * Change hack names and edit comments based on feedback. + * Improve the vortex hack to make it safe and slow it down + * Test hack that breaks the vortex timer, it's still wrong + * Add 30 fps version as an option, doesn't suffer from softlock + * oops + * Updated text on AdHoc wiki page button. + * Updated "Use original port" tip message. + * Add Fixed60FPShack for GOW, should work smoother than the ForceMax60FPS. Also more stable than it's CWCheat version, unfortunately because of that, still causes softlock in GOW:GOS vortex stage. + * Disable ForceMax60FPS for GOW:GOS to avoid softlock + * Show a tip message when enabling "UPnP use original port" option. + * Update zh_TW.ini + * Update zh_CN.ini + * Update th_TH.ini + * Hide Hostname progressView when not showing any message to get a better view of the list. + * Slightly reduce Hostname progressView's margin as it took too much space. + * Added psp.gameplayer.club public Adhoc Server. + * Additional game IDs for The Simpsons Game + * Apply the mali workaround for The Simpsons Game. + * Merge pull request #15629 from Halo-Michael/lang + * Add ARM64_NEON compile arch flag + * ffmpeg: fix compile bug + * Mask some part of public IP in the logs for security reason. + * Update zh_CN.ini + * More translation key + * Add numpad enter as a confirm + * Update all + * Remove driver version check for Mali stencil/discard bug. See #13833 + * Update eng lang + * Oops, revert debugging commit + * Workaround for strange driver hang on Raspberry Pi 3. + * Keep ye olde else case + * Avoid including signal.h from CommonFuncs.h + * Allow to repeat a "single" button + * Out of performance paranoia, limit the Mali workaround to known affected games + * Work around the Mali stencil discard bug the same way as the Adreno one. + * Show detected GPU driver bugs in system info + * Increase the difficulty of the GPU driver test + * Build: Use cinttypes not inttypes.h. + * Update zh_TW.ini + * Update zh_CN.ini + * Update gamecontrollerdb.txt + * Update zh_CN.ini + * Update zh_TW.ini + * Fix-ja_jp.ini + * Fix-ja_jp.ini + * Small struct layout optimization (stencil is never more than 8 bits, can sneak in next to the load actions) + * Rename Vulkan attachment operation variables. Prep for supporting specifying store op. + * Inline the CoreParameters accessor (eliminates a function call for each access) + * Shaun White Snowboarding: Use compat flag to eliminate a readback per frame + * Echochrome lines: Remove UV offsets, avoid reading the destination (much better codegen) + * Update zh_CN.ini + * Update zh_TW.ini + * Update zh_CN.ini + * Add centered line drawing for Echochrome. + * Take the absolute value when measuring pixel size for line expansion. + * Bump gradle again, .gitignore update for VS 2022 + * Add TODO to see if we can make better choices of time sources + * add additional rescaps + * fix indentation + * add runFullTrust, broadFileSystemAccess and expandedResources rescaps + * Add guard + * Update zh_CN.ini + * Add search on SDL + * Update id_ID.ini + * Update zh_CN.ini + * Small Change + * Update zh_TW.ini + * Update + * Indonesian Translation Update + * Fix commending wrong lines + * GLES: Fix sw transform flip in skip buffer mode. + * Update defaultshaders.ini + * Add files via upload + * Update upscale_spline36.fsh + * Return by copy + * Log missing lines on shutdown, update lang files + * Fix CpuCore select + * Merge update into master + * Savedata: Allow no key when not using SECURE modes. + * Osk: Fix fullscreen check. + * UI: Abandon focus movement on returning from pause. + * Update zh_CN.ini + * Config: Don't save --fullscreen unless changed. + * Fix themes as well + * Postshader: Avoid treating directories as files. + * Update th_TH.ini + * Update zh_TW.ini + * GPU: Fix skip buffer effects rendering. + * GPU: Hook Sol Trigger func to flush texture. + * Crash: Recover from unaligned CPU access. + * Update pt_BR.ini + * Add files via upload + * Update zh_CN.ini + * Detects connection establishment to AdhocServer by combining select and getpeername to increase accuracy. + * We might be missing this in the past. + * Fix sceNetAdhocPtpConnect issue on Windows (ie. Tales of VS) due to getpeername on Windows can return a success when the other side haven't Accepted the connection yet. + * Fix sceNetAdhocPtpConnect issue on non-Windows (ie. Taiko no Tatsujin Portable DX on Linux) due to select can flags an unconnected socket as writeable & readable on Linux. + * Added a note regarding SO_NOSIGPIPE + * Avoid compilation warning + * Avoid a possibility of fatal signal on non-Windows + * Added a note regarding select on Linux. + * (libretro) Add option to detect and notifiy frontend of internal frame rate changes (60 <-> 30 <-> 20 fps, etc.) + restore 'Duplicate Frames in 30 Hz Games' core option + * Update zh_CN.ini + * Limit minimum timeout to [0, 100, 200, ..., 5000] + * Fix missing GPU and run tools + * For Libretro -DUSE_DISCORD=1 is not added in CMakeLists.txt + * Let port offset cover the entire range in steps of 1k + * Remove options for network chat, since these only work for standalone + * Make MAC address configurable and randomize when not set + * Fix update of chat screen position + * Disable UPnP by default + * Correct indentation and some strings + * Expose networking options in libretro core + * Unused lang tool + * Update Russian translation + * Display: Ignore lagsync when auto frameskip on. + * Tweak -latomic preconditions in CMakeLists.txt. See comments on 14e9aab96e111 + * softgpu: Correct stencil debugging. + * softgpu: Fix viewport flag clean/dirty. + * Lang: Remove some duplicate lines in German. + * uptated German Translation + * Enable -latomic on Linux. Fixes build on Raspberry Pi 3b for me. + * Update .gitignore + * Replacement: Allow texture dump without replace. + * Fix shader blending regression (Vulkan) + * libkirk: Set Segher Boessenkool's original copyright notice + * Fix discrepancy in how the LMODE flag is calculated between VS and FS + * Vulkan: Correct size for tex mip level readback. + * Vulkan: Fix depth transition barrier. + * Remove the rather problematic limit on framebuffer copies + * Remove log output + * Combine barriers before blits/copies + * Convert some barriers to use the new barrier framework. + * Add utility class VulkanBarrier + * Combine renderpass "pretransitions" for barriers. + * Vulkan: Follow sampler best practices with auto max quality. + * Bump vulkan headers to SDK 1.3.211 + * Bump gradle + * Better default control config for the Retroid handheld Android console + * Convert the blueToAlpha bool to a usageFlag. KEEP instead of CLEAR for depth during reinterpret. + * Oops, gotta update the cached framebuffer format in the other drawing paths too + * Finish BlueToAlpha functionality + * BlueToAlpha hack to avoid clearing the envmap in split/second :( + * Add option to redirect blue to alpha if 565 mode is rendered and mask is 0x0FFFFF. + * Move framebufFormat to gstate_c, so we can override it + * Dirty more state after reinterpret + * Minor cleanups in framebuffer manager + * Detect ISO files from file contents if extensions are wrong. + * Allow to set InternalScreenRotation per game + * Logs unimplemented syscalls as error + * Partially fixes SenseMe app (will need HTTP/HTTPS support to be fully working) + * Vero4k doesn't report its NEON support - so check for underlying Arm Cortex-A53. + * GLES: Reverse alphamask when comparing to CLUT. + * Simplify alphasum checking for DXT textures, and fix a regression + * Make the existing ReinterpretFramebuffers/ShaderColorBitmask path work for Split/Second + * Fix lazy loading of metadata + * UWP: Fix loading the default theme. + * Fix regression with CLUT16 textures. See #12188 + * Minor cleanups + * Fix mip level bug in Vulkan texture cache. + * Make UI a bit more descriptive + * Add key bind to hotswap internal screen rotation + * Fix checks for the "fake mipmap" situation. + * UI: Prevent save textures if not replacing. + * Vulkan: Save textures using a memory copy. + * Replacement: Avoid recheck saving within 5s. + * Replacement: Move IO checks to saving thread. + * Vulkan: Allow saving const tex levels. + * Add the slow comment to one more place + * Save textures on background tasks when texture dumping is enabled. + * Fix hang when choosing "Skip" in first-time init. Warning fix, comment add. + * Disable the compat.ini workaround for Crash Team Racing. Reported by Panderner to not be needed anymore. + * Add contributed data about a couple of prototypes to auto-expand memory to 64MB + * UWP crashfix - serialize accesses to each loader + * Crashfix + * Fix typo and optimize locking pattern in Event, make it the same as LimitedWaitable. + * Buildfix + * Try a clang pragma to avoid overeager auto-vectorization + * Buildfixing, correct NEON type usage + * SIMD-optimize CheckMask16 / CopyAndSumMask16 + * Delete a lot of specialized alpha checking code. + * Address feedback + * One more NEON optimization + * Quick NEON optimization of CheckMask32 + * SSE optimize a common case for video + * Yet another bugfix + * Debug log (remove before merge) + * Bugfixes, handle the rest of the cases + * Simplification and some cleanup + * Remove redundant line + * Fix DeIndexTexture + * Comments + * AND together colors while decoding, and then check against fullAlphaMask. + * Fix some NEON code that had bad compile-time checks (and some didn't compile) + * ARM32: Remove a lot of non-NEON fallback paths + * Always use the stable quick tex hash. Doesn't actually make a difference except on new CPU archs. + * Remove more function defines + * Merge separate NEON functions into the normal functions. + * Avoid creating a fileloader during shutdown. + * Make the number of threads flexible in the threadmanager test. Increase to 9, just because. + * Replacement: Avoid potential destructor crash. + * ThreadManager: Improve waitable destruction. + * Fix race condition in LimitedWaitable between Notify and Wait + * Fix misuses of cond.wait (should return true when you want to stop waiting!) + * OK, this does crash + * Time the stress test + * Add a scheduling stress test to TestThreadManager.cpp. + * Bump Android target version to 12L. + * Gradle bump + * Net: Ignore SIGPIPE errors on POSIX systems. + * Switch: Fix GL build typo. + * NetAdhoc: Fix warning. + * Switch: Avoid loading Vulkan. + * Detect number of cores in build script + * Allow to change atlas per theme + * Increase the timeout when the connection to the AdhocServer is already in progress, avoids cutting off the connection if it took longer than expected but almost completed. + * Added a function to get SO_ERROR on a socket. + * Updated getLocalIp function to detects local IP better. + * Use the most recent posix socket id during blocking simulation, since there is a possibility for the socket id to changes after recreated. + * Recreate the socket before attempting to connect again after ECONNREFUSED/ENETUNREACH error, since reusing the socket only works on Windows. + * Caches IP-specific Port Offset to avoid recalculating them. + * Use reference when possible to avoid copying. + * Initialize some socket-related structs to avoid unexpected behavior. (May fix Unknown Source Port issue) + * Shouldn't return as soon as possible when a blocking PtpConnect is getting ECONNREFUSED error, since it should be treated as ETIMEDOUT on non-Windows platform + * Ignore disconnected error on PtpFlush. Fixes disconnection issue on some games (ie. R-Type Command) + * Debugger: Keep separate tag dirty for RAM. + * Debugger: Fast path write tag lookup. + * softgpu: Combine memcpy into single. + * Debugger: Improve tag on copies. + * softgpu: Avoid unnecessary clearMode checks. + * Manual generate Android APK + * Add a check to the elf loader, preventing a crash on bad module info addr + * softgpu: Plug bad leak of bin queue data. + * softgpu: Reduce memory usage on 32-bit. + * softgpu: Enure aligned bin queues. + * softgpu: Avoid some unaligned access on x86_32. + * Loaders: Fix 64-bit conversion warnings on 32-bit. + * Global: Fix some int/size_t conversion warnings. + * Fix grid align + * Add version to game screen + * Short circuit the texture replacement code + * Don't track small wide-stride block transfers + * softgpu: Fix build with basic logging. + * Windows: Correct root vol handling for UNC paths. + * Windows: Create SYSTEM directory early. + * Show game crashed + * GLES: Remove old invalidation code. + * GLES: Add asserts for correct blend usage. + * GLES: Use EXT not GLES3 for dualsrc blend support. + * Debugger: Track unchanged VRAM to avoid copy. + * Also delete vulkan_funcs.hpp, oops + * Debugger: Improve drawing range in softgpu. + * softgpu: Use region as a second scissor. + * softgpu: Correct accounting for pixel center. + * samplerjit: Correct nearest negative texture clamp. + * softgpu: Directly implement rectangle drawing. + * softgpu: Prepare dedicated rectangle path. + * softgpu: Cleanup some now unused state. + * softgpu: Remove incorrect offsetting for X/Y. + * softgpu: Use a const for subpixel screenpos factor. + * softgpu: Remove offset from screenpos. + * Upgrade to Vulkan 1.3 headers (and remove the .hpp files) + * Gradle upgrade again. Remove some redundant version checks + * Allow to set default background color in a theme + * Fix live update + * Make the untextured uniform buffer the same size. + * Disable darkening background, also allow to over saturate up to 2 + * Add UI Tint/Saturation settings + * samplerjit: Fix non-SSE4 bugs in jit. + * samplerjit: Fix non-AVX three-op shift. + * Unhide "Cache ISO in RAM" on ARM platform. Not the right block level. We should probably check for available RAM instead. + * More fixes for themes + * Merge item and button style, remove hightlighed and few other fixes + * Fix disabled bg + * Allow translation + * Add themes to assets + * GE Debugger: Highlight changed matrix values. + * GE Debugger: Highlight changed state values. + * GE Debugger: Track last gstate on stepping. + * Debugger: Allow custom draw and generic lists. + * Debugger: Avoid scroll on breakpoint add. + * Debugger: Cleanup some header pollution. + * softgpu: Track dirty vs really dirty per buffer. + * softgpu: Allow tri -> rect in transform. + * softgpu: Turn more tri strips into rects. + * softgpu: Combine tris to rects with ignored z too. + * softgpu: Cleanup todos on perspective correctness. + * softgpu: Avoid modifying source vertex data. + * Adress feedback + * remove more prebuilt stuff and huge rapidjson testsuite (unused) + * also delete pre-built ffmpeg windows libraries + * Keep comment up to date + * fix version string generation + * fix version number display + * try 2: remove pre-built ffmpeg and vulkan libraries + * remove pre-built ffmpeg and vulkan libraries + * fix mime type of xz tarball + * rename action title + * rename to tarball.yml + * use recursive submodules and checkoutv2 + * Forgot pragma, silly me + * Allow custom UI themes + * GE Debugger: Avoid crash on Step Draw with flush. + * Debugger: Improve memory search result draw speed. + * Debugger: Search scratchpad/ram/vram consistently. + * CI: add github action to build source release tarball with submodules + * GE Debugger: Correct icon list cleanup on shutdown. + * Add missing Config.h include + * GE Debugger: Optimize slab lookup. + * GE Debugger: Skip overhead without breakpoints. + * Generate dockerfile in GHA + * Vulkan: Avoid crash on double DeviceLost. + * Android: Correct truncation behavior. + * GE Debugger: Add filter to skip prim calls. + * Debugger: Avoid write tag lookup on small alloc. + * Debugger: Use detailed meminfo with breakpoints. + * GE Debugger: Allow modify/breakpoint on matrix. + * GE Debugger: Cleanup right click on state. + * GE Debugger: Allow setting secondary values. + * GE Debugger: Add breakpoint icon in state. + * GE Debugger: Cleanup vertex list redraw. + * GE Debugger: Flush downcount when changing lists. + * GE Debugger: Reduce playback command noise. + * samplerjit: Precalculate DXT1/3/5 offsets. + * Delete pauth_tool, useless since pauth keys known + * samplerjit: Reduce DXT5 decode code size a bit. + * samplerjit: Optimize DXT calc using BMI2. + * samplerjit: Avoid RCX gymanstics with BMI2. + * samplerjit: Decode colors with BMI2. + * softjit: Use PEXT to downsample colors. + * softjit: Use BMI2 to speed up dst color loads. + * softjit: Use BMI to simplify some masking. + * x86jit: Use BMI2 for variable shifts. + * softgpu: Avoid flush on END. + * Remove double flush in GE_CMD_END + * GPUCommon: Remove misleading flag FLUSHBEFORE. + * HLE: Fix some type convert warnings. + * Global: Reduce includes of common headers. + * Global: Fix some type conversion warnings. + * Common: Remove some unused files. + * Core: Skip stats reset while stepping. + * Display: Account for stepping time in graphs. + * Display: Move core counters/stats to HW file. + * Display: Move listeners to separate HW file. + * Headless: Improve total test time on Windows. + * Headless: Improve test init time. + * Debugger: Encourage better meminfo accounting. + * Build: Fix Windows unit tests. + * Kernel: Adjust timings of LwMutex and EventFlag. + * Build: Ignore nullable completeness on Android. + * UI: Cleanup some reference warnings. + * samplerjit: Fix non-shared CLUT on Linux. + * samplerjit: Validate compile in a unit test. + * samplerjit: Correct level lookup in nearest. + * samplerjit: Fix reg corruption in DXT funcs. + * samplerjit: Compile sampler funcs together. + * softjit: Add tests for compile success. + * Common: Avoid assert hang in CI. + * softjit: Fix stencil bug running out of regs. + * Build: Enable ccache for GitHub Actions. + * Build: Run unit tests after build. + * Headless: Remove stack corruption workaround. + * Headless: Add flag to run failing tests. + * unittest: Skip D3D shader compile on non-Windows. + * unittest: Correct Windows-specific Path tests. + * softgpu: Skip new CLUT if identical. + * unittest: Use more consistent output executable. + * unittest: Correct failing unit test. + * samplerjit: Reduce register waste. + * samplerjit: Avoid frac uv transfer to gen regs. + * Added another attempt to detect GETIFADDRS availability, which is more common on newer OS + * We might be missing the SIOCGIFADDR + * Added logs for error within SIOCGIFCONF section. + * Change a confusing variable naming + * Expose 2xBRZ texture filter option. + * Mp3: Delay low level init. + * Mp3: Output PCM data using even/odd buffer slot. + * Mp3: Allow decode with NULL out pointer. + * Mp3: Cleanup AuCtx constructor. + * Kernel: Fail threads with invalid priority. + * UWP buildfix + * Quick hack attempt around build failures + * Rtc: Fix day of week stack bounds issue. + * Savedata: Fix indeterminate timing of field update. + * Ge: Fix sceGeBreak error case. + * softgpu: Cull no-pos and through s8 pos verts. + * softgpu: Fix cull in throughmode. + * Vulkan: Add support for listing all the available surface formats in system info. + * samplerjit: Reduce transfers in nearest texel calc. + * samplerjit: Better vectorize UV linear calc. + * samplerjit: Reuse XMM reg for sizes. + * softgpu: Fix pixel ID for invalid blend factors. + * softjit: Fix src blend factor handling. + * GE Debugger: Allow GPU stepping while CPU stepping. + * GPU: Allow relative jumps and calls. + * GPU: Log and report when region1 is non-zero. + * Common: Fix conversion of 4444 to BGRA. + * softgpu: Correct clear mode dither. + * softgpu: Correct simple rectangles with mipmaps. + * samplerjit: Optimize texenv blending a bit. + * softgpu: Correct UV rotation for transformed rects. + * softgpu: Reuse transform state. + * softgpu: Add dirty flags for rasterization state. + * softgpu: Use dirty flags for render overlap checks. + * softgpu: Fix state race on screen offset. + * softgpu: Add a table and initial dirty flags. + * GE Debugger: Add option to auto flush. + * GE Debugger: Cut down on primary buttons. + * GE Debugger: Add a menu to the GE debugger window. + * samplerjit: Oops, forgot about 64 unpack. + * samplerjit: Fix Linux stack corruption. + * Build: Fix some minor warnings. + * Headless: Return an exit code on test failure. + * Common: Print assert failures to stderr. + * softgpu: Define constexpr var for older C++. + * x86jit: Check CALL dist for safe memory funcs. + * samplerjit: Fix some Linux register issues. + * Headless: Fix font tests on Linux. + * Build: Speed up macOS actions build a bit. + * Headless: Update passing tests. + * Clean up hardware tasseletion setting + * Show the software rendering option on all platforms + * softgpu: Avoid flush texturing from stride. + * softgpu: Flush framebuf only on change. + * softgpu: Remove bin asserts. + * softgpu: Use persistent bin task state. + * softgpu: Pack vertexdata a bit better. + * Reducing MatchingEvents processing delays. Fixes timing issue on a few games (Super Pocket Tennis, Thrillville Off the Rails, Lord of Arcana, Gundam: Senjou No Kizuna Portable, StarWars The Force Unleashed) + * Report sceNetAdhocMatchingInit to find out a list of games that use AdhocMatching library to be tested with. + * Discard HELLO Events when in the middle of Joining + * Updated notifyMatchingHandler comments + * Updated MatchingArgs comment + * Added comment related to EVENT_HELLO + * samplerjit: Reduce prolog/epilog spill. + * softjit: Simplify constant writes. + * softjit: Centralize some common funcs from sampler. + * softjit: Switch to constant pool. + * samplerjit: Simplify AVX shift-copies. + * UI: