openblas/openblas.spec

581 lines
16 KiB
RPMSpec
Raw Normal View History

#
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
# spec file
#
# 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
# upon. The license for this file, and modifications and additions to the
# file, is the same license as for the pristine package itself (unless the
# license for the pristine package is not an Open Source License, in which
# case the license is the MIT License). An "Open Source License" is a
# license that conforms to the Open Source Definition (Version 1.9)
# published by the Open Source Initiative.
# Please submit bugfixes or comments via https://bugs.opensuse.org/
#
%global flavor @BUILD_FLAVOR@%{nil}
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
%define _vers 0_3_20
%define vers 0.3.20
%define pname openblas
%bcond_with ringdisabled
%if "%flavor" == ""
%define package_name %{pname}
ExclusiveArch: do_not_build
%endif
%global build_flags USE_THREAD=1 USE_OPENMP=1
%if "%flavor" == "serial"
%define build_flags USE_THREAD=0 USE_OPENMP=0
%define openblas_so_prio 20
# we build devel packages only from one flavor
%define build_devel 1
%{bcond_with hpc}
- Update to version 0.3.14 common: * Fixed a race condition on thread shutdown in non-OpenMP builds * Fixed custom BUFFERSIZE option getting ignored in gmake builds * Fixed CMAKE compilation of the TRMM kernels for GENERIC platforms * Added CBLAS interfaces for CROTG, ZROTG, CSROT and ZDROT * Improved performance of OMATCOPY_RT across all platforms * Changed perl scripts to use env instead of a hardcoded /usr/bin/perl * Fixed potential misreading of the GCC compiler version in the build scripts * Fixed convergence problems in LAPACK complex GGEV/GGES (Reference-LAPACK #477) * Reduced the stacksize requirements for running the LAPACK testsuite (Reference-LAPACK #335) RISC V: * Fixed compilation on RISCV (missing entry in getarch) POWER: * Fixed compilation for DYNAMIC_ARCH with clang and with older gcc versions * Added support for compilation on FreeBSD/ppc64le * Added optimized POWER10 kernels for SSCAL, DSCAL, CSCAL, ZSCAL * Added optimized POWER10 kernels for SROT, DROT, CDOT, SASUM, DASUM * Improved SSWAP, DSWAP, CSWAP, ZSWAP performance on POWER10 * Improved SCOPY and CCOPY performance on POWER10 * Improved SGEMM and DGEMM performance on POWER10 * Added support for compilation with the NVIDIA HPC compiler x86_64: * Added an optimized bfloat16 GEMM kernel for Cooperlake * Added CPUID autodetection for Intel Rocket Lake and Tiger Lake cpus * Improved the performance of SASUM,DASUM,SROT,DROT on AMD Ryzen cpus * Added support for compilation with the NAG Fortran compiler * Fixed recognition of the AMD AOCC compiler * Fixed compilation for DYNAMIC_ARCH with clang on Windows * Added support for running the BLAS/CBLAS tests on Windows OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=120
2021-03-18 09:47:05 +01:00
%endif
%if "%flavor" == "pthreads"
- Update to version 0.3.14 common: * Fixed a race condition on thread shutdown in non-OpenMP builds * Fixed custom BUFFERSIZE option getting ignored in gmake builds * Fixed CMAKE compilation of the TRMM kernels for GENERIC platforms * Added CBLAS interfaces for CROTG, ZROTG, CSROT and ZDROT * Improved performance of OMATCOPY_RT across all platforms * Changed perl scripts to use env instead of a hardcoded /usr/bin/perl * Fixed potential misreading of the GCC compiler version in the build scripts * Fixed convergence problems in LAPACK complex GGEV/GGES (Reference-LAPACK #477) * Reduced the stacksize requirements for running the LAPACK testsuite (Reference-LAPACK #335) RISC V: * Fixed compilation on RISCV (missing entry in