- Update to 1.15.0.
* New loop unroll pragmas: #pragma unroll and #pragma nounroll
directives provide loop unrolling optimization hints to the
compiler. This pragma may be used immediately before a loop
statement. Currently, this functionality is limited to uniform
for and do-while.
* More efficient packed_[load|store]_active() stdlib functions
implementation (up to 2.5x faster), which now supports 64 bit types.
* New cpus: icelake-server, tigerlake , alderlake, sapphirerapids.
* Several stability fixes related to SOA types, bool varying type
initialization, broken alignment information, type scoping.
* Compile time improvements.
* Initial support of ahead of time compilation to oneAPI Level Zero
binary format using --emit-zebin switch. You can use this binary
from ISPC Runtime by setting
* ISPCRT_USE_ZEBIN env variable to 1. Please note that SPIR-V format
is still a recommended and default way.
* Initial function pointers implementation.
* Global atomics support.
* Double math functions support.
* Memory functions support.
* Reworked masking approach. We disabled genx hardware mask by
default and use a software mask by default.
* Improved address spaces differentiation.
* Initial debug support.
* TGLLP (TigerLake mobile CPU) support (--cpu=TGLLP).
OBS-URL: https://build.opensuse.org/request/show/886453
OBS-URL: https://build.opensuse.org/package/show/devel:tools:compiler/ispc?expand=0&rev=11
Note, we don't do any GPU builds yet, because
- API is not stable
- additional dependencies (that are not available, yet)
- Update to version v1.14.1:
A minor ISPC update with a bug fix for AVX512 detection problem on macOS
(for more details see issue #1854) and update of GPU version to use Level0 v1.0.
CPU binaries are based on patched LLVM 10.0.1.
- Update to version v1.14.0:
An ISPC release with several improvements for CPU and initial support of Intel
graphics hardware architectures. The binaries in this release include CPU versions
for Windows, Linux, and macOS, as previous releases, plus a GPU-enabled Linux binary,
which supports both CPU and GPU. CPU binaries are based on patched LLVM 10.0.1.
CPU changes include:
+ new avx2-i8x32, avx2-i16x16, avx512skx-i8x64, avx512skx-i16x32 targets.
"generic" targets were removed.
+ several stability fixes, including bugs discovered during fuzzing ISPC by YARPGen.
+ integer division performance improvements.
+ support for __vectorcall calling convention on Windows x64 (enabled by
'--vectorcall')
OBS-URL: https://build.opensuse.org/request/show/846843
OBS-URL: https://build.opensuse.org/package/show/devel:tools:compiler/ispc?expand=0&rev=9