Reset ZIP install errors for new ZIPs. + * softgpu: Guide more SSE light factor handling. + * softgpu: Simplify mask check. + * softgpu: Reduce copying during clipping. + * Debugger: Fix crash in software renderer. + * softgpu: Fix block transfer flush detection. + * softgpu: Use SSE in lighting param computation. + * softgpu: Precompute some of screen space multiply. + * softpu: Premultiply matrix transforms. + * softgpu: Precompute state for vertex transform. + * softgpu: Correct texsize flush annotation. + * softgpu: Heuristic to avoid over-draining. + * softgpu: Reduce wasted memory. + * softgpu: Precompute lighting parameters. + * softgpu: Tune some queue sizes for perf. + * softgpu: Expose flush reasons/times in debug stats. + * softgpu: Reduce flushing for smaller textures. + * softgpu: Flush block xfer only on overlap too. + * softgpu: Decorate some stats for flushes. + * softgpu: Avoid texture/CLUT flush unless overlap. + * softgpu: Flush on offset changes. + * softgpu: Track CLUTs as states for binning. + * softgpu: Allow binning across prim calls. + * softgpu: Fix alpha blend with one/zero. + * softgpu: Correct dither matrix lookup. + * softgpu: Correct logic op state update. + * softgpu: Fix off-by-one size limit on bin queues. + * softgpu: Tune queue push/pop to reduce overhead. + * Debugger: Avoid flushing meminfo on write lookup. + * softgpu: Use cached state for screen offset. + * softgpu: Cache CLUT params in sampler state. + * softgpu: Move texenv color into sampler state. + * softgpu: Move tex size to cached state. + * softgpu: Remove z from DrawingCoords. + * softgpu: Move c++ tex func to sampler. + * softgpu: Use cached sampler state outside jit. + * softgpu: Cache line drawing state. + * softgpu: Use cached state for clear write mask. + * softgpu: Move fixed blend factor to draw pix state. + * softgpu: Cache colortest params in draw pix state. + * softgpu: Cache alpha/stencil test masks in state. + * softgpu: Cache logicOp in draw pixel state. + * softgpu: Use cached write mask in draw pixel. + * softgpu: Cache minz/maxz in draw pixel state. + * softgpu: Cache strides in draw pixel state. + * softgpu: Cache fog color draw pixel state. + * softgpu: Remove disable of cached pixel state. + * softgpu: Avoid double calculating screenpos. + * softgpu: Cleanup push/pop atomic handling. + * softgpu: Remove old task. + * softgpu: Stop storing model pos. + * softgpu: Stop storing normal/worldnormal/worldpos. + * softgpu: Enqueue batches of prims when binning. + * softgpu: Adjust binning thresholds. + * softgpu: Avoid thread ordering hazard. + * softgpu: Move threading into BinManager. + * softgpu: Handle scissor/range in BinManager. + * softgpu: Add class to manage and enqueue for bins. + * Fix buffer overflow issue on GameMode. + * GameMode Scheduler should only runs when there are active Master. + * Prevent creating another GameMode Master, since we don't support more than one Master data, and there is no ID related to Master. + * Fix an issue where more than one GameMode Scheduler can be scheduled. + * Fix an issue where sceNetAdhocGameModeCreateReplica blocked current thread for initial sync while Master data hasn't been created yet. + * Debugger: Cache list PC for softgpu tagging. + * softgpu: Centralize more prim drawing state. + * Updated sceNetAdhocGameModeUpdateReplica (Fix Bomberman Panic Bomber) + * Fix invalid argument issue of sceNetAdhocPtpAccept on Bomberman Panic Bomber. + * softgpu: Move tex filter setting check to state. + * softgpu: Move texture addresses to prim state. + * softgpu: Prepare more state per prim call. + * softgpu: Use func IDs instead of gstate more. + * softgpu: Compute func IDs once per batch of verts. + * Avoid confusion for translator + * Add files via upload + * Vulkan: Avoid crash on bad shader. + * softgpu: Output normals to GE debugger. + * Use System_GetPropertyInt to report the keyboard layout + * UI: Mention SDL in the credits when it's being used alongside Qt + * Update pt_BR.ini + * softgpu: Correct decal doubling without alpha. + * samplerjit: Skip processing alpha if unused. + * Add files via upload + * samplerjit: Fix alpha for 565 in linear lookup. + * softgpu: Skip threading when rendering to self. + * softgpu: Draw top left of rectangles first. + * GPU: Allow gcc/clang/icc runtime SSE4 usage. + * UI: Use comparison screenshot as frame dump icon. + * Headless: Fix compare of smaller buffers. + * softgpu: Use some SSE4 in triangle interpolation. + * softgpu: Limit minX/maxX per line. + * softgpu: Avoid rsqrt path for normals. + * softgpu: Clarify internal matrix multiply usage. + * Remove QWEmct from credits as requested + * softgpu: Retain floats in diffuse/specular. + * softgpu: Cleanup -NAN and diffuse factor. + * softgpu: Correct NAN spotlight exponent/direction. + * softgpu: Correct handling of NAN attenuation. + * softgpu: More closely match PSP light rounding. + * softgpu: Reuse SSE/NEON matrix code. + * softgpu: Use common SIMD matrix multiplies. + * softgpu: Separate calculation of S/T. + * UI: Fix some sign/size comparison warnings. + * samplerjit: Fix standard bufw check. + * samplerjit: Fix DXT compilation. + * samplerjit: Correct linear compile failure. + * softgpu: Correct mirroring in fastpath+nearest. + * softgpu: Improve usage of SSE for lighting. + * softjit: Keep mip S/T calc in SIMD. + * softjit: Use RIP access in color/depth off. + * softjit: Optimize depth range checks. + * softjit: Add describes here too. + * samplerjit: Remove old linear nearest paths. + * samplerjit: Avoid gather if overread could crash. + * samplerjit: Use VPGATHERDD for all types. + * samplerjit: Apply gather lookup to all CLUT4. + * samplerjit: Use VPGATHERDD for simple CLUT4 loads. + * samplerjit: Avoid a couple more copies in AVX. + * softgpu: Skip sample lookup if masked. + * softgpu: Stop specializing on miplevels. + * softgpu: Separately profile verts and lighting. + * samplerjit: Use nearest func in fast path too. + * softjit: Avoid const temp registers. + * sofjit: Correctly poison memory. + * softjit: Small bloom optimization. + * softjit: Optimize common case bloom blending. + * softgpu: Add code for tracking GPU writes. + * samplerjit: Decode colors in parallel. + * softgpu: Reduce interpolation if not needed. + * samplerjit: Implement nearest in jit. + * samplerjit: Refactor and reuse constant pool. + * samplerjit: Update nearest args, temp disable jit. + * softjit: Small optimizations. + * samplerjit: Create a separate fetch func. + * samplerjit: Refactor sampler ID calculation. + * softgpu: Correct off-by-one scissor mask. + * softgpu: Cleanup min/max tri range handling. + * GPU: Avoid spline crashes on bad data. + * softgpu: Skip processing scissored triangles. + * samplerjit: Correct wrong bufw at mip levels. + * samplerjit: Oops, fix Linux mipmap handling. + * PPGe: Disable dither in UI drawing. + * softjit: Fix regcache error when clearing. + * GPU: Fix bone matrix CALL opt corruption. + * Debugger: Correct delayed symbol listbox updates. + * samplerjit: Tiny dependency optimizations. + * samplerjit: Small AVX optimization to modulate. + * softgpu: Skip zero size triangles. + * samplerjit: Use SSSE3/SSE4 in linear filtering. + * x64jit: Add AVX2-only instructions. + * x64jit: Add AVX-only instructions. + * x64jit: Add AVX/AVX2 encodings. + * samplerjit: Pass funcs as an argument. + * samplerjit: Add an alternate profiling method. + * samplerjit: Correct level lookup offset. + * samplerjit: Minor SSE4 optimizations. + * samplerjit: Set stackArgPos_ early. + * samplerjit: Move texture function into jit. + * samplerjit: Lookup both mip tex values. + * samplerjit: Use regcache for linear filtering. + * samplerjit: Calculate mip level U/V/offsets. + * samplerjit: Take texptr/bufw as arrays. + * samplerjit: Handle mipmap width/height in S/T calc. + * samplerjit: Move S/T calculation into jit. + * samplerjit: Add more bits for S/T, skip multiply. + * softgpu: Correct non-SSE T clamp. + * softgpu: Reduce mipmap fraction to 4 bits. + * softgpu: Correct nearest filtering too. + * softgpu: Remove no longer possible report. + * softgpu: Make linear filtering more accurate. + * samplerjit: Handle unswizzled offsets too. + * samplerjit: Calculate texel byte offset as vector. + * samplerjit: Blend linear using integers. + * samplerjit: Accumulate results in an XMM. + * samplerjit: Pass u/v coords as vector. + * Debugger: Don't try to hash invalid RAM. + * Missing stdarg.h include + * Vulkan: Update profiler formatting for this. + * Make the Vulkan GPU log profiler a runtime developer setting. + * Add format string support to Vulkan log-profiler + * jit: Use a hash for invalidate/clear all. + * Vulkan: Disable profiler by default again. + * Improve the check for "slow" GPUs (with regards to tex scaling). See #15238 + * Path: Check for PSP case insensitively. + * Vulkan: Be more restrictive about texture upscaling on "slow" devices. + * GPU: Remove explicit rect/line depth cull. + * Fix text wrapping on PromptScreen by improving the layout + * Vulkan: Don't generate mipmaps for huge textures (probably menu backgrounds). Other minor fixes. + * Oops, accidentally enabled the logging GPU profiler by default. + * Address feedback (preserve alpha-ignoring equality checks) + * Minor fixes + * Turn video upscaling off. + * Optimize the xBRZ texture scaling shaders by 30% on Adreno, less on Mali + * Add hardcoded bool toggle for upscaling video, fix log, remove a comment, minor opt + * fmt + * Add a 2x2 downsampled version of 4xBRZ as 2xBRZ (save memory bandwidth) + * Make a copy of tex_4xbrz + * Debugger: Debounce updates to disasm text. + * Debugger: Never wait inactive on window update. + * Debugger: Remove double debug window updates. + * Debugger: Debounce list control updates. + * Debugger: Fix crash on ill-timed breakpoint update. + * Debugger: Debounce redraws using timers. + * Bump gradle version again + * Fixes when profiler is disabled. + * Switch to macros to easily disable the new profiling. Add missing scope + * Add a simple GPU profiler to profile individual events, rather than full passes. + * Build: Correct AppleClang warning handling. + * softjit: Fix a function arg template warning. + * softjit: Fix an enum type comparion error. + * Build: Avoid compiler flag warning. + * Build: Fix some format truncation warnings. + * interp: Correct vrndf2 range. + * Vulkan: Ignore VMA nullability warnings. + * UI: Clarify error message. + * Build: Remove old-style android/assets/. + * softjit: Skip a common case CMP. + * softjit: Throw away regs allocated in conditionals. + * Split the atlas into three, well two, but two different cases for the fonts. + * Name uniform buffers, add more asserts. Used this to track down the bug fixed in the previous commit. + * Fix an ordering issue causing us to hit some asserts + * Vulkan: Document lifetime for desc set pool. + * Vulkan: Fix use-after-free on shutdown. + * Vulkan: Use new desc sets in Draw and VulkanUtil. + * Vulkan: Refactor common desc pool handling. + * Vulkan: Avoid desc set error in tests. + * samplerjit: Fix DXT1/DXT5 register releasing. + * Vulkan is strict about scissor rect, so let's clamp centrally. + * FramebufferManagerCommon: Fix some issues where currentRenderVfb_ could get out of sync. + * Fix Tekken 5 Dark Resurrection Multiplayer + * ThreadManager: Check for exchange failure. + * Updated GetPtpStat + * Updated GetPdpStat + * Added a note regarding FIONREAD behavior on different platforms. + * Discard PDP packets from an IP that can't be resolved to Player's MAC address (simple version) to prevent confusing the game, since the source MAC won't be updated and may contains invalid MAC. Added TODO to handle unresolvable packets properly. + * Added another returned error code on PdpCreate + * Allow Sending PDP to non-existing MAC address (except invalid mac address). Fixes an issue on NBA Ballers Rebound where players unable to enter Multiplayer Menu again after getting connection issue. + * Vulkan: More scissor dimension checks. See #15207 + * Vulkan: Correct the max level to generate mipmap calculation + * Very minor logging improvement + * Detect and specify the max available Vulkan version, up to 1.2. + * (libretro) Force enable g_Config.bRenderDuplicateFrames + minor audio buffer tweaks + * Use system headers when building with system miniupnpc + * ThreadManager: Use separate pool for IO blocking. + * ThreadManager: Make TaskType a property of a Task. + * Android: Optimize content URI exists check. + * softgpu: Correct alpha on add + invalid texfuncs. + * softgpu: Match texenv blend texfunc accurately. + * softgpu: Make decal tex func more accurate. + * softgpu: Correct accuracy on fast path modulate. + * softgpu: Correct accuracy of MODULATE texfunc. + * (Libretro) Reimplement audio - * Move audio callback to the main thread (fixes the longstanding hang issues when using OpenGL) * 64ms on desktop isn't enough due to excess buffering going on in this emulator. Set default to 128ms instead. * Different audio approach is necessary due to the following: it sets a fixed framerate of 60, then sends either too many or too few audio samples per frame in order to force the frontend to run at the correct speed (.e.g. making the frontend run in slow motion for games that run at 30fps) - this is undesirable for libretro, we want a consistent amount of audio samples per batch. See further comments in libretro.cpp for more information + * softgpu: Template away uncommon mip usage. + * softgpu: Remove useless switch by UV gen mode. + * softgpu: Avoid texlevel check when maxlevel is 0. + * softgpu: Keep arguments in vectors for sampling. + * Debugger: Reduce meminfo block check hazard. + * Debugger: Speed up GetFunctionSize(). + * samplerjit: Refactor nearest using reg cache. + * Common: Correct shm_open() call. + * Psmf: Check for invalid stream/EP map entry counts. + * Vulkan: Cleanup pipeline var shadowing. + * softjit: Make it an error to unlock a temp. + * softjit: Get rid of pointless AGE00 tests. + * softjit: Normalize some stencil test patterns. + * softjit: Avoid tests for greater than 0. + * softjit: Reduce ditherMatrix to 8-bit. + * softjit: Oops, correct allocation typo. + * Update Tagalog (tg_PH) with new translations by Gamemulatorer. Fixes #15189 + * Display the graphics API on logoscreen on more platforms + * ext: Update vulkan headers + * ext: Allow use system glew with system ffmpeg (macOS) + * ext: Support building with system miniUPnPc + * cmake: Generate git-version.cpp in the build dir + * ext/SPIRV-Cross-build: Update cmake version + * softjit: Fix locked stencil reg. + * headless: Enable softgpu jit for tests. + * softjit: Centralize argument register allocation. + * softjit: Fix fog typo causing locking bug. + * jit: Lock around changes to the jit pointer. + * softjit: Fix PixelFuncID arg on non-Windows x64. + * Math3D: Allow construction from NEON vectors. + * sofjit: Move common types to reg cache header. + * softjit: Rename reg cache so it can be reused. + * softjit: Cleanup, add other arch types to regcache. + * softjit: Disallow use of register after unlock. + * softjit: Simplify regcache usage as purpose only. + * samplerjit: Allow disabling linear too, oops. + * softjit: Decide stack offset on compile. + * softjit: Ensure all regs are released. + * softjit: Manage args in the register cache. + * softjit: Fix disabled cache. + * ThreadManager: Protect against teardown hang/leak. + * ThreadManager: Lock notify on global queue add. + * ThreadManager: Use atomic task optimistically. + * SDL: Respect inflight frames setting in GL. + * jit: More atomically update pointer. + * Make CRT animation speed configurable + * softjit: Reduce jit pool size a bit. + * softjit: Precalculate write mask and dither. + * softjit: Log any failed compiles. + * softjit: Keep color 16-bit when useful. + * softjit: Optimize common blend inverse alpha case. + * softjit: Add alpha blending factors. + * softjit: Cache zero vector to avoid recreating. + * softjit: Prefer easier to refill regs. + * softjit: Minor tweak to reg order for XCHG. + * softjit: Implement min/max/absdiff blending. + * softjit: Use SSE4.1 for fog and dither a bit. + * softjit: Fix dithering bug. + * softjit: Optimize typical alpha/depth test. + * softjit: Apply logic ops. + * softjit: Apply color write mask. + * softjit: Refactor color conv to dedicated funcs. + * softjit: Add config setting to enable/disable. + * softjit: Initial color write. + * softjit: Add dithering. + * softjit: Keep x and y args for dither. + * softjit: Depth and stencil testing. + * softjit: Optimize some imm sizes. + * softjit: Add helpers to get framebuf offsets. + * softjit: Add levels of register locking. + * softjit: Add color test. + * softjit: Add alpha test. + * softjit: Initial depthrange, fog. + * softjit: Add register cache for softjit. + * softjit: Switch to the __vectorcall convention. + * softjit: Add stubbed DrawPixel for x64. + * softgpu: Correct alpha blend subtract on negative. + * softgpu: Use KEEP for any invalid stencil ops. + * vertexjit: Optimize 5551 read a bit on x86/x64. + * sofgpu: Disable alpha blend for invalid equations. + * softgpu: Accurately alpha blend. + * softgpu: Use 0 alpha for 565 alpha blending. + * softgpu: Correct tagging of vertexjit. + * softgpu: Fix stencil DECR on 5551. + * softgpu: Use ALWAYS for alpha/depth test in clear. + * Add review suggestion. + * Destroy backbuffer when shutting down vulkan conext. + * PPGe: Prevent special ampersand handling. + * Build with correct preprocessor flag + * Remove the old allocator visualizer. Keep the setting but hide it. Other feedback + * Buildfixes + * Fix bugs in deletion. Delete redundant functions. + * Switch framebuffers to use VMA for allocation as well + * Convert the PushBuffer to use VMA. + * Change the PushBuffer API a bit to not take explicit memory types. + * Remove the VulkanDeviceAllocator + * More integration, use VMA in VulkanTexture + * Start integrating VMA + * Add VMA to ext/ (AMD's Vulkan Memory Allocator) + * Minor refactor + * Comment improvement, restore accidental deletion + * Add a hidden setting to turn off the shader cache, for shader compile performance work + * softgpu: Fix PixelFuncID size. + * Vulkan: Add a single background thread for pipeline creation + * VulkanRenderManager: Add deferred pipeline creation (to get it off the CPU thread) + * softgpu: Add func to tersely name pixel funcs. + * softgpu: Template draw pixel on fb format. + * softgpu: Move draw pixel code to separate file. + * Vulkan: Fix crash on shutdown after device lost. + * softgpu: Clamp and round fog by mantissa bits. + * softgpu: Confirm mask doesn't hit stencil REPLACE. + * softgpu: Force off alpha blend if uselessly on. + * softgpu: Specialize sprite based on pixel func ID. + * softgpu: Use pixel func ID to draw pixels. + * softgpu: Add func to calculate pixel func ID. + * Threading: Remove a level of indirection from Promise and Mailbox. + * On desktop, change an empty current directory to "/" on startup. + * Prevent chat menu touch to fall throght + * Fix crash on close content after cached GPU pointer cleanup. + * UI: Switch to font texture for debug stats. + * softgpu: Clear by dividing out subpixel first. + * softgpu: Add some comments and cleanup. + * softgpu: Tune thread minimums better. + * sfotgpu: Use threads to apply clears. + * softgpu: Combine sliced rectangles. + * softgpu: Split out rectangle path for combining. + * softgpu: Fixup range cull for fans and fast path. + * softgpu: Fast path rectangles as fans. + * softgpu: Use range loops for sprite fast path. + * Vulkan: Don't cache vulkan_ with draw context. + * GPU: Cleanup some extra pointers in fb managers. + * softgpu: Check flat z in fast path. + * softgpu: Fast path triangles without textures. + * Debugger: Fix crash on stepping/break. + * softgpu: Fix leaked range flag on cull. + * Add shortcut for content_uri and native paths in CleanRecent. Saves 150ms. + * Also delay-load the VFPU dialog, just because + * Windows debugger: Load the dialogs on demand. + * Common: UWP buildfix. + * Buildfix + * Use the "new" API SetThreadDescription to set thread names on Windows. + * Last update (id_ID.ini) + * Minor fixes + * Update id_ID.ini + * Update es_LA.ini + * Update zh_CN.ini + * Update id_ID.ini + * Id + * id_ID.ini + * Display: Prevent lock around listener calls. + * Update *id_ID.ini* and some fixes + * On Android 11 and later, let's not fallback to OpenGL at all. + * Add tip to Lazy texture caching button. Replaces #14183 + * softgpu: Correct scissor for pixel centers. + * Headless: Generate a difference highlighting png. + * Headless: Move screenshot compare to class. + * Headless: Fix ppdmp compare. + * Remove ARM Mali blacklisting for 4xBRZ, works now + * Remove unused variable + * Comment fix, remove assert + * Remove support for 16-bit textures from the upscaling shader, to shorten it for benefit of mobile GPUs + * Fix edge conditions in 4xXBZR + * Speed up the 4xBRZ upscaler with 16x and the MMPX one with 4x by not doing redundant work + * Switch texture scaling shaders to a fixed scale model, preparing for the next change. + * Remove the copy-compute-shader path + * SDL: Read assets from Resources on macOS like on iOS + * UI: Support a second texture for just fonts. + * Fix DPAD size calculation + +------------------------------------------------------------------- +Sun Nov 20 00:23:00 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.13.2~git20221119: + * Orient Save State / Load State buttons