getarch) POWER: * Fixed compilation for DYNAMIC_ARCH with clang and with older gcc versions * Added support for compilation on FreeBSD/ppc64le * Added optimized POWER10 kernels for SSCAL, DSCAL, CSCAL, ZSCAL * Added optimized POWER10 kernels for SROT, DROT, CDOT, SASUM, DASUM * Improved SSWAP, DSWAP, CSWAP, ZSWAP performance on POWER10 * Improved SCOPY and CCOPY performance on POWER10 * Improved SGEMM and DGEMM performance on POWER10 * Added support for compilation with the NVIDIA HPC compiler x86_64: * Added an optimized bfloat16 GEMM kernel for Cooperlake * Added CPUID autodetection for Intel Rocket Lake and Tiger Lake cpus * Improved the performance of SASUM,DASUM,SROT,DROT on AMD Ryzen cpus * Added support for compilation with the NAG Fortran compiler * Fixed recognition of the AMD AOCC compiler * Fixed compilation for DYNAMIC_ARCH with clang on Windows * Added support for running the BLAS/CBLAS tests on Windows OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=120
2021-03-18 09:47:05 +01:00
%define build_flags USE_THREAD=1 USE_OPENMP=0
%ifarch %ix86 x86_64
Accepting request 966356 from home:eeich:branches:science - Build PPC64LE libraries with the lastest gcc available to take advantage of instruction sets in later CPUs used in the CPU specific kernels (jsc#SLE-18143, bsc#1197721). For fortran use the stock compiler to avoid compatibility issues between different versions of libfortran. This is relevant for Leap/SLE only. It may be dropped once gcc < 10 is no longer supported. - Do the same for x86_64 on SLE to make sure Cooperlake support is built properly. - Remove: * Do-not-attempt-to-check-host-CPU-if-TARGET-is-set.patch * Create-independent-kernel-Makfile-configuration-when-building-DYNAMIC_ARCH.patch * For-DYNAMIC_ARCH-don-t-use-sbgemm_r-as-parameter.c-doesn-t-get-build.patch Instead, add from upstream: * Define-sbgemm_r-to-fix-DYNAMIC_ARCH-builds.patch * Remove-extraneous-and-wrong-definition-of-sbgemm_r-on-x86_64.patch * Fix-checks-for-AVX512-and-atomics.patch * Revert-AVX512-capability-check-from-PR-1980-moved-to-build.patch * Use-CC-and-full-command-line-instead-of-hard-coding-gcc-for-AVX512-checking.patch * Utilize-compiler-AVX512-capability-info-from-c_check-when-building-getarch.patch BLAS/LAPACK implementation (boo#1177260). 'openblas-common-devel' will provide the removed 'openblas-devel-headers' while the arch specific 'preferred' flavor will provide the removed 'openblas-devel'. OBS-URL: https://build.opensuse.org/request/show/966356 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=131
2022-04-01 09:59:26 +02:00
%define arch_flavor 1
%define openblas_so_prio 50
%else
%define openblas_so_prio 20
%endif
%{bcond_with hpc}
%endif
%if "%flavor" == "openmp"
%ifarch %ix86 x86_64
%define openblas_so_prio 20
%else
Accepting request 966356 from home:eeich:branches:science - Build PPC64LE libraries with the lastest gcc available to take advantage of instruction sets in later CPUs used in the CPU specific kernels (jsc#SLE-18143, bsc#1197721). For fortran use the stock compiler to avoid compatibility issues between different versions of libfortran. This is relevant for Leap/SLE only. It may be dropped once gcc < 10 is no longer supported. - Do the same for x86_64 on SLE to make sure Cooperlake support is built properly. - Remove: * Do-not-attempt-to-check-host-CPU-if-TARGET-is-set.patch * Create-independent-kernel-Makfile-configuration-when-building-DYNAMIC_ARCH.patch * For-DYNAMIC_ARCH-don-t-use-sbgemm_r-as-parameter.c-doesn-t-get-build.patch Instead, add from upstream: * Define-sbgemm_r-to-fix-DYNAMIC_ARCH-builds.patch * Remove-extraneous-and-wrong-definition-of-sbgemm_r-on-x86_64.patch * Fix-checks-for-AVX512-and-atomics.patch * Revert-AVX512-capability-check-from-PR-1980-moved-to-build.patch * Use-CC-and-full-command-line-instead-of-hard-coding-gcc-for-AVX512-checking.patch * Utilize-compiler-AVX512-capability-info-from-c_check-when-building-getarch.patch BLAS/LAPACK implementation (boo#1177260). 