horizontally + * Debugger: Combine mem info on duplicates. + * irjit: Consistently check vec4 safety. + * irjit: Allow VV2Op SIMD with exact overlap. + * irjit: Correct VV2Op SIMD check. + * GPU: Prefer raster for depth blit if supported. + * Typo fixed + * OpenXR - Cursor aspect ratio + * OpenXR - Accept scrolling from both controllers + * OpenXR - Camera adjust issues resolved + * OpenXR - Force PSP menu aspect ratio + * OpenXR - Better dialog support + * OpenXR - Allow higher 2D canvas distance + * OpenXR - Use axis integration for camera adjust only + * OpenXR - Use mouse control only + * OpenXR - Support axis + * OpenXR - Support camera adjust using joystick + * Update zh_TW.ini + * Update zh_TW.ini + * Update zh_TW.ini + * interp: Support memory breakpoints too. + * interp: Centralize memory size handling. + * interp: Allow breakpoints in release mode. + * interp: Allow resume from breakpoint. + * Debugger: Keep flag for any breakpoints. + * Twinbee Portable: Add config flag to avoid the system languages the game doesn't work with + * softgpu: Use threads on self-render if safe. + * UI: Show standard back button on touch layout. + * softgpu: Fix off-by-one rendering after half-pixel. + * softgpu: Avoid splitting rectangles for fog. + * softgpu: Correct tex on fogged rectangles. + * GPU: Automatically reduce depth range on == test. + * GPU: Remove unused GPU_USE flag. + * GPU: Centralize more GPU_USE flags, like depth. + * GPU: Add a small error-compensation to depth clip. + * Vulkan: Mark Adreno bug resolved on recent driver. + * Vulkan: Correct alpha in Adreno workaround. + * GPU: Convert alphamask to FF/00 for 5551. + * Fix grid align + +------------------------------------------------------------------- +Sun Nov 13 00:21:33 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.13.2~git20221112: + * UI: Avoid enum/float comparisons. + * VR: Remove unnecessary deprecated header. + * SaveState: Switch from deprecated is_pod. + * GPU: Correct Marvel copy hook size check. + * GPU: Hook US version of Marvel Alliance upload. + * OpenXR - Freezing on Pico 4 fixed + * OpenXR - Test mirroring over 30 various games + * OpenXR - Get all mirroring variants + * Revert "OpenXR - Disable range culling properly" + * GPU: Correct equal depth checks. + * OpenXR - View matrix identity check breaking several games fixed + * OpenXR - Don't get mirroring from identity matrix + * OpenXR - Do not apply head rotation on identity matrix + * GPU: Correct vertex decoder in software transform. + * Mpeg: Correct YUV order from decode. + * Mpeg: Fix ConvertToYuv420 return value. + * GPU: Avoid enabling depth test pointlessly. + * GPU: Prefer scaling depth to 16-bit if using 24. + * GPU: Avoid clears for non-simple depth values. + * TexCache: Cleanup BGRA flag dirtying. + * Fix GLSL compilation issues on really old Adreno drivers. + * GPU: Avoid self memcpy(). + * OpenGL ES: Crash as early as possible if things are bad + * Update zh_TW.ini + * Space added + * TexCache: Correct 16->32 for CLUT4 with shift. + * TexCache: Fix 16->32 colors with CLUT start pos. + * TexCache: Align expandClut buffer, cleanup sizes. + * Restore parallel GLSL builds + * Lift length limit on Android logging + * Log source code for failed pipeline creations + * Avoid drawing with failed-to-create pipelines + * Make Mailbox handle null values. Fixes hang. + * Output shader descriptions directly in the source code + * Improve VkResult logging + * Warning fix in VR code + * OpenXR - Disable range culling properly + * OpenXR - Allow more extreme camera adjusts + * OpenXR - GTA Liberty city stereo fixed + * Properly buildfix GL/VR again + * OpenXR - Camera sliders removed from settings + * Fix sky plane fix (broke in my last VR refactor) + * OpenXR - Camera adjustment using PSP keys + * OpenXR - Keep status of PSP keys + * Restore assert + * Experiment (Vulkan): Temporarily disable GLSL parallel compilation on Android. For #16341 investigation. + * Disable remaining false positive. + * Turn off boundSecondaryIsInputAttachment_ when we don't have one. + * Vulkan: Use the very same view as input attachment and color attachment, not just the same image + * Update zh_TW.ini + * Update zh_CN.ini + * Remove unnecessary assert + * OpenXR - Move configs out of VRRenderer + * OpenXR - Move matrix calculation out of VRRenderer + * OpenXR - Simplify projection matrix creation + * OpenXR - Matrix indexing fixed + * OpenXR - Get rid of separate projection matrices + * OpenXR - Get rid of ovrMatrix4f structure + * Partially rescue translation by copying from Hack Settings + * Setting should be reported + * Blind libretro fix + * Android fix + * Blind Qt fix + * lang: Remove obsolete keys, add new untranslated ones :( + * Change "Simulate block transfer" to "Skip GPU Readbacks". Group the speed hacks together + * Change "Rendering Mode" to just a "Skip buffer effects" checkbox. Reuse translations. + * libretro fix, more games + * GPU: Correct softgpu curve issue. + * GPU: Always skin in decode for software transform. + * softgpu: Always use software skinning. + * GPU: Purify vertTypeIsSkinningEnabled(). + * Windows ARM/ARM64 buildfix + * Remove now-unused lang keys + * Change the "Retain changed textures" option into a compat.ini option. + * GPU: Use skinned position always in bounding check. + * GPU: Cleanup GetVertTypeID() usage. + * Linking fix for normal Android devices + * UWP build fix. Had to reshuffle some stuff and compile some Vulkan files. + * Add a dynamic loader for OpenXR so the VR code can link. Make it all build on Windows. + * Typo fix + * Make a lot more code VR build for all platforms + * Fix type of predictedDisplayTime + * 6dof fix + * Rename 6DOF_PRECISE to HAS_UNIT_SCALE + * Remove ugly float conversion hacks from VR config + * Finish previous buildfix... + * Upgrade gradle, specify NDK version in android/build.gradle + * Yet another gradle upgrade + +------------------------------------------------------------------- +Sun Nov 06 00:22:09 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.13.2~git20221105: + * Comment fixes (feedback) + * Fix parsing ampersands in server listings for streaming + * Drop some D3D11 screen rotation support (was only used on Windows Phone) + * VK/D3D11: Move the rarely used "u_rotation" uniform to the frame uniform buffer. + * Switch RenderPassType to be a "proper" bitfield enum. + * Support Android file intents with up to 3 dots in path + * Code formatting fixed + * OpenXR - Get platform flags in runtime + * OpenXR - Introduce platform flags + * OpenXR - Fix ifdef ANDROID + * OpenXR - Add an option to disable 72Hz update + * OpenXR - Enable camera movement and stereo in flatscreen mode + * [Libretro] Update core options to v2 + add a few new ones + * Build: Use Makefile for libretro. + * Build: Enable libzip on libretro. + * GE Debugger: Respect unchanged VRAM each frame. + * GE Debugger: Try harder to identify unchanged VRAM. + * GE Debugger: Dump textures only if possibly used. + * GE Debugger: Restrict marking of dirty VRAM better. + * Vulkan: Fix descriptor set binding in geometry shader generator. Should fix #16315 + * OpenXR - Settings code cleanup + * Replacement: Avoid issues with c++14 data(). + * UI: Install textures as a zip if supported. + * OpenXR - Enable stereo for Split/Second + * OpenXR - Workaround for not working inputbox + * Memory initialization fixed + * OpenXR - Let users set the movement length + * OpenXR - Option to disable controller mapping + * OpenXR - Map controller motions on keys + * irjit: Correct prefix validation. + * interp: Correct dprefix accuracy for vrot. + * irjit: Fix unordered float compares. + * Replacement: Don't double count memory usage. + * Replacement: Allow use of textures.zip for Android. + * Replacement: Lock around shared data access. + * Replacement: Cache file data across cache keys. + * Replacement: Log only missing explicit filenames. + * Replacement: Decimate aggressively with high usage. + * Replacement: Reduce IO checks on startup. + * Replacement: Avoid needless exists check per tex. + * interp: Correct vscl/vmscl t prefix handling. + * irjit: Handle vrot overlap more correctly. + * SaveState: Restore replacements in only one place. + * Debugger: Allow currently-invalid memory reference. + * Debugger: Fix input.analog.send validation. + * UWP: Avoid some float conversion warnings. + * Windows: Avoid compile param conflict warning. + * Draw: Correct ambiguous comparison overload. + * typo fixed + * OpenXR - Add new virtual key for camera adjust + * OpenXR - Camera side adjust added + * OpenXR - Camera height adjust added + * OpenXR - Do not process key actions during camera adjust + * OpenXR - Allow updating camera in realtime + * OpenXR - Make VR camera parameters floating numbers + * Update zh_TW.ini + * Update zh_TW.ini + * Update zh_TW.ini + +------------------------------------------------------------------- +Sun Oct 30 00:22:37 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.13.2~git20221029: + * OpenXR - Stereo support for Outrun 2006 added + * Vulkan: Log additional queue command. + * armips: Update to latest. + * headless: Update tests. + * Vulkan: Fixes validation errors with "null" textures + * Address feedback + * Need to request the proper type of uniform buffer descriptors. + * Add some extra checking for valid stereo shader. + * Actually bind a global uniform buffer, too. Not yet used. + * Vulkan: Reserve descriptor set 0 for frame-global data, move everything else to set 1 + * jit: Reduce some include pollution. + * Fix UB + * Format string fix + * Shadergen fix for OpenGL + * Crashfix in settings menu (Android only, for whatever reason) + * Fix validation issue with clears, remove redundant code + * Clean up GPU_Vulkan::BuildReportingInfo a bit + * Improve compatibility checks for stereo rendering. + * Fix for stereo with textureProj (emulated projection). + * Fix depal texture binding bug + * Name more image views. Very useful with a little patch to the Vulkan debug layers. + * Fix rendering in non-stereo mode + * Pass shadergen tests + * Address feedback + * Stereo now works through shader depal too + * Fix more cases, GTA works now. + * Use arrays more consistently. + * Always use array textures for framebuffers in Vulkan for simplicity. + * wip + * Minor shaderwriter refactor + * Stereo on mobile fixes + * Some refactoring of framebuffer views, layer issues, more work. + * More work. Things are starting to work now. + * Add a couple of stereo display shaders, add infrastructure, compile stereo pipelines + * Non-stereo bugfix + * Start stereo postshader work + * More multiview work + * Misc multiview hackery + * thin3d: Replace hint at future MRT support with basic multi layer support + * OpenXR - Lego Star Wars III stereo support added + * OpenXR - 6DoF and stereo in Lego Star Wars II fixed + * Build: Fix Android/macOS ccache. + * Build: Update deprecated actions. + * headless: Update tests. + * GE Debugger: Correct UV display with prescale. + * Minor ShaderWriter refactor, prep for later changes + * GE Debugger: Allow reading tex on D3D11. + * Vulkan: Use v2 feature checks, and add check for multiview features. + * Testfix + * Minor refactor of ShaderWriter flags. Extracted from the stereo PR + * softgpu: Better approximate slope mip level mode (#16276) + * x86jit: Correct vh2f NAN handling (#16275) + * headless: Update tests. + * VR: Always pass the bounding box check. + * GPU: Account for scissor/viewport in bound test. + * GPU: Respect depth clamp in bounding box check. + * GPU: Allow/use indices in bounding box check. + * GPU: Correct large morph vertex advance. + * GPU: Consume cycles in bounding box check. + * GPU: Correct bounding box for larger counts. + * GE Debugger: List verts on bounding box command. + +------------------------------------------------------------------- +Sun Oct 23 00:23:54 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.13.2~git20221022: + * Make it possible to specify extensions for Vulkan in the ShaderWriter + * Remove duplicate line in libretro makefile. + * OpenXR - Wrong condition removed (#16271) + * OpenXR - Refactor compatvr.ini file (#16270) + * Apply Kyria's updates to tr_TR (Turkish) translation + * Minor refactoring in PresentationCommon + * PresentationCommon: Use refcounts to simplify shader module management + * Vulkan: Fix frame ordering issue with postprocessing shaders + * GPU: Respect matrix and reverse flag w/o normals. + * softgpu: Multiply prev normal by world matrix. + * softgpu: Respect negate normal flag without norm. + * Update zh_TW.ini + * Update zh_CN.ini + * OpenXR - Camera distance adjust added + * OpenXR - Rearrange VR settings + * Update zh_CN.ini + * Some more codestyle cleanup + * Change wrap_mode to clamp for bicubic upscaler + * Libretro buildfix attempt again + * Remove "attachment" parameter from BindFramebufferAsTexture everywhere. + * Libretro buildfix attempt + * Vulkan: Correct some enum switch warnings. + * Kernel: Fix reported StopThread error. + * headless: Update passing tests. + * headless: Fix crash running some tests on Vulkan. + * headless: Allow screenshot to not be 512x272. + * SDL: fix glew on wayland by ignore glx + * OpenXR - Use only the first projection matrix in frame for mirroing detection + * OpenXR - Ensure scene analyze is called the same way as before + * Build/warning fix + * Remove newly added dependencies on PPSSPP + * Don't use PPSSPP-specific stuff in Common/VR for now. + * Buildfixes + * Android non-gradle buildfix + * CMakeLists.txt fix + * Make as much as possible of the VR code build on all platforms + * Use GPU "use" flags to replace IsVRBuild in the renderer. It remains elsewhere. + * Reorder the GPU USE flags a bit + * OpenXR - Move VR passes count calculation + * OpenXR - Reduce uniform calls + * OpenXR - Simplify projection matrix analyze + * OpenXR - Get rid of VR tweaks + * OpenXR - Move matrix composing out of VR tweaks + * Two more renames to make things read better + * Rename GPU_SUPPORTS_ to GPU_USE_ + * Rename gstate_c.Supports to gstate_c.Use + * GPU: Replace logic ops with blend for simple cases. + * softgpu: Correct linear interp for uneven positions. + * softgpu: Correct drawing outside TL of rectangle. + * softgpu: Make triangle fan rect detection generic. + * OpenXR - Add VR mode for stereo flat screen rendering + * Savedata: Update filelist on file erasure. + * Kernel: Stop reporting invalid semaphore names. + * Kernel: Stop reporting invalid mutex names. + * Code cleanup, optional param renamed: "recreate" -> "remove" + * Fix save states + * OOP + * know unknown in sceMp4Create + * Add draft mp4msv module + * Kernel: Add reason to reported stop error. + * Windows: Improve reporting of WASAPI errors. + * Io: Fail open earlier from ms0: w/ no access flags. + * jit: Run invalidates immediately. + * jit: Consistently check range on invalidate. + * jit: Ignore zero byte icache invalidates. + * Removed the bugged part of hle.func.scan + new comments + * OpenXR - Disable VUlkan for now + * OpenXR - Create VR framebuffer for Vulkan + * OpenXR - Hookup VR API into Vulkan + * Comments updated, default value for parameter "recreate" set to false + * Added optional parameter to hle.func.scan, improved code of hle.func.removeRange + * OpenXR - Vulkan support in progress + * OpenXR - Initial work on Vulkan support + +------------------------------------------------------------------- +Sun Oct 16 00:17:05 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.13.2~git20221015: + * D3D9: Correct scissor state cache in Draw. + * headless: Update tests. + * softgpu: Cull a triangle with all negative w. + * softgpu: Fix crash on screenshot w/o display. + * Vulkan: fix wayland swapchain size 1x1 + * typo + * OpenXR - Ensure the app gets unloaded when killed by OS + * Revert "Vulkan: Create depth/stencil buffers on demand" + * Add ShaderId utility function to be used for some sanity checking. + * Add simple way to add debug annotation in the middle of the command stream. Vulkan-only. + * GPU: Respect stencil state in shader blend. + * Followup to #16205, fix one more instance of the problem. + * Removes the option, autodetect instead - only enable if the GPU doesn't support bitwise ops. + * ArmJit: Save/restore downcount where needed, we missed a few cases. + * Revert "Disables "ForceCheck" on jit invalidation on ARM32, introduced in #16194" + * Touchpad/mousewheel scrolling acted very weirdly on Mac due to 0 deltas. Make it work better. + * Disables "ForceCheck" on jit invalidation on ARM32, introduced in #16194 + * Warning fixes + * GE Debugger: Display if tex is framebuf. + * GE Debugger: Use a modern GL context. + * GE Debugger: Remove ancient unecessary GL calls. + * GE Debugger: Normalize framebuffer texture preview. + * GE Debugger: Centralize current fb tex fetch code. + * D3D9: Remove block transfer code overrides. + * D3D9: Implement CopyFramebufferToMemorySync(). + * Vulkan: Fix initial depth copy w/o existing depth. + * MIPS: Fix non standard layout offsets. + * GLES: Add debug readback of stencil data. + * GLES: Refactor depth pipeline create. + * GPU: Rename stencil write pipeline. + * GLES: Allow stencil texturing if supported. + * it_IT - Italian localization: Apply giona888's latest changes + * Experimentally revert "Delete some misguided FPU control code, that we definitely don't want" + * Comment about VR uniforms + * Don't use fragmentShaderInt32Support as a replacement for checking for bitwiseOps + * Don't expand alphaColorRef to 128 bytes on backends where we don't need to. + * glslang: Set a missing resource limit. Avoids spurious errors in shader unit test + * Revert "Revert "Loader: Validate offsets and truncation in ELF."" + * Update elfSize to uncompressed elf size when needed. Fixes infinite loading in Wipeout. + * Revert "Loader: Validate offsets and truncation in ELF." + * Track min and max GPU time, init time, render CPU time. + * Delete some misguided FPU control code, that we definitely don't want + * Vulkan (trivial): Fix numDraws stat when merging render passes. Shorten a name. + * GLES: Simplify, enable debug depth readback. + * GLES: Use Draw for depth readback shader. + * GLES: Enable EXT_depth_clamp where supported. + * GLES: Refactor depth shader download. + * GLES: Depth download cleanup. + * Windows: Retain log console on graphics restart. + * jit: Defer invalidations made while running. + * GE Debugger: Reset on shutdown. + * Loader: Get rid of overalloc on module exec. + * Loader: Validate offsets and truncation in ELF. + * D3D11: Correct depth readback. + * Savedata: Cache SFO data during processing. + * Savedata: Skip and defer texture exist check. + * Savedata: Combine file info and dir listing calls. + * Io: Provide directory existence with listing. + * Android: Differentiate empty dir from failure. + * Savedata: Reduce some duplicate file info lookups. + * Font: Reduce internal font loading IO. + * Io: Defer GetFileInfo() until it's needed. + * Io: Reduce operations inside GetFileInfo(). + * Module: Cleanup file read pattern. + * GPU: Rename readback and buffer write operations. + * D3D9: Correct depth readback. + * Depal from dynamic CLUT: When detecting bounds, be more conservative. + * Vulkan: Create depth/stencil buffers on demand + * Fix depal bounds with dynamic CLUT. Fixes lens flare glitches in Ridge Racer + * Remote API: hle.func.removeRange added + * GLES: Hook up depth download. + * Vulkan: Disable geometry shaders for Mali <= 18. + * GPU: Correct sizing account on block transfer. + * GPU: Hook Gods Eater Burst avatar read. + * GPU: Consider depth buffers in block transfer. + * Jpeg: Cleanup debug dump file access. + * Jpeg: Encode and decode YCbCr as 2x2. + * Jpeg: Correct some logging of return types. + * Debugger: Fix off-by-one in func scanning. + * GPU: Refuse to set fb_address == z_address. + * GPU: Ignore depth when masked and ALWAYS. + * GLES: Use GL_LUMINANCE on GLES for indexed tex. + * D3D9: Add simple rendered CLUT handling. + * D3D9: Correct sampler states for binding 1+. + * Jpeg: Correct colorspace conversion funcs. + * Jpeg: Decode buffers using configured stride. + * Jpeg: Improve error handling for YCbCr decode. + * Jpeg: Retain previous width/height. + * Jpeg: Improve timing/errors for GetOutputInfo. + * Jpeg: Play it safe with load/unload handling. + * Jpeg: Validate CreateMJpeg() parameters. + * Jpeg: Cleanup Init/Finish error checks. + * Fix assertion issue when playing Cars over public adhoc server. + +------------------------------------------------------------------- +Sun Oct 09 00:16:36 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.13.2~git20221008: + * OpenXR - Revision of VR rendering + * Psmf: Fix save states. + * Fix as @hrydgard suggest + * Typo fix, thanks iota97 + * Shader uniforms (VK/D3D11): Fix issue where we could overwrite the fourth component padding. + * GPU: Clip against neg Z even w/o cull support. + * GPU: Verify generated shader buffer length. + * GPU: Optimize clip distances needed. + * Vulkan: Correct clamped Z clip when clipping neg Z. + * Vulkan: Avoid max_vertices=12 if unnecessary. + * Vulkan: Clip clamped depth in geometry shader. + * Forget remove the include + * Store crc change to u32 + * Reporting: Always report hleReportDebug. + * Reporting: Avoid accessing ticks before start. + * Vulkan: Generate indices in clipping. + * Compat: Remove DisableRangeCulling. + * Vulkan: Clip to neg z in the geometry shader. + * Improve Freedesktop integration * Move scalable icon install from `${prefix}/share/pixmaps` (`pixmaps` has been deprecated long time ago) to `${prefix}/share/icons/hicolor/scalable/apps` * Add `Keywords`, `Categories`, and `MimeType` keys to `.desktop` file * Add shared MIME-info on [`application/x-compressed-iso`](https://en.wikipedia.org/wiki/.CSO "Compressed ISO") MIME type + * Try to build fix + * Fix Kurohyou again. See #9576 + * GPU: Skip matching a framebuf for RAM. + * GPU: Use flags to fix triggered upload/download. + * Add crc record to psmfplayer + * Add crc record into mpeg module + * Add crc calculation in loading module + * Vertex decoder: Add fallback for non-SSE4.1 + * Fix another crash with non-buffered rendering + * More libretro buildfix, thanks stuken + * Attempt to fix libretro build + * Fix validation error with the GPU profiler + * Improve Vulkan profiler (show framebuffer tag for render passes) + * Fixed syntax accessing method reviewed by Unknown W. Brackets + * softgpu: Minor opt, ignore unused z_stride. + * GPU: Normalize framebuf addresses. + * GE Debugger: Ignore mirrors for target in record. + * GE Debugger: Correct mask in target breakpoints. + * GPU: Mask away unused bits in framebuf/zbuf ptr. + * Vulkan: Cleanup unused geometry shader vars. + * Vulkan: Allow configuring geometry shaders on/off. + * Vulkan: Block geometry shaders on older Mali. + * Vulkan: Use geo clip distance only where supported. + * Vulkan: Enable geo shader for culling. + * Vulkan: Correct geometry shader culling. + * Vulkan: Cull in geoshader, hack to on for now. + * Vulkan: Basic geoshader code generation. + * Vulkan: Update shader cache format for geo shaders. + * Vulkan: Add geometry shader ID tracking. + * Vulkan: Clean up shader module tag. + * GPU: Add dirtying for geo shader state. + * UI: Prevent rendering during reboot. + * assets/lang: translate some key phrases in pl_PL.ini + * headless: Update tests. + * GE Debugger: Fix memcpy/memset recording. + * GE Debugger: Record the Edram translation value. + * Ge: Report and save Edram translation value. + * Debugger: Fix game.reset by adding a reboot status. + * Build: Ignore no symbols link warning. + * Build: Fix debug build on Windows 32-bit. + * Tighten up some color format checks with displays and copies + * Better readability and optimization insertion into container by replacing 'insert' -> 'emplace', 'push_back' -> 'emplace_back' + * Fixed bug and memleaks + * Code readability, vec reserve() and remove excess c_str() + * Correct cleaning string and remove unused vars + * Forget remove unused include + * Change to mpegLibVersion + * Trying to improve sceMpegAvcDecodeYCbCr + +------------------------------------------------------------------- +Sun Oct 02 00:17:05 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.13.2~git20221001: + * GPU: Add geometry path for shader writer. + * unittest: Cleanup stage/glsl version. + * Draw: Add COLOR1 semantic. + * ES2 crash fix: Don't draw depth if lacking fragment shader depth write. + * Build: Raise macOS/iOS min version for c++17. + * armips: Use native std::filesystem on Windows/UWP. + * armips: Also update UWP and ARM64, etc. + * armips: Update to UTF-8/c++17 armips. + * Core: Remove disallowed register. + * Build: Update zstd to latest dev. + * armips: Update to latest. + * GE Debugger: Correct rounded coords in vertex list. + * softjit: Fix dst blend shift. + * softgpu: Avoid projecting textures in common case. + * GPU: Fix missed dirtying when fast loading tgen. + * GPU: Restore matrices with dirtying. + * Use compat.ini to avoid readbacks in Silent Hill: Shattered Memories as well. + * Remove redundant call to SetMenu + * Add hidden ini file setting that you can use to hide the menu bar. + * Typo fix + * Update README.md + * OpenXR - Add README.md about the SDK + * Update the Vulkan debug names when reassigning depth buffers. + * Add Silent Hill: Origins to compatibility setting BlockTransferAllowCreateFB. + * If a framebuffer starts using a different depth buffer than before, re-point. + * Quiet the Vulkan miniprofiler (for texture uploads etc) a bit + * softgpu: Correct matrix value update wrapping. + * Ge: Move matrix reading into GPU. + * GLES: Fix colortest/logicop uint/int conversion. + * Debugger: Add API to scan memory for funcs. + * Debugger: Fix doc typo for disasm/assemble names. + * OpenXR - Fix FBO colors on Pico + * OpenXR - Fix the build on Quest + * OpenXR - Support for Pico controllers + * Vulkan: Better debug names for RENDER passes. + * Remove alpha ignore in xbrz texture shaders. + * OpenXR - Initialization on Pico fixed + * OpenXR - Code formatting + * Better bit scrambling when computing draw call IDs for vertex cache. + * softgpu: Handle rectangle texture projection. + * softgpu: Implement triangle texture projection. + * softgpu: Expand texture coords to include q. + * softgpu: Split clippos out of rasterization vert. + * GPU: Correct flat normal projection mapping. + * GPU: Correct normalized zero normal proj map. + * softgpu: Save last tc/normal in vertex reading. + * Fix color test. + * Add compatibility flag for loading pixels on framebuffer create using nearest filtering + * Mpeg: Correct Au update size. + * GPU: Apply color test mask as a uint. + * GPU: Consistently use uvec3 for colortest. + * ivec->uvec, comment fix + * ShaderUniforms: cleanup, put every "4-float" on a line for clarity + * Shuffle constants around, squeezing them into gaps. Saves another 16 bytes. + * Fragment shader uniforms: Pack color mask in 32 bits instead of expand to 128 bits. + * Fold the "materialUpdate" flag into the light ubershader part. + * Add some recent flags to FragmentShaderDesc + * Remove suspicious dirty flag + * Prepare for dynamic mat update + * Minor debugger fix + * Always do the vertex shader part of the fog computation. + * GPU: Fix SSE4 Vec3f normalize. + * Oops! Also, testfix + * Fix light ubershader for D3D11 and OpenGL, GLES unsigned/signed stuff + * Experiment: Generate "Ubershaders" that can handle all lighting configurations + * Replace faulty layout transition deduplication algorithm. + * Vulkan: Add debug names to our command buffers, so you can see which one errored in validation + * Minor cleanup in TinySet + * (UWP) Remove expandedResources + * Remove debug log. Add some new debug log though, unrelated to this PR, for fb clut + fb texture. Plus a couple asserts. + * OpenXR - Pico specific code refactor + * softgpu: Avoid waiting for a thread to drain. + * softgpu: Increase queued prims. + * softgpu: Fix triangle strip with partial rects. + * softgpu: Simply 5551 blending fast path. + * softpu: Refactor out 5551 fast path checks. + * softgpu: Avoid fast path in another wrong case. + * D3D9: Allow INTZ depth buffers more correctly. + * Vulkan: Avoid allocating depth images for stuff like temp copies, depal buffers etc. + * Bump the shader cache version + * Vulkan: Don't compile pipeline variants that don't make sense given their flags. + * OpenXR - Initialize Pico subsystems + * OpenXR - Disable performance extension on Pico + * OpenXR - Support multiple loaders + * OpenXR - Add manifest for Pico + * OpenXR - Rename Quest build variant + +------------------------------------------------------------------- +Sun Sep 25 00:17:45 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.13.2~git20220924: + * GE Debugger: Add option to track pixel in preview. + * GE Debugger: Copy images optionally with alpha. + * GE Debugger: Add a direct copy for preview images. + * OpenXR - PreprocessStep function implemented + * OpenXR - Simplify skyplane detection in GTA + * OpenXR - Get rid of an ugly if condition + * GPU: Cleanup unused CheckAlpha() funcs. + * softjit: Skip reading dst pixel where blended out. + * softgpu: Optimize rectangle sampling/blending. + * softgpu: Ignore a needless color test case. + * GE Debugger: Tag frame dump replay VRAM writes. + * GE Debugger: Fix small tex/clut recopying. + * HLE: Fix bad logging without latestSyscall. + * Vulkan: Avoid crash on two backbuffer steps. + * Remove ill-formed asserts + * More tweaks, fix crash on exit (double-join thread) + * Small tweak + * Can't seem to get around this condition variable... Oh well. + * Readback fix + * Greatly simplify the thread synchronization. + * Flatten the Run function, to make it easier to reorganize. + * Make functions private that should be private + * Best practices layer warning: Don't pass in unused clear values. (probably zero impact, heh). + * Handle depth buffer write-after-write hazards in the queue runner + * Fix leak of shader modules from thin3d. + * Another uninitialized variable (VAI minihash/hash) + * Make valgrind happy + * Fix memory leak (of Promise objects) in VK pipeline wrappers + * Add Burnout Dominator to list of games that require buffered rendering + * GPU: Upload depth only on first usage. + * softgpu: Fix self-render detect in Ridge Racer. + * softgpu: Cache reused indexed verts. + * softgpu: Skip matrix multiply for fog factor calc. + * softgpu: Remove unnecessary state param. + * Texture/Framebuffer match: Ignore stride if texHeight == 1. Fixes Ridge Racer lens flares. + * Use Unknown's idea from #16081 instead of the UploadDepthForCLUTTexture compat.ini flag. + * Fix Vulkan regression of #16075 due to silly typo. + * Fix corruption of Ridge Racer speedometers with AutoMaxQuality enabled. + * Cleanup logic + * Fix for render pass merge + * If depth or stencil are cleared in a renderpass, set the pipeline flag. + * Vulkan: Don't have renderpasses store/load depth buffers when we don't use them + * Preserve depth buffer on framebuffer resize, if has been used. + * More lenient heuristic for uploading depth buffers. Still behind compat flag. See #11100 + * Minor cleanups around dirtying of render state + * softgpu: Avoid unnecessary flushing for curves. + * GLES: Avoid resizing tessellation data textures. + * GPU: Correct neg normal for hwtess texgen. + * GE Debugger: Default auto flush enabled. + * GPU: Use hardware tess params for texgen. + * Dirty more state after depal + * Don't try to replace or scale CLUT8-on-GPU textures. + * Shouldn't 'floor' there + * CLUTs can be loaded from small rectangular textures. Need to linearize. + * OpenXR - Stereo mirroring fixed, disable stereo when no world scale defined + * Vulkan: Avoid crash in headless on finish. + * Kernel: Match index lookup behavior for tls. + * GPU: Simplify depth clamped clip planes. + * GPU: Clip clamped depth accounting for perspective. + * GPU: Account for w properly in lines, fixing width. + * softgpu: Narrow blend check for fast path further. + * headless: Use display buf for compare screenshot. + * GPU: Correct display framebuffer reading. + * softgpu: Fix display framebuffer read. + * Fix BGRA issue on D3D with GPU CLUT textures + * softgpu: Correct clear rect off by one issues. + * OpenXR - Compile fix + * UI: Fix controller autoconf list result. + * headless: Update tests. + * Kernel: Disallow partitions 8/10 to match tests. + * OpenXR - Support for all headset's IPDs fixed + * Kernel: Allow volatile for MsgPipe buffers. + * Kernel: Allow volatile for Tlspl objects. + * Throw in assorted warning fixes + * Kernel: Allow kernel/volatile VPLs as well. + * Submit main command buffer before acquire. Fix sync bug. + * When taking screenshots, make sure that the readback command happens on the right cmdbuf. + * Kernel: Allow allocating FPLs in volatile memory. + * Minor code cleanups (hasBegun is now redundant, for example) + * Rewrite some present logic for slightly more clarity. Fixes bad logic and a minor race condition. + * Various refactoring + * Unify depth texture and framebuffer fetch checks + * Unify range culling detection + * Start unifying setting of the GPU feature flags, now that thin3d has feature detection. + * Kernel: Load volatile memblocks from save states. + * Kernel: Check sceKernelAllocPartitionMemory args. + * Kernel: Fix save states for volatile memory alloc. + * GE Debugger: Include rendered CLUTs in frame dumps. + * Address feedback + * Vulkan: Remove the useThread bool entirely. + * Start moving stuff out to VulkanFrameData.cpp + * softgpu: Skip flushing on no change. + * softgpu: Skip more triangles with equal verts. + * Windows: Normalize paths for OpenDirectory(). + * GE Debugger: Save current clut in frame dumps. + * HLE: sceKernelAllocPartitionMemory volatile memory support (partition 5) + * Gamedata: Cleanup install request args/validate. + * Fix the UI setting "Buffer graphics commands". Was off by 1. + * GPU: Clip depth properly when also clamping. + * GLES: Support more clip distances. + * softgpu: Correct line early z checks. + * GE Debugger: Allow search by state name. + * GE Debugger: Show time spent stepping. + * softgpu: Fix s8 primitives in throughmode. + * headless: Add some new and passing tests. + * GPU: Fix crash on imm vert triangles. + * softgpu: Correct secondary color on imm verts. + * softgpu: Fix state/continuation for imm prims. + * softgpu: Support fog and color1 on imm verts. + * softgpu: Refactor imm draws to bypass vert read. + * softgpu: Simplify vertex reading. + * softgpu: Simplify vertex range culling. + * softgpu: Explicitly flush on sync and output. + * GPU: Minor cleanup of unnecessary virtual. + * Quick GE debugger stepping fix, after the recent vulkan swapchain acquire change + * Update pspautotest submodule + * Update ffmpeg submodule (no big news, just readme updates and a gcc buildfix) + * Consider the Adreno and Mali stencil-discard bugs the same. Use the better check from the Mali bug. + * Show GPU device ID in system information if available. + * OpenXR - Move skyplane compat flag into compatvr.ini + * OpenXR - Refactor variable name + * OpenXR - Compile fix + * OpenXR - Do not mashup Common/GPU + * OpenXR - Refactor + * OpenXR - Use fog color to replace the sky + * OpenXR - Refactor VR compat integration + * Force clearing framebuffer for GTA games + +------------------------------------------------------------------- +Sun Sep 18 00:23:17 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.13.2~git20220917: + * headless: Benchmark a test or frame dump. + * headless: Add option for MSE pass threshold. + * headless: Switch screenshot error to MSE. + * Headless: Small cleanup of options handling. + * softgpu: Run early Z tests in fast rect path. + * UI: Fix some logic dependent on screen tags. + * Remove assert (this happens when playing framedumps) + * Finish the switch to late acquire, moving the swapchain responsibility to the queue runner. + * RunSteps cleanup + * Move FrameData out of VulkanRenderManager + * Move back buffer rendering to a separate "present" command buffer + * Remove splitSubmit setting. Now we submit init commands before the acquire. + * Flag shader blending as broken on Adreno for now. + * Need to have input attachments in the descriptor pool. + * Remove logspam + * Show render pass types in GPU profiler + * Use subpass dependencies to implement shader framebuffer read in Vulkan. + * Add two more extensions to device extension lookup + * Careful with min filtering of framebuffer textures only if auto max quality is on. + * Increase log level of ppsspp version + * Ah, need to enable the Mali workaround for Outrun's water, too. + * Logic op fix, min filter fix. Also remove some unnecessary indentation. Fixes the water on Adreno (no logic) + * Log the screen tag when logging scissor errors. + * Add tags to all our UI screens. Useful to debug UI problems. + * Fix const pointer conversion error for RISC-V + * D3D9 crashfix (flare doesn't render correctly though) + * Copy the texels directly in LoadCLUT + * Small simplification + * Address feedback, change roles of the two dynamic clut fbos to eliminate the need for state (see next commit) + * Seems Need for Speed: Shift has exactly the same lens flare, enable the same compat settings. + * OpenXR - Better variable/methods naming + * OpenGL and D3D11 fixes + * OpenXR - Add code comments + * Log improvements + * Ridge Racer fix + * Get depal-from-dynamic-CLUT working + * Set texcache entry status flag if CLUT needs to be dynamic + * Add way to bind cached textures to a DrawContext + * Allow binding "native" texture objects to thin3d + * Start implementing ApplyTextureDepal + * Cleanup in ge_constants.h, add virtual CLUT8 buffer format + * OpenGL fix + * Comment + * Force low resolution for effects in Burnout Dominator. Fixes lens flare at high res + * Enable depth uploads on render-to-clut-buffer. Esoteric but needed for #11100. Compat flag for now. + * Enable DownloadCLUT and ShaderBitmask for Burnout Dominator + * Detect the CLUT8/8888 situation + * Buffer copy matching workaround + * OpenXR - Ensure the head pose is locked during frame rendering + * Switch multiple bools as parameters to DecodeTextureLevel to a flags enum. + * Fix regression in Tantalus games. See #15914 + * GE Debugger: Allow fb copies while stepping. + * GPU: Fix simulating logicop with blend and shader. + * Unify CLUT texture size at 512 to reduce complexity. Worth the extra bytes. + * Extract function GetReinterpretPipeline + * Savedata: Return errors more correctly on load. + * softgpu: Force correct alignment on 32-bit. + * Common: Fix some type compare warnings on 32-bit. + * softgpu: Remove mostly unused screen offset state. + * softgpu: Cache texture bufws at 16 bit. + * softgpu: Reduce size of VertexData texture coords. + * Non-buffered rendering crashfix + * OpenXR - Crash when using 3D stereo fixed + * OpenXR - Requested header change in compatvr.ini + * OpenXR - Use per game stereo separation file + * Address feedback + * Better shader compile logging + * Don't start blending in the shader just because there's bitmasks, if blend is disabled. + * Debug-name shader modules with their shader desc strings + * softgpu: Ignore stencil test mask in ALWAYS mode. + * softgpu: Include early Z in func description. + * softgpu: Detect longer strips of rectangles. + * softgpu: Detect full triangles as rectangles. + * softgpu: Refactor triangle cull processing. + * softgpu: Store vertex colors as packed RGBA8. + * softgpu: Enable early Z tests a bit more often. + * softgpu: Reduce some minor bin item field sizes. + * README: Move some older releases to history. + * GE Debugger: Add some constants to expressions. + * GE Debugger: Add fields to register expressions. + * Add Rewrite to DrawSyncEatCycles hack, softlocks without it + * GE Debugger: Eval breakpoints as if register set. + * GE Debugger: Provide addrs in breakpoint conds. + * Remove CLUT matching case that shouldn't be there yet. + * softgpu: Switch vert continue buffer to members. + * softjit: Implement accurate fog color blending. + * softgpu: Correct accuracy of fog calculation. + * softgpu: Restrict sprite fast path to throughmode. + * Shadergen issues found by the test + * Switch more texture matching calculations to use bytes for matching. + * Texture/framebuffer matching: Do margin check in bytes. + * Add ShaderDepalMode enum, use in shader ID. Replaces smoothed bit. + * Implement shader depal for D3D11. + * DownloadFramebufferForCLUT: Apply the fix mentioned in #8590 + * CLUT download: Replace the broken block transfer check with a new compat flag. + * vertexjit: Use consistent skinInDecode. + * softgpu: Check depth test early on simple stencil. + * softgpu: Clamp/wrap textures at 512 pixels. + * GE Debugger: Correct confusing block transfer size. + * D3D11: Prevent crash on larger textures. + * TexCache: Allow valid bufw sizes. + * Vulkan: Avoid crash if large texture attempted. + * GE Debugger: Fix crash on bp before list running. + * GE Debugger: Fix crash stepping with large tex. + * softgpu: Avoid calling unordered coords tl/br. + * softgpu: Avoid over-aggressive rect conversion. + * GE Debugger: Fix logging of texture size. + * softgpu: Allow almost flat rectangles to go fast. + * softgpu: Refactor duplicate rectangle compat check. + +------------------------------------------------------------------- +Sun Sep 11 00:15:34 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.13.2~git20220910: + * Add logging for LoadCLUT from framebuffer. See #15923 . Logs nicely in Burnout Dominator. + * GE Debugger: Correct cond break removal warnings. + * Vulkan: Wake compile thread when enqueued. + * Update version to 1.13.2 + * Update README.md for v1.13.2 + * CI: make indentation in tarball.yml consistent + * CI: also upload release source tarball on tag push + * GPU: Dirty params when converting viewport state. + * softgpu: Correct morph weight updates. + * Add Emulator to Categories in ppsspp.desktop.in + * Vulkan: Declare gl_Position as invariant. See #11583 + * readme link + * D3D9: Correct viewport offset sign. + * D3D11: Correct viewport offset sign handling. + * Remove a per-audio-frame allocation in sceAudio + * Initialize a bunch of Dialog structs + * Font: Correct uninitialized values. + * Atrac: Avoid accessing uninitialized stream status. + * Atrac: Avoid an uninitialized value. + * softgpu: Correct bad munmap() size. + * atrac3: Apply the overallocation workaround in a couple more places. Fixes #15788 . + * Atrac3+: Allocate some extra bytes for the data buffer to workaround decoder bug. + * atrac3: Apply the overallocation workaround in a couple more places. Fixes #15788 . + * Warning fix + * atrac3: Allocate some extra bytes for the data buffer to workaround decoder bug. + * Match FragmentShaderFlags definition with the other PR. + * Some more debug names to vulkan objects + * UI: Resize after visiting developer settings. + * GPU: Make NotifyConfigChanged() calls consistent. + * Build: Another attempt to fix libretro. + * Fix