'openblas-common-devel' will provide the removed 'openblas-devel-headers' while the arch specific 'preferred' flavor will provide the removed 'openblas-devel'. OBS-URL: https://build.opensuse.org/request/show/966356 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=131
2022-04-01 09:59:26 +02:00
%define arch_flavor 1
%define openblas_so_prio 50
%endif
%{bcond_with hpc}
- Update to version 0.3.14 common: * Fixed a race condition on thread shutdown in non-OpenMP builds * Fixed custom BUFFERSIZE option getting ignored in gmake builds * Fixed CMAKE compilation of the TRMM kernels for GENERIC platforms * Added CBLAS interfaces for CROTG, ZROTG, CSROT and ZDROT * Improved performance of OMATCOPY_RT across all platforms * Changed perl scripts to use env instead of a hardcoded /usr/bin/perl * Fixed potential misreading of the GCC compiler version in the build scripts * Fixed convergence problems in LAPACK complex GGEV/GGES (Reference-LAPACK #477) * Reduced the stacksize requirements for running the LAPACK testsuite (Reference-LAPACK #335) RISC V: * Fixed compilation on RISCV (missing entry in getarch) POWER: * Fixed compilation for DYNAMIC_ARCH with clang and with older gcc versions * Added support for compilation on FreeBSD/ppc64le * Added optimized POWER10 kernels for SSCAL, DSCAL, CSCAL, ZSCAL * Added optimized POWER10 kernels for SROT, DROT, CDOT, SASUM, DASUM * Improved SSWAP, DSWAP, CSWAP, ZSWAP performance on POWER10 * Improved SCOPY and CCOPY performance on POWER10 * Improved SGEMM and DGEMM performance on POWER10 * Added support for compilation with the NVIDIA HPC compiler x86_64: * Added an optimized bfloat16 GEMM kernel for Cooperlake * Added CPUID autodetection for Intel Rocket Lake and Tiger Lake cpus * Improved the performance of SASUM,DASUM,SROT,DROT on AMD Ryzen cpus * Added support for compilation with the NAG Fortran compiler * Fixed recognition of the AMD AOCC compiler * Fixed compilation for DYNAMIC_ARCH with clang on Windows * Added support for running the BLAS/CBLAS tests on Windows OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=120
2021-03-18 09:47:05 +01:00
%endif
%if "%flavor" == "gnu-hpc"
%define compiler_family gnu
%undefine c_f_ver
%{bcond_without hpc}
%endif
%if "%flavor" == "gnu-hpc-pthreads"
%define compiler_family gnu
%undefine c_f_ver
%define ext pthreads
%define build_flags USE_THREAD=1 USE_OPENMP=0
%{bcond_without hpc}
%endif
%if "%flavor" == "gnu7-hpc"
%define compiler_family gnu
%define c_f_ver 7
%{bcond_without hpc}
%endif
%if "%flavor" == "gnu7-hpc-pthreads"
%define compiler_family gnu
%define c_f_ver 7
%define ext pthreads
%define build_flags USE_THREAD=1 USE_OPENMP=0
%{bcond_without hpc}
%endif
%if "%flavor" == "gnu8-hpc"
%define compiler_family gnu
%define c_f_ver 8
%{bcond_without hpc}
%endif
%if "%flavor" == "gnu8-hpc-pthreads"
%define compiler_family gnu
%define c_f_ver 8
%define ext pthreads
%define build_flags USE_THREAD=1 USE_OPENMP=0
%{bcond_without hpc}
%endif
%if "%flavor" == "gnu9-hpc"
%define compiler_family gnu
%define c_f_ver 9
%{bcond_without hpc}
%endif
%if "%flavor" == "gnu9-hpc-pthreads"
%define compiler_family gnu
%define c_f_ver 9
%define ext pthreads
%define build_flags USE_THREAD=1 USE_OPENMP=0
%{bcond_without hpc}
%endif
%if "%flavor" == "gnu10-hpc"
%define compiler_family gnu
%define c_f_ver 10
%{bcond_without hpc}
%endif
%if "%flavor" == "gnu10-hpc-pthreads"