compiler warnings + * Assorted cleanup + * Fix for desc struct + * Vulkan: Improve tagging of pipelines for debugging purposes + * Extend lifetime of ShaderModule objects since we might need to recreate pipelines for different render passes + * Fix pipeline lifetime issue, misc. + * Correct merging of render passes. However, we have lifetime issues. + * Lifetime fixes, cleanups + * Fix shader caching. + * Also create renderpass variants on demand + * Create framebuffer variants on demand + * Split each renderpass/framebuffer into multiple "compatibility classes" (RenderPassType). + * Another attempt at fixing libretro, this time the win build failed + * GE Debugger: Show imm prim flag detail in disasm. + * GPU: Handle more flags on imm prim command. + * softgpu: Use vertType prim override for flags. + * GPU: Allow usage of texturing in immediate verts. + * GPU: Handle immediate prims more accurately. + * GE Debugger: Count imm prims as prims. + * softgpu: Fix immediate rendering crash. + * GPU: Fix crash on overlap copy w/o cur target. + * GPU: Fix CanDepalettize() missing false paths. + * Build: Add additional zstd file for libretro. + * d3d9: Correct decode of 16-bit textures. + * Add Lubos to credits list + * GE Debugger: Add more missing GE disasm. + * softgpu: Implement matrix data wrap-around. + * Ge: Read matrix num/data back like hardware. + * Build: Include all zstd files in libretro. + * GE Debugger: Add conditions to cmd breakpoints. + * GE Debugger: Add address breakpoint conditions. + * GE Debugger: Allow expressions for goto address. + * Remove impossible check + * Buildfix, warn fixes while at it + * NFS speed effect: Improve framebuffer choice logging, tighten up stride checks. + * Correct libretro buildfix (hopefully) + * OpenXR - Config for stereo separation added + * GE Debugger: Add a table of register metadata. + * OpenXR - Stereo matrices fixed + * libretro build fix, see #15968 + * Remove reference to non-existent file from Common.vcxproj + * The Darkstalkers subpixel check should be == 0. Fixes #14953 again + * OpenXR - Fixing FOV better + * Fix for ARM 32-bit. + * Fix case problem in libretro make file, see #15964 + * Remove the assembly implementation of 4x4 matrix mul, in favor of intrinsics. + * Disable BlueToAlpha mode when logic op is present. Re-enable for Outrun and DiRT. + * Vulkan: Update to latest VMA. + * Build: Update xxhash to v0.8.1. + * Build: Update to zstd v1.5.3. + * Build: Split ppsspp_common out for Android bins. + * Water effect fix #2: If ForceLowerResolutionForEffects, don't upscale textures participating in small-resolution passes. + * Water effect fix #1: Force "Lower resolution for effects" on in Outrun and DiRT 2 + * Fix "Lower resolution for effects" in Ratchet & Clank. Bad option for this game though. + * Fix the new logic-op-in-shader on OpenGL and D3D11 + * OpenXR - Mismatch in stereo and normal uniform fixed + * OpenXR - Do not include lens projection in nonVR builds + * OpenXR - Split projection matrix for color and depth buffer + * Shader unit test fix + * Preserve alpha in shader logic op emulation. Include logic op in check for packUnorm4x8. + * OpenXR - Mismatched call order fixed + * OpenXR - Multipass rendering fixed + * OpenXR - Stereo without multiview added + * Turn on logic ops in shader when shader bitmasking is used. + * Shader bit setup and code generation for logic-ops-in-shader + * headless: Fix startup on latest Android versions. + * arm64jit: Handle branch/jump in branch delay slots. + * x86jit: Refactor and fix jump in branch delay slot. + * irjit: Handle branch/jump in branch delay slots. + * OpenXR - FBO intialization better + * compile fix + * OpenXR - Viewport init better + * OpenXR - Emulate mouse cursor + * OpenXR - Multiview refactor to make it optional + * OpenXR - Compile fix + * OpenXR - Swapchain acquiring fixed + * OpenXR - Flag to easily disable multiview shaders added + * OpenXR - Keep using GLES2 headers for now + * OpenXR - Dummy integration compile fix + * OpenXR - OPENXR ifdef removal + * OpenXR - Enable performance mode + * Assert removed + * OpenXR - Step closer to identify multiview FBO issue + * OpenXR - Use OpenGL ES 3 and ensure FBO pair calls + * OpenXR - Initialization errors fixed + * OpenXR - Multiview in progress + * OpenXR - Multiview uniforms working + * OpenXR - Disable multiview for now + * OpenXR - GL and OXR logging added + * OpenXR - Integrate multiview into system + * OpenXR - Config for stereoscopic vision added + * OpenXR - Unused parameter removed + * OpenXR - VR renderer adjustments + * OpenXR - Framebuffer multiview used + +------------------------------------------------------------------- +Sun Sep 04 11:50:45 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.13.1~git20220904: + * UI: Only force ASCII messages if they're ASCII. + * Fix dialog rendering (PPGe). Fixes #15954 + * spaces->tabs + * Fix GLES check + * Additional cleanup + * Add log functions to state structs + * Move the computation of simulate-logic shader type earlier + * Some more renaming and refactoring, extracted from a future PR + * Channel mask refactor + * Correct LogicOp API support detection + * Minor refactor with pipelines in QueueRunner (makes more information available for easier debugging) + * Don't parallelize shader compiles in debug mode, sync in debug allocator kills performance. + * interp: Handle jumps in branch delay slots better. + * softgpu: Correct texture check in fastpath. + * softgpu: Validate alpha test properly for fastpath. + * softgpu: Avoid rect fast path with subpixel offset. + * Kernel: Use PSPPointer for status objects. + * Io: Use PSPPointer instead of struct temps. + * Font: Remove needless ReadStruct(). + * Ge: Switch from ReadStruct to PSPPointer. + * Psmf: Use PSPPointer instead of WriteStruct. + * Net: Remove usage of Memory::WriteStruct(). + * HLE: Remove misc usage of WriteStruct(). + * PPGe: Cleanup memory writes for clarity. + * Fix lying comment, log level reduction + * Promise: Add SpawnEmpty (couldn't come up with a better name), fix bug with multiple waiters + * Switch Vulkan pipelines to use promises for synchronization + * VS 2022 warning fix + * More buildfixing + * Buildfix + * Buildfix + * Fix silly logic, fixes the shader leak. + * Work towards finding the leak + * Don't block on shader creation until it's time to create the pipeline. + * Use shader module promises as keys in pipeline cache instead of the actual shader modules. + * Create shader modules (and compile GLSL) on worker threads + * minor sv translation fix + * Use promises to hold shader modules, for later asyncification + * Remove confusing resetFramebufferRead flag from secondary framebuffer binding + * Remove setting "Disable slower framebuffer effects" + * Remove unused function + * langtool: Run fmt, remove wrong comments + * langtool: Add easy command to remove a key. + * unitest: Correct wrong free, initialize data. + * vertexjit: Update unit test, oops. + * Avoid recomputing the shaderblend setup in ComputeFragmentShaderID. + * Small steps + * Some reordering + * More renaming + * Plumb the computed pipeline state into ComputeFragmentShaderID. + * Some renaming. Store the BlendState/MaskState. + * vertexjit: Remove unused ReadUV() cases. + * vertexjit: Remove unused ReadPos() cases. + * vertexjit: Clamp through float pos during decode. + * Replacement: Fix reload of replaced textures. + * Replacement: Detect OOM even with replacements. + * Vulkan: Prevent saving tex during replace load. + * Revert "Merge pull request #15930 from sum2012/kernel-minor" + * Fix image format when dumping textures. + * Fix texture saving for Vulkan + * Display paths with backslashes on Windows. Show texture path when saving. + * Shrink VkRenderData from 88 to 64 bytes. + * Another buildfix attempt + * libretro buildfix + * Remove allocation from the heaviest MemBlockInfo path (BlockTransfer). Required some refactoring. + * Remove a per-audio-frame allocation in sceAudio + * Shrink the DeferredDrawcall struct, because why not. Assorted cleanup + * Shrink the TransitionRequest struct, use TinySet + * TinySet bugfix, add "append" + * fbtex matching: Shrink the matching struct, use TinySet instead of vector to avoid an allocation. + * Stricter checks for negative framebuffer offsets. Fixes #15937 + * Merge GetFramebufferCandidates and GetBestCandidateIndex into one function. + * Much faster framebuffer matching (no more ToString every time there's more than 1 candidate, etc). + * Rename Slow Framebuf effects to Shader Blending (might remove). Reassign readbacks to BlockTransferGPU. + * KillZone: Fix margin detection on the title screen, fixing some glitches. Need more fixes though to fix the shimmer effect. + * Extract the split code to a function + * Quicker way to update the through projection matrix with the split framebuffer offset + * Cleanup (more to do, though) + * Allow signed texture offsets on framebuffers + * Offset the draws properly. Now just some texel clamping left to fix + * Check X coordinate of through-mode drawcalls to figure out which of the splits to render to + * Tweak ForceMax60Fps to work for Killzone + * Split the framebuffer in Killzone, with a compatibility flag. + * Small tweak in texture matching. Fixes #15927 + * Fix sceKernelUnlockMutex timing + * Remove "DisableReadbacks" compat flag. MotoGP no longer needs it after BlockTransferAllowCreateFB was added. + * Fix race condition in replacement texture loading + * Implement shader blending for D3D9 + * Clean up mip size calculations + * GE Debugger: Improve display list disasm. + * Address feedback + * Add support for reading depth buffers to the PackFramebufferSync function. + * Add another stat, for self-tex + * Remove the reinterpret compat flag, now always on + * Fix OpenGL bug when framebuffer fetch is available. Add two new stats. + * Better framebuffer checks, remove all ways that framebuffer formats can change. + * Use better pattern for enum switches + * Some Vulkan renderpass load/store optimizations. + * Depal: Set scissor tightly around processed area. + * UI: Validate UI scissor inside bounds. + * GPU: Make Spongebob hack checks more readable. + * GPU: Recheck depth/stencil after Spongebob hack. + * GPU: Add a way to defer dirty rechecks. + * GPU: Fix possible shift truncation. + * Update the Spongebob workaround check to work with MX vs ATV: Reflex. See #15907 + * riscv: Add a simple unit test. + * More spongebob check tweaks, assert fix + * Remove irrelevant stencil state checks + * Address additional feedback, thanks! + * Safer spongebob depth inverse check, includes color check and depth func + * Unify the spongebob depth inverse check + * riscv: Emit compressed instructions. + * riscv: Improve sign reduce/immediate readability. + * riscv: Emit ADD/SUB/etc. for ADDW/SUBW/etc. on R32. + * riscv: Include an LI helper in the emitter. + * Comments and fixes + * Block transfer bug fix + * OpenGL graphics fixes + * Comment fixes + * Reinterpret is the proper fix for God of War's shadows. + * Fix UV offsets in 16->32 reinterpret. Better, but something is still not quite right. + * GL bugfix, log cleanup + * Enable reinterprets and create-buffer-on-copy for the Tantalus games + * Fix problem matching framebuffers for reinterpret + * Logging improvements + * Remove some old logging that's not very interesting anymore + * Log improvement + * When cloning framebuffers for reinterpret, adjust width according to bpp + * Framebuffer texture matching: Remove heuristics other than seqCount + * Accept color textures as reinterpretable between 32 and 16-bit formats + * Implement the stencil/alpha reverse trick for all backends + * Fix bug in reinterpret float impl + * Comment fix, cleanup + * Minor D3D11 refactoring + * Fix Katamari + * Bugfix, better logging + * Quick hack to widen the depal texture if 0x200000 swizzle is needed + * More successful swizzle impl. Not perfect + * Unsuccessful swizzle implementation + * wip + * Initial implementation of 32/16-bit color reinterpret blits. + * Refactor reinterpret to prepare for 16/32-bit + * Prepare for 16/32 reinterpret + * Detect need for 16<->32-bit reinterprets + * Work in bytes in the framebuffer matching algorithm for block copies, avoid margins + +------------------------------------------------------------------- +Sun Aug 28 00:27:22 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.13.1~git20220826: + * Finish the depal refactoring. + * Fix shader compilation error in depth->565 + * riscv: Emit CSR manipulation instructions. + * riscv: Validate FixupBranch usage better. + * riscv: Emit float instructions. + * riscv: Emit atomic operations. + * riscv: Emit mul/div instructions. + * riscv: Emit fence and 64-bit instructions. + * riscv: Emit 32-bit 3-op instructions, validate GPR. + * riscv: Emit 32-bit load/store and imm ops. + * riscv: Emit standard B/U/J type ops. + * riscv: Add initial emitter shell. + * Remove support for framebuffers changing stride (already unreachable). + * Revert minimum stride + * Add missing terminator to VR settings array. Better logging. + * More comment fixes, cleanups + * Comments + * Comment updates + * Merge the two find FindTransferFramebuffer* methods + * Additional renaming + * Rename variables to make the two functions match + * Split up FindTransferFramebuffers + * Tighten the check a bit more + * Reorder some parameters. Detect buffers in stride gaps. + * Remove unused function decl + * Fix smoothed CLUT lookups. Shouldn't go beyond the mask's reach. + * Add floating point version of reinterpret shaders + * Fix BlueToAlpha together with reinterpret. Add back fake reinterpret for now. + * Reimplement texture format reinterpretation + * Fix smoothed depal on GLES. Don't enable filtering if not using. + * Fix checks for depth texture support, GLES syntax errors + * OpenGL: Fix regular depal (shader depal still worked) that broke recently + * Remove duplicate uniform queries + * Fix wrong width/height values being stored in thin3d textures + * Drive-by texture slot management cleanup + * OSD: Scale large multiline messages to fit better on screen (like shader logging) + * GPU: More consistently clear ptrs on DeviceLost(). + * GPU: Correct Draw2D::DeviceRestore(). + * GE Debugger: Fix bad read on step at start of VRAM. + * GE Debugger: Add separate step based on vsync. + * GPU: Centralize SetDisplayFramebuffer(). + * GE Debugger: Record 1 flip if no display calls. + * Vulkan: Very minor chnages after checking Best Practices with new validation layer + * Use Draw2D for depal shaders (except the actual blit, for now) + * Break out Draw2D in a class + * Switch reinterpret shaders over to the Draw2D framework. + * More Draw2D refactoring + * Small Draw2D refactor + * Rename VirtualFramebuffer->format to fb_format to line up with fb_address and fb_stride (fundamental parameters) + * Rename fmt->fb_format + * D3D9: Correct viewport offset sign. + * D3D11: Correct viewport offset sign handling. + * Copy color from overlapping framebuffers on bind, under certain conditions. + * Implement smoothed depal for the "old" depal path as well. + * Better checks for smoothed depal + * Remove SmoothedDepal compat setting, instead detect the ramp directly. + * Special case depal lookups for Test Drive's strange usage. + * More renaming. Fix shader view for Vulkan + * DepalettizeCommon -> TextureShaderCommon. Simplifications. + * Convert TextureShaderApplier to a member function in DepalCache. + * Simplify depal shader apply code a bit. + * Replacement: Read files only within time budget. + * Replacement: Be more consistent about base level. + * Vulkan: Cleanup minor logging. + * Config: Default PSP model to SLIM on all platforms. + * Linux: Consider Vulkan if OpenGL init fails. + * GLES: Correct preview of mip levels > 0. + * GPU: Share CLUTs if no extended CLUT. + * Headless: Set softgpu config flag. + * Headless: Fix comparison for oversized buffers. + * UI: Show return address for exec crashes. + * Core: Show exception on misaligned jump. + * irjit: Validate alignment in slow memory mode. + * irjit: Add constructs for validing mem access. + * More D3D9 work on depth textures. Something still missing. + * Address more feedback, cleanup scale parameter + * Remove leftover resolutionScale depal config parameter + * Compatibility check + * Use a DX9-compatible method to pass texwidth and height. the effect doesn't work on DX9 yet though. + * Fixes for D3D11 + * Remove debug code, cleanup + * Fix glitch when changing render resolution + * Really not sure how we can detect whether a depth deswizzle is needed :( Hopefully temporary flag... + * Texture bind channel fix. Still need to figure out a way to decide to swizzle or not. + * Fix graphics in Ratchet & Clank. WIP + * Fix naming of temp fbos to be in PSP pixels + * Allow binding depth as 565 by going through depal. + * softgpu: Correct morph weight updates. + * softgpu: Flush on transfer to pending tex read. + * softgpu: Restrict CLUT to proper size. + * GE Debugger: Include extended CLUT in frame dumps. + * GPU: Restrict mip CLUT enhancement a bit. + * GE Debugger: Correct 16-bit CLUT display. + * GPU: Correct depal CLUT texture for 5551/565. + * GPU: Purify ConvertViewportAndScissor(). + * GPU: Dirty params when converting viewport state. + * OpenXR - Refactor display framerate + * OpenXR - Resuming renderer fixed, haptics support added + * OpenXR - Recenter changes pitch of the 2D screens + * OpenXR - Ensure PSP aspect ratio for 2D content + * OpenXR - Option to set canvas distance added + * OpenXR - 2D mode hack mappable (using screen keycode) + * OpenXR - Default config values fixed + * OpenXR - Support for multiple projection matrices in the scene + * OpenXR - Rotation mirroring fixed in a few games + * OpenXR - Remove OPENXR ifdef from UI + * OpenXR - Better detecting of 2D menus + * OpenXR - Detect 3D content better + * OpenXR - Detecting UI mode improved + * OpenXR - 72Hz support added + * OpenXR - Unused variable removed + * OpenXR - HUD is enabled based on gstate + * OpenXR - Forgotten file added + * OpenXR - Introduce VR tweaks + * OpenXR - HUD support added + * OpenXR - Estimate world unit scale + * OpenXR - Head orientation in some games fixed + * OpenXR - Refactor the common math methods + * OpenXR - 6DoF head movement support added + * OpenXR - Renderer support for axis mirror added + * OpenXR - Switching between 2D/VR fixed + * OpenXR - VR settings added, better UI detection + * OpenXR - Automatically switch between fullscreen/VR + * OpenXR - FoV fix for Quest 1 + * OpenXR - Basic look around implementation + * OpenXR - Enable 6DoF mode when in game + +------------------------------------------------------------------- +Sun Aug 21 00:28:47 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.13.1~git20220820: + * Address feedback to PR #15858 + * Restore the clearing optimization, avoiding unnecessary depth copies + * Fix deferred-depth for bezier/spline. Move updating of last_frame_depth_render to GPUCommon. + * Minor fixes. Everything seems fine now. + * Implement copying color to depth on load. However, weird stuff is happening.. + * Extract depth copies to CopyToDepthFromOverlappingFramebuffers + * Remove the color-to-depth mode + * Defer depth copies until depth buffer is actually used. + * Texture bind channel fix + * More consistent handling of lack of depth texture support. Small log improvement. + * Fix bug in framebuffer matching where the wrong format was checked for depth matching. Fixes #15867 + * Update ko_KR.ini + * Update ko_KR.ini + * Update ko_KR.ini + * Update ko_KR.ini + * Initialize a bunch of Dialog structs + * Fix dependency of RISC-V targets on common JIT code + * Update ko_KR.ini + * Bias towards the most recently written framebuffer + * Remove restrictions on VRAM addresses for depth and color + * Always attempt to match both depth and color + * Pre-reject DXT formats, Misc + * Show seqCount in candidate list + * Stride fix in matching + * Framebuffer matching simplification + * Fix the raster depth blit again, which I recently broke. + * Typo fix + * Silverfall: Add the US game ID too + * Add Silverfall to [AtracLoopHack]. At least we're gathering a good list of test cases once we have an idea for a real fix... + * Use sequence numbers instead of a tracking array for depth buffers + * Revise comments according to feedback + * Oops, remove wrong comments + * Comment updates + * Clean up and comment framebuffer struct better, add bind sequence numbers + * After recording a GE dump, open an explorer window pointing at the file + * Remove the DX9 namespace, for consistency with the other backends + * Vulkan: Restore a minor old forgotten optimization + * More removing redundant stuff + * Add a newline to "Leaving main" message. + * Refactor 2D pipeline creation + * Minor cleanup in reinterpret + * Rely on refcounting for stencil upload shader modules + * thin3d: Properly AddRef and Release shader modules from the pipeline in backends that need it + * add bin and elf to picker as they now work + * Remove some unused code in FramebufferManagerD3D11.cpp/h + * remove storage fileloader and storage folderbrowser + * Remove deprecated function (mostly) + * Remove some unnecessary bool returns. + * Move out the reinterpret stuff to ReinterpretFramebuffer.cpp. Draw2D refactor. + * Assorted minor cleanup + * Revert "Pause the debugger on TRANSFERSTART (treat that as a prim)" + * remove unneeded switch case as un broken cases fall through + * GPU: Avoid using depth buffers for blending. + * GE Debugger: Fix crash on 0 count prim. + * GE Debugger: Correct base accounting in playback. + * GPU: Skip fb create upload when clearing. + * GE Debugger: Save tab configuration. + * GE Debugger: Cleanup duplicate dlist redraws. + * GE Debugger: Correct tab updating for right panel. + * GE Debugger: Actually allow swapping tabs. + * GE Debugger: Add context menu to control tabs. + * GE Debugger: Show/hide tabs if empty. + * GE Debugger: Track tabs per panel. + * Move the display list view to its own column on the right + * Add plain Create method to CtrlDisplayListView + * Pause the debugger on TRANSFERSTART (treat that as a prim) + * Add a second set of tabs to the Windows GE debugger. Also reorder stuff. + * oop + * oop + * oop + * Use compat.ini instead + * Add atrac hack setting + * Net: Cleanup initialization syntax. + * GLES: Correct some variable shadowing. + * libpng: Correct func pointer type warning. + * Global: Cleanup some unreferenced warnings. + * Font: Fix bad memory read on invalid font data. + * Assign value to version where appropriate + * Update + * A better way to get property + * Merge update into master + +------------------------------------------------------------------- +Sun Aug 14 00:30:25 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.13.1~git20220813: + * Font: Correct uninitialized values. + * Atrac: Avoid accessing uninitialized stream status. + * Build: Correct ccache keys. + * Atrac: Avoid an uninitialized value. + * softgpu: Ensure TCs are initialized. + * softgpu: Correct bad munmap() size. + * x86jit: Avoid an instruction form Valgrind hates. + * Headless: Correct bad pointer without gfx context. + * Core: Correct shutdown on late init failure. + * Try to fix shader compilation error on Adreno reported by Lubos + * Debugger: Bump GE dump version to ignore bad flag. + * Debugger: Correct GE frame dump VRAM dirty flag. + * Debugger: Fix crash on bad request. + * GPU: Avoid stencil force pass when writing depth. + * GPU: Write stencil fail to alpha is RGB masked. + * Fix feature checks + * Enable framebuffer depth blits through raster. Fixes games like Jeanne D'Arc on D3D9. + * Some enum renaming, move RasterChannel to GPU.h. + * Careful with mipmaps on old OpenGL ES. Lost this check in the refactor, made it clearer than before at least. + * Texcoord attrs must be called a_texcoord0, this should be fixed. Oh well. + * Convert depal shaders to use the ShaderWriter. Add tests for depal shaders. + * Cleanup + * Warning fix + * Remove BGRA color swap in depal for D3D, no longer needed (and breaks Simpsons) + * Update zh_CN.ini + * Headless: Cleanup unused/leak warnings. + * Change indentation to tabs. + * UI: Show better warning on MAC reset. + * Lang: Fix some mismatched lang keys. + * Build: Remove no-format. + * Common: Avoid format warning with log timestamp. + * GLES: Correct alpha check for depal framebuffers. + * Add -Wformat (see issue #15792), no idea if ok. + * Oops, this iOS check was still required. + * More framebuffer blit fixes. + * Improve the NPOT check + * Forgot these... + * Make GLQueueRunner/RenderManager aware of device caps, use it to check for NPOT and some other stuff + * Remove optimization in BlitUsingRaster that shouldn't have been there (redundant, lacked feature check) + * Reimplement bicubic upscaling. + * GLES: Correct semantic name for Draw2D. + * GPU: Clarify new framebuf stencil upload. + * Dialog: Add missing header for MSVC 2017. + * GPU: Skip alpha stencil upload when safe. + * GPU: Cleanup remnants of GLES-only stencil upload. + * Fixed infinite loop and crash which could occur when using "Import from cheat.db". + +------------------------------------------------------------------- +Sun Aug 07 00:32:10 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.13.1~git20220807: + * Fix feature check for wrapping of non-pow-2 textures + * Feature detection fix on old GL, more compiler warning fixing + * D3D9: Fix a lot of stuff + * More cleanup, fixes. + * More cleanup + * Buildfixes + * Lock out shaderDepal from D3D for now, will implement later. + * Feature detection fixes + * Delete all the duplicate implementations of ApplyTextureFramebuffer + * Remove Vulkan2D + * Move the depal cache instance ownership to TextureCacheCommon + * Compiles on Vulkan, but doesn't yet work. + * Switch over D3D9 as well. Buggy, will debug after unification + * Works with D3D11, except the depth texturing case + * Simplify + * Fixy fix + * D3D11 prep + * Move the TextureShaderApplier out to DepalettizeCommon + * Move more stuff to Common + * OpenGL: Fix shader depal too. + * Move to Common + * Mostly works for OpenGL. + * Port the OpenGL depal path to thin3d + * Fix some compiler warnings + * ShaderWriter: Fix GLSL 1.0 output + * Fix corrupted colors when using software upscaling with OpenGL. + * Use safe_string when printing %s with Memory::GetCharPointer. + * Prefer using "%04x" over "%04hx". + * State handling reordering in D3D11 and D3D9 backends. + * Fix logging warning issue. + * Update /lang/ru_RU.ini + * Fix include issue. + * Updated copyright's year. + * Replace the dummy PSN name with Nickname on sceNpInit. + * Moving sceNpMatching2-stuff to sceNp2.cpp/h. PS: Register_sceNpMatching2 not being added to HLETables because currently the callback didn't work properly yet and could crash the game. + * Implement NP Signin Dialog (faked signin) + * Updated NP Library PS: Some of registered modules should probably moved to a separate files (ie. sceNp2.cpp). + * Added a FunctionWrapper used by sceNpService + * UWP: Fix missing font file causing missing text fields. Fixes #15771 + * Delete more unused code + * Fixes, state invalidation + * Delete unused stuff in all backends + * Remove the D3D9 and Vulkan BlitFramebuffer functions + * Replace D3D11's BlitFramebuffer with the common one + * Convert FramebufferBlit from the GL backend to use thin3d + * D3D11: Remove more custom drawing + * Fix upscaling in Vulkan. Fixes #15779 + * UWP buildfix + * Reimplement filtering in DrawActiveTexture + * Delete the OpenGL implementation of DrawActiveTexture. + * Delete the Vulkan implementation of DrawActiveTexture, fixups to shader gen + * Delete the D3D9 impl of DrawActiveTexture + * Move to common (todo: merge with above) + * Convert the D3D11 DrawActiveTexture to thin3d + * Draw2D + * Need a separate cap bit for stencil blits to prevent misuse on D3D9 + * Cleanup and testfixes + * Unit test, comment + * Rebind the framebuffer only if we're in the blit path. + * Bufferless path for D3D11/Vulkan + * Get D3D shaders to build (though not working correctly) + * Switch to ShaderWriter. Works in OpenGL + * Unify stencil buffer upload using Draw. Only OpenGL tested yet (shaders need adaptation). + * Implement DrawUP for D3D11 + * GLES: FramebufferBlit: Don't blit if the target is the current framebuffer. + * Clean up blit/copy feature detection. Don't need fast GPU flags for these. + * Fix crash in shader upscaling. This variable shadowed the one that now lives in Common. + * Use BlueToAlpha in more games instead of ReinterpretFramebuffers/ShaderColorBitmask + * Fix Vulkan as well + * Fix for D3D11 and GL, but not Vulkan for some weird reason + * Avoid upscaling textures used for color-to-depth. + * Disable texture filtering when drawing color into depth + * D3D9 fix + * OpenGL fix + * D3D11 fix + * Implement color-to-depth for Vulkan, start implementing for D3D11 + * Add logging for some non-traditional uses of framebuffers. + * Fix small breakage in Vulkan texture decoder + * Fix Macross glitches. + * Vulkan: Don't try to auto-mipmap 3D textures in max quality filter mode + * Fixes UWP building issue after Retargeting Solution to a different SDK version. PS: This won't work when selecting the "latest installed version" tho, so a specific version need to be selected when retargeting. + * OpenXR - Use a build variant instead of a module + * CI: fix tag regex to match also shorter tags like v1.13 + * Address feedback + * Check for color/Z clashes + * Track depth buffers separately from framebuffers to track previous use for copies + * Give the mip bias its own uniform flag. + * Possible flicker fix for Macross + * Revert mistaken change in D3D9 fragment shader gen + * Support 3D textures in OpenGL ES, add feature check + * Testfix, cleanup + * Implement 3D texturing in the OpenGL backend too. Assorted fixes. + * Remove some dead code + * Implement 3D textures for D3D9 as well + * Implement 3D textures for D3D11 + * Some safety checks to avoid bad combinations, probably not really needed in practice. + * Quick attempt at fixing the Macross glitch + * Fix the mip bias to take the texture depth and sampling offset into account. + * Implement new discovery of a larger CLUT than we thought, plus non-shared CLUTs for 8-bit indices. + * Initial implementation of 3D texturing through equal-size mips (see #6357) + * CI: make indentation in tarball.yml consistent + * Reporting: Disable when debugger methods used. + * GPU: Small scaling variable cleanup. + * CI: also upload release source tarball on tag push + * Revert debug change + * Unify the GLES LoadTextureLevel with the D3D ones. + * Prepare to convert the GLES functions too + * Share LoadTextureLevel between the two D3D backends. + * Cleanup + * Part 2 of the above + * D3D: Part 1 of breaking out the API stuff + * libretro buildfix + * UWP buildfix + * Simplify CPU upscaling code + * Vulkan stencil fixes + * Make the D3D9 backend handle stencil state in more similar ways to the others. + * More thin3d d3d11 fixes + * thin3d: Make writeMask and compareMask dynamic in all backends. D3D11 needs emulation. + * thin3d: Remove dual sided stencil, not supported by all backends anyway + * Remove some indentation + * Use PrepareBuildTexture from the Vulkan backend too + * Change from maxLevel to levelsToLoad, for better readability. Cleanup. + * Define texture loading plan semantics better, pre-port some from Vulkan + * Vulkan: One step closer to using PrepareBuildTexture + * Update PrepareBuildTexture + * Use PrepareBuildTexture from DX9 backend + * Use PrepareBuildTexture from D3D11 backend + * Break out the preparation part of BuildTexture from GLES + * Additional cleanup + * Get the GLES BuildTexture slightly closer + * Make the BuildTexture of DX9 and D3D11 backends more similar + * Remove indentation, useless code + * Remove the DX API access from LoadTextureLevel + * Remove fake-mipmap logic from LoadTextureLevel + * Minor code simplification + * Remove unused parameters to some conversion functions + * Remove unused useBGRA parameter + * Remove the ReplacedTextureFormat enum + * Remove support for 16-bit replacement texture formats + * Update ru_RU.ini + * readme link + * Windows: Add a simple window message to get the base pointer. + * Update assets/lang/ru_RU.ini + * Update version for 1.13.1 + * Update README.md for 1.13.1 + * Replacement: Cleanup leak on replacement reload. + * Replacement: Clear cache on disable. + * irjit: Correct another PurgeTemps case. + * irjit: Clarify PurgeTemps, guard a couple ops. + * Add confirmation dialog when generating a new Mac address + * better pt-pt translation + * Fix bug in blue-to-alpha - alpha blending could be on when it shouldn't be. + * Reporting: Avoid crash if socket connect fails. + * Some zero init, add a null check in TextureReplacer. See #15730 + * OpenXR - Move VR folder under Common + * OpenXR - Static added to global variables + * OpenXR - Version info refactor + * OpenXR - Integrate SYSPROP_HAS_FILE_BROWSER + * Update version for 1.13 release + * README.md update for the 1.13 release + * Simple hack to hide the search button in settings on small screens. Needs a better solution. + * Build: Update to NDK r21e. + * Switch to using the D3D9 state cache in thin3d, to avoid confusion + * Clarity improvement in dx state cache stencil func + * Remove unnecessary complexity from the D3D9 state cache + * Remove bool that was always true in state mapping. + * irjit: Correct initialization warning. + * irjit: Add test for double clobber in #15713. + * Headless: Ignore __testcompare.png in tree. + * irjit: Simplify more arithmetic to Movs. + * irjit: Add tests for IR passes. + * Revert the x-offset bpp fix temporarily. + * Add GPU stat for number of depth copies per frame + * OpenXR - Code formatting fixed + * OpenXR - Disable HW scaler on VR headsets + * OpenXR - Default device config added + * OpenXR - Hide unsupported menus + * OpenXR - VR code formatting fixed + * OpenXR - Finalize VR renderer interface + * Libretro fix + * Windows fixes + * Remove unnecessary parameters from MakePixelsTexture + * Delete some unused code. + * Separate logging for GetPointer/GetPointerWrite + * Split GetPointer into two versions, to help with const correctness + * OpenXR - Pragma once instead of ifdefs + * Allows "merging" render targets that overlap on the Y access. Fixes #7295 (Juiced 2) + * OpenXR - Code formatting header removal + * OpenXR - Code formatting global variables + * OpenXR - Code formatting { + * OpenXR - Button repeating implemented + * OpenXR - Controller mapping better + * irjit: Prevent corruption on purge temps pass. + * Add UWP Platform to Issue Templates + * (UWP) Update Dummy Manifest + * (UWP) Remove Reference to Deleted File. + * (UWP) Delete Old Key + * (UWP) Building Improvements + * Remove redundant file from packages. + * Prevent Dummy file from appearing in packages. + * Remove a redundant mutex lock in logging. + * Main screen: Display .ppdmp filenames on buttons in grid mode + * Remove log noise + * Mali driver bug workaround for driver 32+. + * OpenXR - Hacky mapping to make the games playable + * OpenXR - UI controls integrated + * OpenXR - UI scale/resolution fixed + * OpenXR - Input class added + * OpenXR - Rendering UI fixed + * OpenXR - Rendering into VR framebuffer works + * OpenXR - Dummy rendering working + * OpenXR - Initial version of VR renderer added + * OpenXR - VR mode initialization added + * OpenXR - VR framebuffer class added + * OpenXR - Hide options which are not relevant for VR + * OpenXR - SDK added + * OpenXR - Quest target into Android Studio project added + +------------------------------------------------------------------- +Sun Jul 17 11:13:12 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220717: + * Osk: Allow upper/lower for all keyboards. + * Debugger: Update to latest web debugger. + * Debugger: Fix SVG content-type for proper display. + * Use the recent Clone Wars fix for Star Wars: Force Unleashed too + * Update zh_CN.ini + * Add a simple compat flag to workaround the Clone Wars issue, #12949 + * Apply leoxxx's spacing changes in zh_TW.ini + * Update zh_CN.ini + * Update zh_CN.ini + * Add Zettai Zetsumei Toshi 3 bypass for softlock on character select screen + * zh_CN translation: Apply spacing rules according to leoxx. See #15660 + * Io: Correct access field on dir listings. + * Config: Keep recentIsos lock owned by Config. + * langtool: Add command move-key to move a line globally from one section to another. Update deps + * Update zh_TW.ini + * Update zh_CN.ini + * Add files via upload + * compat.ini for Mali stencil bug: Add the Suikoden game too + * Mali workaround - add compat.ini entry for Mahou Shoujo Nanoha A's Portable. + * Config: Cleanup recent on a thread. + * Config: Move recent ISOs access to accessors. + * Io: Show message if files are outside valid ISO. + * irjit: Allow unaligned loads by default. + * UI: Reduce IO for texture ini on dev settings. + * Improve comment, variable name + * Vulkan fragment shaders: Don't opportunistically emit "layout (early_fragment_tests)" + * Slightly improve logging of failed block device reads + * Fixes Kenka Bancho too. + * Tweak the Mali stencil-test workaround to work for Surf's Up. See #15016 + * Dynamic update + +------------------------------------------------------------------- +Mon Jul 11 09:07:34 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220710: + * Remove unmotivated check preventing texture ini creation on mobile. (#15647) + * Common: add explicit for BSDs after 3ed74350129c + * Fix comment + * Double quote + * Be explicit on the possibility of failure + * Use hashtag for RGBA format + * Change theme endianess + * Update zh_CN.ini (#15657) + * Add check that makes PPSSPP work on Xbox through the debugger + * Replace Win32 file IO with UWP safe variants and add support for getting drives to UWP build (#15652) + * compat.ini: Add R-Type games to IntraVRAMBlockTransferAllowCreateFB. + * Compat: Fix backwards check. + * UI: Add small deadzone for analog speed. + * UI: Clamp analog speed to 0 on opposite value. + * Update zh_CN.ini + * Update zh_TW.ini + * UI: Reset views when analog speed mapped. + * UI: Allow cancel from analog speed mapping. + * Change comment based on new findings. + * UWP: Skip pad name lookup. + * UI: Add analog speed limit mapping. + * Core: Track names of connected pad devices. + * interp: Fix constant typo. + * Compat: Add setting to skip compat flags. + * Reporting: Log when vertex count truncated. + * Updated wiki page key in all translation INIs. + * Remove left over + * Change hack names and edit comments based on feedback. + * Improve the vortex hack to make it safe and slow it down + * Test hack that breaks the vortex timer, it's still wrong + * Add 30 fps version as an option, doesn't suffer from softlock + * oops + * Updated text on AdHoc wiki page button. + * Updated "Use original port" tip message. + * Add Fixed60FPShack for GOW, should work smoother than the ForceMax60FPS. Also more stable than it's CWCheat version, unfortunately because of that, still causes softlock in GOW:GOS vortex stage. + * Disable ForceMax60FPS for GOW:GOS to avoid softlock + * Show a tip message when enabling "UPnP use original port" option. + * Update zh_TW.ini + * Update zh_CN.ini + * Update th_TH.ini + * Hide Hostname progressView when not showing any message to get a better view of the list. + * Slightly reduce Hostname progressView's margin as it took too much space. + * Added psp.gameplayer.club public Adhoc Server. + * Additional game IDs for The Simpsons Game + * Apply the mali workaround for The Simpsons Game. + * Merge pull request #15629 from Halo-Michael/lang + * Add ARM64_NEON compile arch flag + * ffmpeg: fix compile bug + * Mask some part of public IP in the logs for security reason. + * Update zh_CN.ini + * More translation key + * Add numpad enter as a confirm + * Update all + * Remove driver version check for Mali stencil/discard bug. See #13833 + * Update eng lang + * Oops, revert debugging commit + * Workaround for strange driver hang on Raspberry Pi 3. + * Keep ye olde else case + * Avoid including signal.h from CommonFuncs.h + * Allow to repeat a "single" button + +------------------------------------------------------------------- +Sun Jun 19 00:37:27 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220618: + * Out of performance paranoia, limit the Mali workaround to known affected games + * Work around the Mali stencil discard bug the same way as the Adreno one. + * Show detected GPU driver bugs in system info + * Increase the difficulty of the GPU driver test + * Build: Use cinttypes not inttypes.h. + * Update zh_TW.ini + * Update zh_CN.ini + * Update gamecontrollerdb.txt + * Update zh_CN.ini + * Update zh_TW.ini + +------------------------------------------------------------------- +Sun Jun 12 14:51:41 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220612: + * Fix-ja_jp.ini + * Fix-ja_jp.ini + * Small struct layout optimization (stencil is never more than 8 bits, can sneak in next to the load actions) + * Rename Vulkan attachment operation variables. Prep for supporting specifying store op. + * Inline the CoreParameters accessor (eliminates a function call for each access) + * Shaun White Snowboarding: Use compat flag to eliminate a readback per frame + * Echochrome lines: Remove UV offsets, avoid reading the destination (much better codegen) + * Update zh_CN.ini + * Update zh_TW.ini + * Update zh_CN.ini + * Add centered line drawing for Echochrome. + * Take the absolute value when measuring pixel size for line expansion. + * Bump gradle again, .gitignore update for VS 2022 + * Add TODO to see if we can make better choices of time sources + * add additional rescaps + * fix indentation + * add runFullTrust, broadFileSystemAccess and expandedResources rescaps + * Add guard + * Update zh_CN.ini + * Add search on SDL + +------------------------------------------------------------------- +Tue Jun 07 13:00:24 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220607: + * Update id_ID.ini + * Update zh_CN.ini + * Small Change + * Update zh_TW.ini + * Update + * Indonesian Translation Update + * Fix commending wrong lines + * GLES: Fix sw transform flip in skip buffer mode. + * Update defaultshaders.ini + * Add files via upload + * Update upscale_spline36.fsh + * Return by copy + * Log missing lines on shutdown, update lang files + +------------------------------------------------------------------- +Mon May 30 12:59:10 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220530: + * Fix CpuCore select + * Savedata: Allow no key when not using SECURE modes. + * Osk: Fix fullscreen check. + * UI: Abandon focus movement on returning from pause. + * Update zh_CN.ini + * Config: Don't save --fullscreen unless changed. + * Fix themes as well + * Postshader: Avoid treating directories as files. + * Update th_TH.ini + +------------------------------------------------------------------- +Thu May 26 15:17:24 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220525: + * Update zh_TW.ini + * GPU: Fix skip buffer effects rendering. + * GPU: Hook Sol Trigger func to flush texture. + * Crash: Recover from unaligned CPU access. + +------------------------------------------------------------------- +Sun May 22 08:32:10 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220519: + * Update pt_BR.ini + * Add files via upload + * Update zh_CN.ini + * Detects connection establishment to AdhocServer by combining select and getpeername to increase accuracy. + * We might be missing this in the past. + * Fix sceNetAdhocPtpConnect issue on Windows (ie. Tales of VS) due to getpeername on Windows can return a success when the other side haven't Accepted the connection yet. + * Fix sceNetAdhocPtpConnect issue on non-Windows (ie. Taiko no Tatsujin Portable DX on Linux) due to select can flags an unconnected socket as writeable & readable on Linux. + * Added a note regarding SO_NOSIGPIPE + * Avoid compilation warning + * Avoid a possibility of fatal signal on non-Windows + * Added a note regarding select on Linux. + * (libretro) Add option to detect and notifiy frontend of internal frame rate changes (60 <-> 30 <-> 20 fps, etc.) + restore 'Duplicate Frames in 30 Hz Games' core option + * Update zh_CN.ini + * Limit minimum timeout to [0, 100, 200, ..., 5000] + * Fix missing GPU and run tools + * For Libretro -DUSE_DISCORD=1 is not added in CMakeLists.txt + * Let port offset cover the entire range in steps of 1k + * Remove options for network chat, since these only work for standalone + * Make MAC address configurable and randomize when not set + * Fix update of chat screen position + * Disable UPnP by default + * Correct indentation and some strings + * Expose networking options in libretro core + * Display: Ignore lagsync when auto frameskip on. + +------------------------------------------------------------------- +Sun May 15 00:42:19 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220512: + * Unused lang tool + * Update Russian translation + * Tweak -latomic preconditions in CMakeLists.txt. See comments on 14e9aab96e111 + * softgpu: Correct stencil debugging. + * softgpu: Fix viewport flag clean/dirty. + * Lang: Remove some duplicate lines in German. + * uptated German Translation + +------------------------------------------------------------------- +Sun May 08 00:21:52 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220507: + * Enable -latomic on Linux. Fixes build on Raspberry Pi 3b for me. + * Update .gitignore + +------------------------------------------------------------------- +Thu May 05 08:50:38 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220503: + * Replacement: Allow texture dump without replace. + * Fix shader blending regression (Vulkan) + * libkirk: Set Segher Boessenkool's original copyright notice + * Fix discrepancy in how the LMODE flag is calculated between VS and FS + * Vulkan: Correct size for tex mip level readback. + * Vulkan: Fix depth transition barrier. + * Remove the rather problematic limit on framebuffer copies + * Remove log output + * Combine barriers before blits/copies + * Convert some barriers to use the new barrier framework. + * Add utility class VulkanBarrier + * Combine renderpass "pretransitions" for barriers. + * Vulkan: Follow sampler best practices with auto max quality. + * Bump vulkan headers to SDK 1.3.211 + * Bump gradle + * Convert the blueToAlpha bool to a usageFlag. KEEP instead of CLEAR for depth during reinterpret. + * Oops, gotta update the cached framebuffer format in the other drawing paths too + * Finish BlueToAlpha functionality + * BlueToAlpha hack to avoid clearing the envmap in split/second :( + * Add option to redirect blue to alpha if 565 mode is rendered and mask is 0x0FFFFF. + * Move framebufFormat to gstate_c, so we can override it + * Dirty more state after reinterpret + * Minor cleanups in framebuffer manager + +------------------------------------------------------------------- +Sun May 01 00:22:49 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220430: + * Better default control config for the Retroid handheld Android console + * Detect ISO files from file contents if extensions are wrong. + * Allow to set InternalScreenRotation per game + * Fix lazy loading of metadata + +------------------------------------------------------------------- +Thu Apr 28 17:23:44 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220427: + * Logs unimplemented syscalls as error + * Partially fixes SenseMe app (will need HTTP/HTTPS support to be fully working) + * Vero4k doesn't report its NEON support - so check for underlying Arm Cortex-A53. + * GLES: Reverse alphamask when comparing to CLUT. + * Simplify alphasum checking for DXT textures, and fix a regression + * Make the existing ReinterpretFramebuffers/ShaderColorBitmask path work for Split/Second + * UWP: Fix loading the default theme. + * Fix regression with CLUT16 textures. See #12188 + * Minor cleanups + * Fix mip level bug in Vulkan texture cache. + * Make UI a bit more descriptive + * Add key bind to hotswap internal screen rotation + * Fix checks for the "fake mipmap" situation. + * Allow to change atlas per theme + +------------------------------------------------------------------- +Tue Apr 19 17:29:06 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220418: + * UI: Prevent save textures if not replacing. + * Vulkan: Save textures using a memory copy. + * Replacement: Avoid recheck saving within 5s. + * Replacement: Move IO checks to saving thread. + * Vulkan: Allow saving const tex levels. + * Add the slow comment to one more place + * Save textures on background tasks when texture dumping is enabled. + * Fix hang when choosing "Skip" in first-time init. Warning fix, comment add. + * Disable the compat.ini workaround for Crash Team Racing. Reported by Panderner to not be needed anymore. + * Add contributed data about a couple of prototypes to auto-expand memory to 64MB + * UWP crashfix - serialize accesses to each loader + * Crashfix + * Fix typo and optimize locking pattern in Event, make it the same as LimitedWaitable. + * Buildfix + * Try a clang pragma to avoid overeager auto-vectorization + * Buildfixing, correct NEON type usage + * SIMD-optimize CheckMask16 / CopyAndSumMask16 + * Delete a lot of specialized alpha checking code. + * Address feedback + * One more NEON optimization + * Quick NEON optimization of CheckMask32 + * SSE optimize a common case for video + * Yet another bugfix + * Debug log (remove before merge) + * Bugfixes, handle the rest of the cases + * Simplification and some cleanup + * Remove redundant line + * Fix DeIndexTexture + * Comments + * AND together colors while decoding, and then check against fullAlphaMask. + * Fix some NEON code that had bad compile-time checks (and some didn't compile) + * ARM32: Remove a lot of non-NEON fallback paths + * Always use the stable quick tex hash. Doesn't actually make a difference except on new CPU archs. + * Remove more function defines + * Merge separate NEON functions into the normal functions. + * Avoid creating a fileloader during shutdown. + * Make the number of threads flexible in the threadmanager test. Increase to 9, just because. + * Replacement: Avoid potential destructor crash. + * ThreadManager: Improve waitable destruction. + +------------------------------------------------------------------- +Sun Apr 10 00:24:09 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220409: + * Fix race condition in LimitedWaitable between Notify and Wait + * Fix misuses of cond.wait (should return true when you want to stop waiting!) + * OK, this does crash + * Time the stress test + * Add a scheduling stress test to TestThreadManager.cpp. + * Bump Android target version to 12L. + * Gradle bump + * Net: Ignore SIGPIPE errors on POSIX systems. + * Switch: Fix GL build typo. + * NetAdhoc: Fix warning. + * Switch: Avoid loading Vulkan. + * Detect number of cores in build script + +------------------------------------------------------------------- +Sun Mar 27 00:22:27 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220324: + * Increase the timeout when the connection to the AdhocServer is already in progress, avoids cutting off the connection if it took longer than expected but almost completed. + * Added a function to get SO_ERROR on a socket. + * Updated getLocalIp function to detects local IP better. + * Use the most recent posix socket id during blocking simulation, since there is a possibility for the socket id to changes after recreated. + * Recreate the socket before attempting to connect again after ECONNREFUSED/ENETUNREACH error, since reusing the socket only works on Windows. + * Caches IP-specific Port Offset to avoid recalculating them. + * Use reference when possible to avoid copying. + * Initialize some socket-related structs to avoid unexpected behavior. (May fix Unknown Source Port issue) + * Shouldn't return as soon as possible when a blocking PtpConnect is getting ECONNREFUSED error, since it should be treated as ETIMEDOUT on non-Windows platform + * Ignore disconnected error on PtpFlush. Fixes disconnection issue on some games (ie. R-Type Command) + +------------------------------------------------------------------- +Tue Mar 22 15:10:33 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220320: + * Debugger: Keep separate tag dirty for RAM. + * Debugger: Fast path write tag lookup. + * softgpu: Combine memcpy into single. + * Debugger: Improve tag on copies. + * softgpu: Avoid unnecessary clearMode checks. + * Manual generate Android APK + +------------------------------------------------------------------- +Sun Mar 20 00:18:38 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220313: + * Add a check to the elf loader, preventing a crash on bad module info addr + * softgpu: Plug bad leak of bin queue data. + * softgpu: Reduce memory usage on 32-bit. + * softgpu: Enure aligned bin queues. + * softgpu: Avoid some unaligned access on x86_32. + * Loaders: Fix 64-bit conversion warnings on 32-bit. + * Global: Fix some int/size_t conversion warnings. + +------------------------------------------------------------------- +Sun Mar 13 00:20:17 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220311: + * Add version to game screen + * Short circuit the texture replacement code + * Don't track small wide-stride block transfers + +------------------------------------------------------------------- +Sun Mar 06 00:20:34 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220305: + * softgpu: Fix build with basic logging. + * Windows: Correct root vol handling for UNC paths. + * Windows: Create SYSTEM directory early. + +------------------------------------------------------------------- +Sun Feb 27 22:22:06 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220224: + * Show game crashed + * GLES: Remove old invalidation code. + * GLES: Add asserts for correct blend usage. + * GLES: Use EXT not GLES3 for dualsrc blend support. + * Debugger: Track unchanged VRAM to avoid copy. + * Also delete vulkan_funcs.hpp, oops + * Debugger: Improve drawing range in softgpu. + * softgpu: Use region as a second scissor. + * softgpu: Correct accounting for pixel center. + * samplerjit: Correct nearest negative texture clamp. + * softgpu: Directly implement rectangle drawing. + * softgpu: Prepare dedicated rectangle path. + * softgpu: Cleanup some now unused state. + * softgpu: Remove incorrect offsetting for X/Y. + * softgpu: Use a const for subpixel screenpos factor. + * softgpu: Remove offset from screenpos. + * Upgrade to Vulkan 1.3 headers (and remove the .hpp files) + +------------------------------------------------------------------- +Sun Feb 20 00:23:56 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220219: + * Gradle upgrade again. Remove some redundant version checks + * Allow to set default background color in a theme + * Fix live update + * Make the untextured uniform buffer the same size. + * Disable darkening background, also allow to over saturate up to 2 + * Add UI Tint/Saturation settings + * samplerjit: Fix non-SSE4 bugs in jit. + * samplerjit: Fix non-AVX three-op shift. + * Unhide "Cache ISO in RAM" on ARM platform. Not the right block level. We should probably check for available RAM instead. + * More fixes for themes + * Merge item and button style, remove hightlighed and few other fixes + * Fix disabled bg + * Allow translation + * Add themes to assets + * GE Debugger: Highlight changed matrix values. + * GE Debugger: Highlight changed state values. + * GE Debugger: Track last gstate on stepping. + * Debugger: Allow custom draw and generic lists. + * Debugger: Avoid scroll on breakpoint add. + * Debugger: Cleanup some header pollution. + * softgpu: Track dirty vs really dirty per buffer. + * softgpu: Allow tri -> rect in transform. + * softgpu: Turn more tri strips into rects. + * softgpu: Combine tris to rects with ignored z too. + * softgpu: Cleanup todos on perspective correctness. + * softgpu: Avoid modifying source vertex data. + * Adress feedback + * remove more prebuilt stuff and huge rapidjson testsuite (unused) + * also delete pre-built ffmpeg windows libraries + * fix version string generation + * fix version number display + * try 2: remove pre-built ffmpeg and vulkan libraries + * remove pre-built ffmpeg and vulkan libraries + * fix mime type of xz tarball + * rename action title + * rename to tarball.yml + * use recursive submodules and checkoutv2 + * Forgot pragma, silly me + * Allow custom UI themes + * CI: add github action to build source release tarball with submodules + +------------------------------------------------------------------- +Sun Feb 13 00:25:44 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220211: + * Keep comment up to date + * GE Debugger: Avoid crash on Step Draw with flush. + * Debugger: Improve memory search result draw speed. + * Debugger: Search scratchpad/ram/vram consistently. + * GE Debugger: Correct icon list cleanup on shutdown. + * Add missing Config.h include + * GE Debugger: Optimize slab lookup. + * GE Debugger: Skip overhead without breakpoints. + * Generate dockerfile in GHA + * Vulkan: Avoid crash on double DeviceLost. + * Android: Correct truncation behavior. + * GE Debugger: Add filter to skip prim calls. + +------------------------------------------------------------------- +Sun Feb 06 23:00:42 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220206: + * Debugger: Avoid write tag lookup on small alloc. + * Debugger: Use detailed meminfo with breakpoints. + * GE Debugger: Allow modify/breakpoint on matrix. + * GE Debugger: Cleanup right click on state. + * GE Debugger: Allow setting secondary values. + * GE Debugger: Add breakpoint icon in state. + * GE Debugger: Cleanup vertex list redraw. + * GE Debugger: Flush downcount when changing lists. + * GE Debugger: Reduce playback command noise. + * samplerjit: Precalculate DXT1/3/5 offsets. + * Delete pauth_tool, useless since pauth keys known + * samplerjit: Reduce DXT5 decode code size a bit. + +------------------------------------------------------------------- +Wed Feb 02 15:12:29 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220201: + * samplerjit: Optimize DXT calc using BMI2. + * samplerjit: Avoid RCX gymanstics with BMI2. + * samplerjit: Decode colors with BMI2. + * softjit: Use PEXT to downsample colors. + * softjit: Use BMI2 to speed up dst color loads. + * softjit: Use BMI to simplify some masking. + * x86jit: Use BMI2 for variable shifts. + * softgpu: Avoid flush on END. + * Remove double flush in GE_CMD_END + * GPUCommon: Remove misleading flag FLUSHBEFORE. + * HLE: Fix some type convert warnings. + * Global: Reduce includes of common headers. + * Global: Fix some type conversion warnings. + * Common: Remove some unused files. + * Core: Skip stats reset while stepping. + * Display: Account for stepping time in graphs. + * Display: Move core counters/stats to HW file. + * Display: Move listeners to separate HW file. + * Headless: Improve total test time on Windows. + * Headless: Improve test init time. + * Debugger: Encourage better meminfo accounting. + * Build: Fix Windows unit tests. + * Kernel: Adjust timings of LwMutex and EventFlag. + * Build: Ignore nullable completeness on Android. + * UI: Cleanup some reference warnings. + * samplerjit: Fix non-shared CLUT on Linux. + * samplerjit: Validate compile in a unit test. + * samplerjit: Correct level lookup in nearest. + * samplerjit: Fix reg corruption in DXT funcs. + * samplerjit: Compile sampler funcs together. + * softjit: Add tests for compile success. + * Common: Avoid assert hang in CI. + * softjit: Fix stencil bug running out of regs. + * Build: Enable ccache for GitHub Actions. + +------------------------------------------------------------------- +Sun Jan 30 00:37:18 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220130: + * Build: Run unit tests after build. + * Headless: Remove stack corruption workaround. + * Headless: Add flag to run failing tests. + * unittest: Skip D3D shader compile on non-Windows. + * unittest: Correct Windows-specific Path tests. + * softgpu: Skip new CLUT if identical. + * unittest: Use more consistent output executable. + * unittest: Correct failing unit test. + * samplerjit: Reduce register waste. + * samplerjit: Avoid frac uv transfer to gen regs. + * Added another attempt to detect GETIFADDRS availability, which is more common on newer OS + * We might be missing the SIOCGIFADDR + * Added logs for error within SIOCGIFCONF section. + * Change a confusing variable naming + * Expose 2xBRZ texture filter option. + * Mp3: Delay low level init. + * Mp3: Output PCM data using even/odd buffer slot. + * Mp3: Allow decode with NULL out pointer. + * Mp3: Cleanup AuCtx constructor. + * Kernel: Fail threads with invalid priority. + * UWP buildfix + * Quick hack attempt around build failures + * Vulkan: Add support for listing all the available surface formats in system info. + +------------------------------------------------------------------- +Wed Jan 26 12:54:03 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220126: + * Rtc: Fix day of week stack bounds issue. + * Savedata: Fix indeterminate timing of field update. + * Ge: Fix sceGeBreak error case. + * softgpu: Cull no-pos and through s8 pos verts. + * softgpu: Fix cull in throughmode. + * samplerjit: Reduce transfers in nearest texel calc. + * samplerjit: Better vectorize UV linear calc. + * samplerjit: Reuse XMM reg for sizes. + * softgpu: Fix pixel ID for invalid blend factors. + * softjit: Fix src blend factor handling. + * GE Debugger: Allow GPU stepping while CPU stepping. + * GPU: Allow relative jumps and calls. + * GPU: Log and report when region1 is non-zero. + * Common: Fix conversion of 4444 to BGRA. + * softgpu: Correct clear mode dither. + * softgpu: Correct simple rectangles with mipmaps. + * samplerjit: Optimize texenv blending a bit. + * softgpu: Correct UV rotation for transformed rects. + * softgpu: Reuse transform state. + * softgpu: Add dirty flags for rasterization state. + * softgpu: Use dirty flags for render overlap checks. + * softgpu: Fix state race on screen offset. + * softgpu: Add a table and initial dirty flags. + +------------------------------------------------------------------- +Sun Jan 23 15:43:26 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220123: + * GE Debugger: Add option to auto flush. + * GE Debugger: Cut down on primary buttons. + * GE Debugger: Add a menu to the GE debugger window. + * samplerjit: Oops, forgot about 64 unpack. + * samplerjit: Fix Linux stack corruption. + * Build: Fix some minor warnings. + * Headless: Return an exit code on test failure. + * Common: Print assert failures to stderr. + * softgpu: Define constexpr var for older C++. + * x86jit: Check CALL dist for safe memory funcs. + * samplerjit: Fix some Linux register issues. + * Headless: Fix font tests on Linux. + * Build: Speed up macOS actions build a bit. + * Headless: Update passing tests. + * Clean up hardware tasseletion setting + * Show the software rendering option on all platforms + * softgpu: Avoid flush texturing from stride. + * softgpu: Flush framebuf only on change. + * softgpu: Remove bin asserts. + * softgpu: Use persistent bin task state. + * softgpu: Pack vertexdata a bit better. + * Reducing MatchingEvents processing delays. Fixes timing issue on a few games (Super Pocket Tennis, Thrillville Off the Rails, Lord of Arcana, Gundam: Senjou No Kizuna Portable, StarWars The Force Unleashed) + * Report sceNetAdhocMatchingInit to find out a list of games that use AdhocMatching library to be tested with. + * Discard HELLO Events when in the middle of Joining + * Updated notifyMatchingHandler comments + * Updated MatchingArgs comment + * Added comment related to EVENT_HELLO + * samplerjit: Reduce prolog/epilog spill. + * softjit: Simplify constant writes. + * softjit: Centralize some common funcs from sampler. + * softjit: Switch to constant pool. + * samplerjit: Simplify AVX shift-copies. + * UI: Reset ZIP install errors for new ZIPs. + * softgpu: Guide more SSE light factor handling. + * softgpu: Simplify mask check. + * softgpu: Reduce copying during clipping. + * Debugger: Fix crash in software renderer. + * softgpu: Fix block transfer flush detection. + * softgpu: Use SSE in lighting param computation. + * softgpu: Precompute some of screen space multiply. + * softpu: Premultiply matrix transforms. + * softgpu: Precompute state for vertex transform. + * softgpu: Correct texsize flush annotation. + * softgpu: Heuristic to