%define compiler_family gnu
%define c_f_ver 10
%define ext pthreads
%define build_flags USE_THREAD=1 USE_OPENMP=0
%{bcond_without hpc}
%endif
%ifarch ppc64le x86_64 s390x
%if 0%{?c_f_ver} > 9
%else
%if 0%{?sle_version} == 150400
%define cc_v 11
%endif
%if 0%{?sle_version} == 150300
%define cc_v 10
%endif
%endif
%endif
%if %{without hpc}
%if 0%{!?package_name:1}
%define package_name %{pname}_%{flavor}
%endif
%define so_v 0
%define p_prefix %_prefix
%define p_includedir %_includedir/%pname
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
%define p_libdir %_libdir/openblas%{?flavor:-%{flavor}}
%define p_cmakedir %{p_libdir}/cmake/%{pname}
%define num_threads 64
%else
# Magic for OBS Staging. Only build the flavors required by
# other packages in the ring.
%if %{with ringdisabled}
ExclusiveArch: do_not_build
%endif
%define package_name %{hpc_package_name %_vers}
%define p_prefix %hpc_prefix
%define p_includedir %hpc_includedir
%define p_libdir %hpc_libdir
%define p_cmakedir %{hpc_libdir}/cmake
%define num_threads 256
%{hpc_init -c %{compiler_family} %{?c_f_ver:-v %{c_f_ver}} %{?ext:-e %{ext}}}
%endif
Name: %{package_name}
Version: %vers
Release: 0
Summary: An optimized BLAS library based on GotoBLAS2
License: BSD-3-Clause
Group: Productivity/Scientific/Math
URL: http://www.openblas.net
Source0: https://github.com/xianyi/OpenBLAS/archive/v%{version}.tar.gz#/OpenBLAS-%{version}.tar.gz
Source1: README.SUSE
Source2: README.HPC.SUSE
Accepting request 966356 from home:eeich:branches:science - Build PPC64LE libraries with the lastest gcc available to take advantage of instruction sets in later CPUs used in the CPU specific kernels (jsc#SLE-18143, bsc#1197721). For fortran use the stock compiler to avoid compatibility issues between different versions of libfortran. This is relevant for Leap/SLE only. It may be dropped once gcc < 10 is no longer supported. - Do the same for x86_64 on SLE to make sure Cooperlake support is built properly. - Remove: * Do-not-attempt-to-check-host-CPU-if-TARGET-is-set.patch * Create-independent-kernel-Makfile-configuration-when-building-DYNAMIC_ARCH.patch * For-DYNAMIC_ARCH-don-t-use-sbgemm_r-as-parameter.c-doesn-t-get-build.patch Instead, add from upstream: * Define-sbgemm_r-to-fix-DYNAMIC_ARCH-builds.patch * Remove-extraneous-and-wrong-definition-of-sbgemm_r-on-x86_64.patch * Fix-checks-for-AVX512-and-atomics.patch * Revert-AVX512-capability-check-from-PR-1980-moved-to-build.patch * Use-CC-and-full-command-line-instead-of-hard-coding-gcc-for-AVX512-checking.patch * Utilize-compiler-AVX512-capability-info-from-c_check-when-building-getarch.patch BLAS/LAPACK implementation (boo#1177260). 'openblas-common-devel' will provide the removed 'openblas-devel-headers' while the arch specific 'preferred' flavor will provide the removed 'openblas-devel'. OBS-URL: https://build.opensuse.org/request/show/966356 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=131
2022-04-01 09:59:26 +02:00
Patch1: Define-sbgemm_r-to-fix-DYNAMIC_ARCH-builds.patch
Patch2: Remove-extraneous-and-wrong-definition-of-sbgemm_r-on-x86_64.patch
Patch3: Do-not-include-symbols-defined-in-driver-others-parameter.c-in-DYNAMIC_BUILD.patch
Patch4: Utilize-compiler-AVX512-capability-info-from-c_check-when-building-getarch.patch
Patch5: Revert-AVX512-capability-check-from-PR-1980-moved-to-build.patch
Patch6: Fix-checks-for-AVX512-and-atomics.patch
Patch7: Use-CC-and-full-command-line-instead-of-hard-coding-gcc-for-AVX512-checking.patch
# PATCH-FIX-UPSTREAM openblas-noexecstack.patch
Patch101: openblas-noexecstack.patch
# PATCH port
Patch102: openblas-s390.patch
Patch103: openblas-ppc64be_up2_p8.patch
BuildRoot: %{_tmppath}/%{name}-%{version}-build
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
#BuildRequires: cmake
BuildRequires: memory-constraints
Accepting request 966356 from home:eeich:branches:science - Build PPC64LE libraries with the lastest gcc available to take advantage of instruction sets in later CPUs used in the CPU specific kernels (jsc#SLE-18143, bsc#1197721). For fortran use the stock compiler to avoid compatibility issues between different versions of libfortran. This is relevant for Leap/SLE only. It may be dropped once gcc < 10 is no longer supported. - Do the same for x86_64 on SLE to make sure Cooperlake support is built properly. - Remove: * Do-not-attempt-to-check-host-CPU-if-TARGET-is-set.patch * Create-independent-kernel-Makfile-configuration-when-building-DYNAMIC_ARCH.patch * For-DYNAMIC_ARCH-don-t-use-sbgemm_r-as-parameter.c-doesn-t-get-build.patch Instead, add from upstream: * Define-sbgemm_r-to-fix-DYNAMIC_ARCH-builds.patch * Remove-extraneous-and-wrong-definition-of-sbgemm_r-on-x86_64.patch * Fix-checks-for-AVX512-and-atomics.patch * Revert-AVX512-capability-check-from-PR-1980-moved-to-build.patch * Use-CC-and-full-command-line-instead-of-hard-coding-gcc-for-AVX512-checking.patch * Utilize-compiler-AVX512-capability-info-from-c_check-when-building-getarch.patch BLAS/LAPACK implementation (boo#1177260). 'openblas-common-devel' will provide the removed 'openblas-devel-headers' while the arch specific 'preferred' flavor will provide the removed 'openblas-devel'. OBS-URL: https://build.opensuse.org/request/show/966356 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=131
2022-04-01 09:59:26 +02:00
%if 0%{?cc_v:1}
BuildRequires: gcc%{?cc_v}-fortran
%endif
%if %{without hpc}
BuildRequires: gcc-fortran
BuildRequires: update-alternatives
Requires(post): update-alternatives
- Update to version 0.3.14 common: * Fixed a race condition on thread shutdown in non-OpenMP builds * Fixed custom BUFFERSIZE option getting ignored in gmake builds * Fixed CMAKE compilation of the TRMM kernels for GENERIC platforms * Added CBLAS interfaces for CROTG, ZROTG, CSROT and ZDROT * Improved performance of OMATCOPY_RT across all platforms * Changed perl scripts to use env instead of a hardcoded /usr/bin/perl * Fixed potential misreading of the GCC compiler version in the build scripts * Fixed convergence problems in LAPACK complex GGEV/GGES (Reference-LAPACK #477) * Reduced the stacksize requirements for running the LAPACK testsuite (Reference-LAPACK #335) RISC V: * Fixed compilation on RISCV (missing entry in getarch) POWER: * Fixed compilation for DYNAMIC_ARCH with clang and with older gcc versions * Added support for compilation on FreeBSD/ppc64le * Added optimized POWER10 kernels for SSCAL, DSCAL, CSCAL, ZSCAL * Added optimized POWER10 kernels for SROT, DROT, CDOT, SASUM, DASUM * Improved SSWAP, DSWAP, CSWAP, ZSWAP performance on POWER10 * Improved SCOPY and CCOPY performance on POWER10 * Improved SGEMM and DGEMM performance on POWER10 * Added support for compilation with the NVIDIA HPC compiler x86_64: * Added an optimized bfloat16 GEMM kernel for Cooperlake * Added CPUID autodetection for Intel Rocket Lake and Tiger Lake cpus * Improved the performance of SASUM,DASUM,SROT,DROT on AMD Ryzen cpus * Added support for compilation with the NAG Fortran compiler * Fixed recognition of the AMD AOCC compiler * Fixed compilation for DYNAMIC_ARCH with clang on Windows * Added support for running the BLAS/CBLAS tests on Windows OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=120
2021-03-18 09:47:05 +01:00
Requires(preun):update-alternatives
%else
BuildRequires: %{compiler_family}%{?c_f_ver}-compilers-hpc-macros-devel
BuildRequires: lua-lmod
BuildRequires: suse-hpc
%global dep_summary %{summary}
%endif
%description
OpenBLAS is an optimized BLAS library based on GotoBLAS2 1.13 BSD version.