avoid over-draining. + * softgpu: Reduce wasted memory. + * softgpu: Precompute lighting parameters. + * softgpu: Tune some queue sizes for perf. + * softgpu: Expose flush reasons/times in debug stats. + * softgpu: Reduce flushing for smaller textures. + * softgpu: Flush block xfer only on overlap too. + * softgpu: Decorate some stats for flushes. + * softgpu: Avoid texture/CLUT flush unless overlap. + * softgpu: Flush on offset changes. + * softgpu: Track CLUTs as states for binning. + * softgpu: Allow binning across prim calls. + * softgpu: Fix alpha blend with one/zero. + * softgpu: Correct dither matrix lookup. + * softgpu: Correct logic op state update. + * softgpu: Fix off-by-one size limit on bin queues. + * softgpu: Tune queue push/pop to reduce overhead. + * Debugger: Avoid flushing meminfo on write lookup. + * softgpu: Use cached state for screen offset. + * softgpu: Cache CLUT params in sampler state. + * softgpu: Move texenv color into sampler state. + * softgpu: Move tex size to cached state. + * softgpu: Remove z from DrawingCoords. + * softgpu: Move c++ tex func to sampler. + * softgpu: Use cached sampler state outside jit. + * softgpu: Cache line drawing state. + * softgpu: Use cached state for clear write mask. + * softgpu: Move fixed blend factor to draw pix state. + * softgpu: Cache colortest params in draw pix state. + * softgpu: Cache alpha/stencil test masks in state. + * softgpu: Cache logicOp in draw pixel state. + * softgpu: Use cached write mask in draw pixel. + * softgpu: Cache minz/maxz in draw pixel state. + * softgpu: Cache strides in draw pixel state. + * softgpu: Cache fog color draw pixel state. + * softgpu: Remove disable of cached pixel state. + * softgpu: Avoid double calculating screenpos. + * softgpu: Cleanup push/pop atomic handling. + * softgpu: Remove old task. + * softgpu: Stop storing model pos. + * softgpu: Stop storing normal/worldnormal/worldpos. + * softgpu: Enqueue batches of prims when binning. + * softgpu: Adjust binning thresholds. + * softgpu: Avoid thread ordering hazard. + * softgpu: Move threading into BinManager. + * softgpu: Handle scissor/range in BinManager. + * softgpu: Add class to manage and enqueue for bins. + * Fix buffer overflow issue on GameMode. + * GameMode Scheduler should only runs when there are active Master. + * Prevent creating another GameMode Master, since we don't support more than one Master data, and there is no ID related to Master. + * Fix an issue where more than one GameMode Scheduler can be scheduled. + * Fix an issue where sceNetAdhocGameModeCreateReplica blocked current thread for initial sync while Master data hasn't been created yet. + * Debugger: Cache list PC for softgpu tagging. + * softgpu: Centralize more prim drawing state. + * Updated sceNetAdhocGameModeUpdateReplica (Fix Bomberman Panic Bomber) + * Fix invalid argument issue of sceNetAdhocPtpAccept on Bomberman Panic Bomber. + * softgpu: Move tex filter setting check to state. + * softgpu: Move texture addresses to prim state. + * softgpu: Prepare more state per prim call. + * softgpu: Use func IDs instead of gstate more. + * softgpu: Compute func IDs once per batch of verts. + * Avoid confusion for translator + * Add files via upload + * Vulkan: Avoid crash on bad shader. + * softgpu: Output normals to GE debugger. + * Use System_GetPropertyInt to report the keyboard layout + * UI: Mention SDL in the credits when it's being used alongside Qt + * Update pt_BR.ini + * softgpu: Correct decal doubling without alpha. + * samplerjit: Skip processing alpha if unused. + * Add files via upload + * samplerjit: Fix alpha for 565 in linear lookup. + +------------------------------------------------------------------- +Sun Jan 09 14:04:25 UTC 2022 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20220109: + * softgpu: Skip threading when rendering to self. + * softgpu: Draw top left of rectangles first. + * GPU: Allow gcc/clang/icc runtime SSE4 usage. + * UI: Use comparison screenshot as frame dump icon. + * Headless: Fix compare of smaller buffers. + * softgpu: Use some SSE4 in triangle interpolation. + * softgpu: Limit minX/maxX per line. + * softgpu: Avoid rsqrt path for normals. + * softgpu: Clarify internal matrix multiply usage. + * Remove QWEmct from credits as requested + * softgpu: Retain floats in diffuse/specular. + * softgpu: Cleanup -NAN and diffuse factor. + * softgpu: Correct NAN spotlight exponent/direction. + * softgpu: Correct handling of NAN attenuation. + * softgpu: More closely match PSP light rounding. + * softgpu: Reuse SSE/NEON matrix code. + * softgpu: Use common SIMD matrix multiplies. + * softgpu: Separate calculation of S/T. + * UI: Fix some sign/size comparison warnings. + * samplerjit: Fix standard bufw check. + * samplerjit: Fix DXT compilation. + * samplerjit: Correct linear compile failure. + * softgpu: Correct mirroring in fastpath+nearest. + * softgpu: Improve usage of SSE for lighting. + * softjit: Keep mip S/T calc in SIMD. + * softjit: Use RIP access in color/depth off. + * softjit: Optimize depth range checks. + * softjit: Add describes here too. + * samplerjit: Remove old linear nearest paths. + * samplerjit: Avoid gather if overread could crash. + * samplerjit: Use VPGATHERDD for all types. + * samplerjit: Apply gather lookup to all CLUT4. + * samplerjit: Use VPGATHERDD for simple CLUT4 loads. + * samplerjit: Avoid a couple more copies in AVX. + * softgpu: Skip sample lookup if masked. + * softgpu: Stop specializing on miplevels. + * softgpu: Separately profile verts and lighting. + * samplerjit: Use nearest func in fast path too. + * softjit: Avoid const temp registers. + * sofjit: Correctly poison memory. + * softjit: Small bloom optimization. + * softjit: Optimize common case bloom blending. + * softgpu: Add code for tracking GPU writes. + * samplerjit: Decode colors in parallel. + * softgpu: Reduce interpolation if not needed. + * samplerjit: Implement nearest in jit. + * samplerjit: Refactor and reuse constant pool. + * samplerjit: Update nearest args, temp disable jit. + * softjit: Small optimizations. + * samplerjit: Create a separate fetch func. + * samplerjit: Refactor sampler ID calculation. + * softgpu: Correct off-by-one scissor mask. + * softgpu: Cleanup min/max tri range handling. + * GPU: Avoid spline crashes on bad data. + * softgpu: Skip processing scissored triangles. + * samplerjit: Correct wrong bufw at mip levels. + * samplerjit: Oops, fix Linux mipmap handling. + * PPGe: Disable dither in UI drawing. + * softjit: Fix regcache error when clearing. + * GPU: Fix bone matrix CALL opt corruption. + * Debugger: Correct delayed symbol listbox updates. + * samplerjit: Tiny dependency optimizations. + * samplerjit: Small AVX optimization to modulate. + * softgpu: Skip zero size triangles. + * samplerjit: Use SSSE3/SSE4 in linear filtering. + * x64jit: Add AVX2-only instructions. + * x64jit: Add AVX-only instructions. + * x64jit: Add AVX/AVX2 encodings. + * samplerjit: Pass funcs as an argument. + * samplerjit: Add an alternate profiling method. + * samplerjit: Correct level lookup offset. + * samplerjit: Minor SSE4 optimizations. + * samplerjit: Set stackArgPos_ early. + * samplerjit: Move texture function into jit. + * samplerjit: Lookup both mip tex values. + * samplerjit: Use regcache for linear filtering. + * samplerjit: Calculate mip level U/V/offsets. + * samplerjit: Take texptr/bufw as arrays. + * samplerjit: Handle mipmap width/height in S/T calc. + * samplerjit: Move S/T calculation into jit. + * samplerjit: Add more bits for S/T, skip multiply. + * softgpu: Correct non-SSE T clamp. + * softgpu: Reduce mipmap fraction to 4 bits. + * softgpu: Correct nearest filtering too. + * softgpu: Remove no longer possible report. + * softgpu: Make linear filtering more accurate. + * samplerjit: Handle unswizzled offsets too. + * samplerjit: Calculate texel byte offset as vector. + * samplerjit: Blend linear using integers. + * samplerjit: Accumulate results in an XMM. + * samplerjit: Pass u/v coords as vector. + +------------------------------------------------------------------- +Sun Dec 26 12:53:38 UTC 2021 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20211221: + * Debugger: Don't try to hash invalid RAM. + * Missing stdarg.h include + * Vulkan: Update profiler formatting for this. + * Make the Vulkan GPU log profiler a runtime developer setting. + * Add format string support to Vulkan log-profiler + * jit: Use a hash for invalidate/clear all. + * Vulkan: Disable profiler by default again. + +------------------------------------------------------------------- +Sun Dec 19 00:58:09 UTC 2021 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20211218: + * Improve the check for "slow" GPUs (with regards to tex scaling). See #15238 + * Path: Check for PSP case insensitively. + * Vulkan: Be more restrictive about texture upscaling on "slow" devices. + +------------------------------------------------------------------- +Thu Dec 16 11:41:22 UTC 2021 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20211214: + * GPU: Remove explicit rect/line depth cull. + * Fix text wrapping on PromptScreen by improving the layout + * Vulkan: Don't generate mipmaps for huge textures (probably menu backgrounds). Other minor fixes. + * Oops, accidentally enabled the logging GPU profiler by default. + * Address feedback (preserve alpha-ignoring equality checks) + * Minor fixes + * Turn video upscaling off. + * Optimize the xBRZ texture scaling shaders by 30% on Adreno, less on Mali + * Add hardcoded bool toggle for upscaling video, fix log, remove a comment, minor opt + * fmt + * Add a 2x2 downsampled version of 4xBRZ as 2xBRZ (save memory bandwidth) + * Make a copy of tex_4xbrz + * Debugger: Debounce updates to disasm text. + * Debugger: Never wait inactive on window update. + * Debugger: Remove double debug window updates. + * Debugger: Debounce list control updates. + * Debugger: Fix crash on ill-timed breakpoint update. + * Debugger: Debounce redraws using timers. + * Bump gradle version again + * Fixes when profiler is disabled. + * Switch to macros to easily disable the new profiling. Add missing scope + * Add a simple GPU profiler to profile individual events, rather than full passes. + * Build: Correct AppleClang warning handling. + * softjit: Fix a function arg template warning. + * softjit: Fix an enum type comparion error. + * Build: Avoid compiler flag warning. + * Build: Fix some format truncation warnings. + * interp: Correct vrndf2 range. + * Vulkan: Ignore VMA nullability warnings. + * UI: Clarify error message. + * Build: Remove old-style android/assets/. + * Split the atlas into three, well two, but two different cases for the fonts. + +------------------------------------------------------------------- +Sun Dec 12 01:01:49 UTC 2021 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20211211: + * softjit: Skip a common case CMP. + * softjit: Throw away regs allocated in conditionals. + * Name uniform buffers, add more asserts. Used this to track down the bug fixed in the previous commit. + * Fix an ordering issue causing us to hit some asserts + * Vulkan: Document lifetime for desc set pool. + * Vulkan: Fix use-after-free on shutdown. + * Vulkan: Use new desc sets in Draw and VulkanUtil. + * Vulkan: Refactor common desc pool handling. + * Vulkan: Avoid desc set error in tests. + * samplerjit: Fix DXT1/DXT5 register releasing. + * Vulkan is strict about scissor rect, so let's clamp centrally. + * FramebufferManagerCommon: Fix some issues where currentRenderVfb_ could get out of sync. + * Fix Tekken 5 Dark Resurrection Multiplayer + * ThreadManager: Check for exchange failure. + * Updated GetPtpStat + * Updated GetPdpStat + * Added a note regarding FIONREAD behavior on different platforms. + * Discard PDP packets from an IP that can't be resolved to Player's MAC address (simple version) to prevent confusing the game, since the source MAC won't be updated and may contains invalid MAC. Added TODO to handle unresolvable packets properly. + * Added another returned error code on PdpCreate + * Allow Sending PDP to non-existing MAC address (except invalid mac address). Fixes an issue on NBA Ballers Rebound where players unable to enter Multiplayer Menu again after getting connection issue. + * Vulkan: More scissor dimension checks. See #15207 + * Vulkan: Correct the max level to generate mipmap calculation + * Very minor logging improvement + * Detect and specify the max available Vulkan version, up to 1.2. + * (libretro) Force enable g_Config.bRenderDuplicateFrames + minor audio buffer tweaks + * Use system headers when building with system miniupnpc + * ThreadManager: Use separate pool for IO blocking. + * ThreadManager: Make TaskType a property of a Task. + * Android: Optimize content URI exists check. + * softgpu: Correct alpha on add + invalid texfuncs. + * softgpu: Match texenv blend texfunc accurately. + * softgpu: Make decal tex func more accurate. + * softgpu: Correct accuracy on fast path modulate. + * softgpu: Correct accuracy of MODULATE texfunc. + * (Libretro) Reimplement audio - * Move audio callback to the main thread (fixes the longstanding hang issues when using OpenGL) * 64ms on desktop isn't enough due to excess buffering going on in this emulator. Set default to 128ms instead. * Different audio approach is necessary due to the following: it sets a fixed framerate of 60, then sends either too many or too few audio samples per frame in order to force the frontend to run at the correct speed (.e.g. making the frontend run in slow motion for games that run at 30fps) - this is undesirable for libretro, we want a consistent amount of audio samples per batch. See further comments in libretro.cpp for more information + * softgpu: Template away uncommon mip usage. + * softgpu: Remove useless switch by UV gen mode. + * softgpu: Avoid texlevel check when maxlevel is 0. + * softgpu: Keep arguments in vectors for sampling. + * Debugger: Reduce meminfo block check hazard. + * Debugger: Speed up GetFunctionSize(). + * samplerjit: Refactor nearest using reg cache. + * Make CRT animation speed configurable + +------------------------------------------------------------------- +Sun Dec 05 00:54:45 UTC 2021 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20211204: + * Common: Correct shm_open() call. + * Psmf: Check for invalid stream/EP map entry counts. + * Vulkan: Cleanup pipeline var shadowing. + * softjit: Make it an error to unlock a temp. + * softjit: Get rid of pointless AGE00 tests. + * softjit: Normalize some stencil test patterns. + * softjit: Avoid tests for greater than 0. + * softjit: Reduce ditherMatrix to 8-bit. + * softjit: Oops, correct allocation typo. + * Update Tagalog (tg_PH) with new translations by Gamemulatorer. Fixes #15189 + * Display the graphics API on logoscreen on more platforms + * ext: Update vulkan headers + * ext: Allow use system glew with system ffmpeg (macOS) + * ext: Support building with system miniUPnPc + * cmake: Generate git-version.cpp in the build dir + * ext/SPIRV-Cross-build: Update cmake version + * softjit: Fix locked stencil reg. + * headless: Enable softgpu jit for tests. + * softjit: Centralize argument register allocation. + * softjit: Fix fog typo causing locking bug. + * jit: Lock around changes to the jit pointer. + * softjit: Fix PixelFuncID arg on non-Windows x64. + * Math3D: Allow construction from NEON vectors. + * sofjit: Move common types to reg cache header. + * softjit: Rename reg cache so it can be reused. + * softjit: Cleanup, add other arch types to regcache. + * softjit: Disallow use of register after unlock. + * softjit: Simplify regcache usage as purpose only. + * samplerjit: Allow disabling linear too, oops. + * softjit: Decide stack offset on compile. + * softjit: Ensure all regs are released. + * softjit: Manage args in the register cache. + * softjit: Fix disabled cache. + * ThreadManager: Protect against teardown hang/leak. + * ThreadManager: Lock notify on global queue add. + * ThreadManager: Use atomic task optimistically. + * jit: More atomically update pointer. + * softjit: Reduce jit pool size a bit. + * softjit: Precalculate write mask and dither. + * softjit: Log any failed compiles. + * softjit: Keep color 16-bit when useful. + * softjit: Optimize common blend inverse alpha case. + * softjit: Add alpha blending factors. + * softjit: Cache zero vector to avoid recreating. + * softjit: Prefer easier to refill regs. + * softjit: Minor tweak to reg order for XCHG. + * softjit: Implement min/max/absdiff blending. + * softjit: Use SSE4.1 for fog and dither a bit. + * softjit: Fix dithering bug. + * softjit: Optimize typical alpha/depth test. + * softjit: Apply logic ops. + * softjit: Apply color write mask. + * softjit: Refactor color conv to dedicated funcs. + * softjit: Add config setting to enable/disable. + * softjit: Initial color write. + * softjit: Add dithering. + * softjit: Keep x and y args for dither. + * softjit: Depth and stencil testing. + * softjit: Optimize some imm sizes. + * softjit: Add helpers to get framebuf offsets. + * softjit: Add levels of register locking. + * softjit: Add color test. + * softjit: Add alpha test. + * softjit: Initial depthrange, fog. + * softjit: Add register cache for softjit. + * softjit: Switch to the __vectorcall convention. + * softjit: Add stubbed DrawPixel for x64. + * PPGe: Prevent special ampersand handling. + +------------------------------------------------------------------- +Sun Nov 28 01:01:16 UTC 2021 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20211127: + * SDL: Respect inflight frames setting in GL. + * softgpu: Correct alpha blend subtract on negative. + * softgpu: Use KEEP for any invalid stencil ops. + * vertexjit: Optimize 5551 read a bit on x86/x64. + * sofgpu: Disable alpha blend for invalid equations. + * softgpu: Accurately alpha blend. + * softgpu: Use 0 alpha for 565 alpha blending. + * softgpu: Correct tagging of vertexjit. + * softgpu: Fix stencil DECR on 5551. + * softgpu: Use ALWAYS for alpha/depth test in clear. + * Add review suggestion. + * Destroy backbuffer when shutting down vulkan conext. + * Build with correct preprocessor flag + * Remove the old allocator visualizer. Keep the setting but hide it. Other feedback + * Buildfixes + * Fix bugs in deletion. Delete redundant functions. + * Switch framebuffers to use VMA for allocation as well + * Convert the PushBuffer to use VMA. + * Change the PushBuffer API a bit to not take explicit memory types. + * Remove the VulkanDeviceAllocator + * More integration, use VMA in VulkanTexture + * Start integrating VMA + * Add VMA to ext/ (AMD's Vulkan Memory Allocator) + * Minor refactor + * Comment improvement, restore accidental deletion + * Add a hidden setting to turn off the shader cache, for shader compile performance work + * softgpu: Fix PixelFuncID size. + * Vulkan: Add a single background thread for pipeline creation + * VulkanRenderManager: Add deferred pipeline creation (to get it off the CPU thread) + * softgpu: Add func to tersely name pixel funcs. + * softgpu: Template draw pixel on fb format. + * softgpu: Move draw pixel code to separate file. + * Vulkan: Fix crash on shutdown after device lost. + * softgpu: Clamp and round fog by mantissa bits. + * softgpu: Confirm mask doesn't hit stencil REPLACE. + * softgpu: Force off alpha blend if uselessly on. + * softgpu: Specialize sprite based on pixel func ID. + * softgpu: Use pixel func ID to draw pixels. + * softgpu: Add func to calculate pixel func ID. + +------------------------------------------------------------------- +Sun Nov 21 00:55:21 UTC 2021 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20211120: + * Threading: Remove a level of indirection from Promise and Mailbox. + * On desktop, change an empty current directory to "/" on startup. + * Prevent chat menu touch to fall throght + * Fix crash on close content after cached GPU pointer cleanup. + * Fix DPAD size calculation + +------------------------------------------------------------------- +Wed Nov 17 16:50:29 UTC 2021 - pousaduarte@gmail.com + +- Update to version 1.12.3~git20211115: + * UI: Switch to font texture for debug stats. + * softgpu: Clear by dividing out subpixel first. + * softgpu: Add some comments and cleanup. + * softgpu: Tune thread minimums better. + * sfotgpu: Use threads to apply clears. + * softgpu: Combine sliced rectangles. + * softgpu: Split out rectangle path for combining. + * softgpu: Fixup range cull for fans and fast path. + * softgpu: Fast path rectangles as fans. + * softgpu: Use range loops for sprite fast path. + * Vulkan: Don't cache vulkan_ with draw context. + * GPU: Cleanup some extra pointers in fb managers. + * softgpu: Check flat z in fast path. + * softgpu: Fast path triangles without textures. + * Debugger: Fix crash on stepping/break. + * softgpu: Fix leaked range flag on cull. + ------------------------------------------------------------------- Sun Nov 14 00:58:18 UTC 2021 - pousaduarte@gmail.com diff --git a/ppsspp.spec b/ppsspp.spec index 9fc6dc2..e77dc02 100644 --- a/ppsspp.spec +++ b/ppsspp.spec @@ -1,7 +1,7 @@ # # spec file for package ppsspp # -# Copyright (c) 2021 SUSE LLC +# Copyright (c) 2022 SUSE LLC # # All modifications and additions to the file contributed by third parties # remain the property of their copyright owners, unless otherwise agreed @@ -18,7 +18,7 @@ %define _lto_cflags %{nil} Name: ppsspp -Version: 1.12.3~git20211114 +Version: 1.13.2~git20221123 Release: 0 Summary: PlayStation Portable Emulator License: GPL-2.0-or-later