%package -n lib%{name}%{?so_v}
Summary: An optimized BLAS library based on GotoBLAS2, %{flavor} version
Group: System/Libraries
%if %{without hpc}
Requires(post): update-alternatives
- Update to version 0.3.14 common: * Fixed a race condition on thread shutdown in non-OpenMP builds * Fixed custom BUFFERSIZE option getting ignored in gmake builds * Fixed CMAKE compilation of the TRMM kernels for GENERIC platforms * Added CBLAS interfaces for CROTG, ZROTG, CSROT and ZDROT * Improved performance of OMATCOPY_RT across all platforms * Changed perl scripts to use env instead of a hardcoded /usr/bin/perl * Fixed potential misreading of the GCC compiler version in the build scripts * Fixed convergence problems in LAPACK complex GGEV/GGES (Reference-LAPACK #477) * Reduced the stacksize requirements for running the LAPACK testsuite (Reference-LAPACK #335) RISC V: * Fixed compilation on RISCV (missing entry in getarch) POWER: * Fixed compilation for DYNAMIC_ARCH with clang and with older gcc versions * Added support for compilation on FreeBSD/ppc64le * Added optimized POWER10 kernels for SSCAL, DSCAL, CSCAL, ZSCAL * Added optimized POWER10 kernels for SROT, DROT, CDOT, SASUM, DASUM * Improved SSWAP, DSWAP, CSWAP, ZSWAP performance on POWER10 * Improved SCOPY and CCOPY performance on POWER10 * Improved SGEMM and DGEMM performance on POWER10 * Added support for compilation with the NVIDIA HPC compiler x86_64: * Added an optimized bfloat16 GEMM kernel for Cooperlake * Added CPUID autodetection for Intel Rocket Lake and Tiger Lake cpus * Improved the performance of SASUM,DASUM,SROT,DROT on AMD Ryzen cpus * Added support for compilation with the NAG Fortran compiler * Fixed recognition of the AMD AOCC compiler * Fixed compilation for DYNAMIC_ARCH with clang on Windows * Added support for running the BLAS/CBLAS tests on Windows OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=120
2021-03-18 09:47:05 +01:00
Requires(preun):update-alternatives
%if "%flavor" == "serial"
Obsoletes: lib%{pname}%{so_v} < %{version}
Provides: lib%{pname}%{so_v} = %{version}
%else
Obsoletes: lib%{pname}0
%endif
%if "%flavor" == "pthreads"
Obsoletes: lib%{pname}p0
%endif
%if "%flavor" == "openmp"
Obsoletes: lib%{pname}o0
%endif
%else # with hpc
%hpc_requires
%endif
%description -n lib%{name}%{?so_v}
OpenBLAS is an optimized BLAS library based on GotoBLAS2 1.13 BSD version.
%{?with_hpc:%{hpc_master_package -l -L}}
%package -n lib%{name}-devel
Summary: Development libraries for OpenBLAS, %{flavor} version
Group: Development/Libraries/C and C++
Requires: lib%{name}%{?so_v} = %{version}
%if %{without hpc}
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
Requires: %{pname}-common-devel = %{version}
%if 0%{?arch_flavor}
Provides: %{pname}-devel = %version
Provides: %{pname}-devel(default) = %version
%else
Provides: %{pname}-devel(other) = %version
Accepting request 966356 from home:eeich:branches:science - Build PPC64LE libraries with the lastest gcc available to take advantage of instruction sets in later CPUs used in the CPU specific kernels (jsc#SLE-18143, bsc#1197721). For fortran use the stock compiler to avoid compatibility issues between different versions of libfortran. This is relevant for Leap/SLE only. It may be dropped once gcc < 10 is no longer supported. - Do the same for x86_64 on SLE to make sure Cooperlake support is built properly. - Remove: * Do-not-attempt-to-check-host-CPU-if-TARGET-is-set.patch * Create-independent-kernel-Makfile-configuration-when-building-DYNAMIC_ARCH.patch * For-DYNAMIC_ARCH-don-t-use-sbgemm_r-as-parameter.c-doesn-t-get-build.patch Instead, add from upstream: * Define-sbgemm_r-to-fix-DYNAMIC_ARCH-builds.patch * Remove-extraneous-and-wrong-definition-of-sbgemm_r-on-x86_64.patch * Fix-checks-for-AVX512-and-atomics.patch * Revert-AVX512-capability-check-from-PR-1980-moved-to-build.patch * Use-CC-and-full-command-line-instead-of-hard-coding-gcc-for-AVX512-checking.patch * Utilize-compiler-AVX512-capability-info-from-c_check-when-building-getarch.patch BLAS/LAPACK implementation (boo#1177260). 'openblas-common-devel' will provide the removed 'openblas-devel-headers' while the arch specific 'preferred' flavor will provide the removed 'openblas-devel'. OBS-URL: https://build.opensuse.org/request/show/966356 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=131
2022-04-01 09:59:26 +02:00
%endif
%else
%hpc_requires_devel
%endif
%description -n lib%{name}-devel
OpenBLAS is an optimized BLAS library based on GotoBLAS2 1.13 BSD version.
This package contains the development libraries for serial OpenBLAS version.
%{?with_hpc:%{hpc_master_package -l -L devel}}
%package devel-static
Summary: Static version of OpenBLAS
Group: Development/Libraries/C and C++
Requires: lib%{name}-devel = %{version}
%description devel-static
OpenBLAS is an optimized BLAS library based on GotoBLAS2 1.13 BSD version.
This package contains the static libraries.
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
%package -n %{pname}-common-devel
Summary: Development headers and libraries for OpenBLAS
Group: Development/Libraries/C and C++
Requires: (%{pname}-devel(default) or %{pname}-devel(other))
Obsoletes: %{pname}-devel < %version
Obsoletes: %{pname}-devel-headers < %version
Provides: %{pname}-devel-headers = %version
Provides: pkgconfig(openblas) = %version
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
%description -n %{pname}-common-devel
OpenBLAS is an optimized BLAS library based on GotoBLAS2 1.13 BSD version.
This package contains headers for OpenBLAS.
%prep
%setup -q -n OpenBLAS-%{version}
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
%autopatch -p1
%ifarch s390
sed -i -e "s@m32@m31@" Makefile.system
%endif
%if %{without hpc}
cp %{SOURCE1} .
%else
cp %{SOURCE2} .
%endif
%build
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
# Limit lto jobs to 1 - -flto=auto together with make -j<m>
# would cause a huge number of build jobs spawned in parallel
%if "%{?_lto_cflags}" != ""
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
%global _lto_cflags -flto=1 -ffat-lto-objects
%endif
# disable lto for ppc64le, boo#1181733
%ifarch ppc64le
%define _lto_cflags %{nil}
%endif
%if %{with hpc}
%hpc_debug
%hpc_setup_compiler
%endif
# Use DYNAMIC_ARCH everywhere - not sure about PPC?
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
# Use DYNAMIC_ARCH to build for multiple targets, use TARGET to specify
# the CPU model assumed for the common code. It should be set to the
# oldest CPU model one expects to encounter.
%global openblas_target DYNAMIC_ARCH=1
# We specify TARGET= to avoid compile-time CPU-detection (boo#1100677)
%ifarch %ix86 x86_64
%global openblas_target %openblas_target TARGET=CORE2
%define openblas_opt BUILD_BFLOAT16=1
%endif
%ifarch aarch64
- Update to version 0.3.14 common: * Fixed a race condition on thread shutdown in non-OpenMP builds * Fixed custom BUFFERSIZE option getting ignored in gmake builds * Fixed CMAKE compilation of the TRMM kernels for GENERIC platforms * Added CBLAS interfaces for CROTG, ZROTG, CSROT and ZDROT * Improved performance of OMATCOPY_RT across all platforms * Changed perl scripts to use env instead of a hardcoded /usr/bin/perl * Fixed potential misreading of the GCC compiler version in the build scripts * Fixed convergence problems in LAPACK complex GGEV/GGES (Reference-LAPACK #477) * Reduced the stacksize requirements for running the LAPACK testsuite (Reference-LAPACK #335) RISC V: * Fixed compilation on RISCV (missing entry in getarch) POWER: * Fixed compilation for DYNAMIC_ARCH with clang and with older gcc versions * Added support for compilation on FreeBSD/ppc64le * Added optimized POWER10 kernels for SSCAL, DSCAL, CSCAL, ZSCAL * Added optimized POWER10 kernels for SROT, DROT, CDOT, SASUM, DASUM * Improved SSWAP, DSWAP, CSWAP, ZSWAP performance on POWER10 * Improved SCOPY and CCOPY performance on POWER10 * Improved SGEMM and DGEMM performance on POWER10 * Added support for compilation with the NVIDIA HPC compiler x86_64: * Added an optimized bfloat16 GEMM kernel for Cooperlake * Added CPUID autodetection for Intel Rocket Lake and Tiger Lake cpus * Improved the performance of SASUM,DASUM,SROT,DROT on AMD Ryzen cpus * Added support for compilation with the NAG Fortran compiler * Fixed recognition of the AMD AOCC compiler * Fixed compilation for DYNAMIC_ARCH with clang on Windows * Added support for running the BLAS/CBLAS tests on Windows OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=120
2021-03-18 09:47:05 +01:00
%global openblas_target %openblas_target TARGET=ARMV8
%define openblas_opt BUILD_BFLOAT16=1
- Update to version 0.3.14 common: * Fixed a race condition on thread shutdown in non-OpenMP builds * Fixed custom BUFFERSIZE option getting ignored in gmake builds * Fixed CMAKE compilation of the TRMM kernels for GENERIC platforms * Added CBLAS interfaces for CROTG, ZROTG, CSROT and ZDROT * Improved performance of OMATCOPY_RT across all platforms * Changed perl scripts to use env instead of a hardcoded /usr/bin/perl * Fixed potential misreading of the GCC compiler version in the build scripts * Fixed convergence problems in LAPACK complex GGEV/GGES (Reference-LAPACK #477) * Reduced the stacksize requirements for running the LAPACK testsuite (Reference-LAPACK #335) RISC V: * Fixed compilation on RISCV (missing entry in getarch) POWER: * Fixed compilation for DYNAMIC_ARCH with clang and with older gcc versions * Added support for compilation on FreeBSD/ppc64le * Added optimized POWER10 kernels for SSCAL, DSCAL, CSCAL, ZSCAL * Added optimized POWER10 kernels for SROT, DROT, CDOT, SASUM, DASUM * Improved SSWAP, DSWAP, CSWAP, ZSWAP performance on POWER10 * Improved SCOPY and CCOPY performance on POWER10 * Improved SGEMM and DGEMM performance on POWER10 * Added support for compilation with the NVIDIA HPC compiler x86_64: * Added an optimized bfloat16 GEMM kernel for Cooperlake * Added CPUID autodetection for Intel Rocket Lake and Tiger Lake cpus * Improved the performance of SASUM,DASUM,SROT,DROT on AMD Ryzen cpus * Added support for compilation with the NAG Fortran compiler * Fixed recognition of the AMD AOCC compiler * Fixed compilation for DYNAMIC_ARCH with clang on Windows * Added support for running the BLAS/CBLAS tests on Windows OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=120
2021-03-18 09:47:05 +01:00
%endif
%ifarch s390 s390x
%global openblas_target %openblas_target TARGET=ZARCH_GENERIC
%endif
%ifarch ppc64le
%global openblas_target %openblas_target TARGET=POWER8
%define openblas_opt BUILD_BFLOAT16=1
%endif
%ifarch ppc64
%global openblas_target %openblas_target TARGET=POWER8
%endif
%ifarch riscv64
%global openblas_target %openblas_target TARGET=RISCV64_GENERIC
%endif
# force -mvsx for ppc64 to avoid build failure:
# ../kernel/power/sasum_microk_power8.c:41:3: error: '__vector' undeclared (first use in this function); did you mean '__cpow'?
# TODO why is it required ? (and not for ppc64le)
%ifarch ppc64
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
%global addopt -mvsx
%endif
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
%global addopt %{?addopt} -fno-strict-aliasing
# Make serial, threaded and OpenMP versions
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
# Calculate process limits
%limit_build -m 1500
[[ -n $_threads ]] && jobs=$_threads
[[ -z $jobs ]] && jobs=1
# NEVER use %%_smp_mflags with top level make:
# set MAKE_NB_JOBS instead and let the build do the work!
make MAKE_NB_JOBS=$jobs %{?openblas_target} %{?build_flags} \
%{?openblas_opt} \
COMMON_OPT="%{optflags} %{?addopt}" \
NUM_THREADS=%{num_threads} V=1 \
OPENBLAS_LIBRARY_DIR=%{p_libdir} \
OPENBLAS_INCLUDE_DIR=%{p_includedir} \
OPENBLAS_CMAKE_DIR=%{p_cmakedir} \
PREFIX=%{p_prefix} \
%{!?with_hpc:LIBNAMESUFFIX=%flavor FC=gfortran CC=gcc%{?cc_v:-%{cc_v}} %{?cc_v:CEXTRALIB=""}} \
%{?with_hpc:%{?cc_v:CC=gcc-%{cc_v} CEXTRALIB=""}}
%install
%if %{with hpc}
%hpc_setup_compiler
%endif
# Install serial library and headers
%make_install %{?build_flags} \
OPENBLAS_LIBRARY_DIR=%{p_libdir} \
OPENBLAS_INCLUDE_DIR=%{p_includedir} \
OPENBLAS_CMAKE_DIR=%{p_cmakedir} \
PREFIX=%{p_prefix} \
%{!?with_hpc:LIBNAMESUFFIX=%flavor}
# Delete info about OBS host cpu
%ifarch %ix86 x86_64
sed -i '/#define OPENBLAS_NEEDBUNDERSCORE/,/#define OPENBLAS_VERSION/{//!d}' \
%{buildroot}%{p_includedir}/openblas_config.h
%endif
%if %{without hpc}
%if 0%{!?build_devel:1}
# We need the includes only once
rm -rf %{buildroot}%{p_includedir}/
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
%endif
# Fix cmake config file
sed -i 's|%{buildroot}||g' %{buildroot}%{p_cmakedir}/*.cmake
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
sed -i 's|_%{flavor}||g' %{buildroot}%{p_cmakedir}/*.cmake
# Put libraries in correct location
rm -rf %{buildroot}%{p_libdir}/lib%{name}*
# Install the serial library
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
install -D -p -m 755 lib%{name}.so %{buildroot}%{p_libdir}/lib%{pname}.so.0
install -D -p -m 644 lib%{name}.a %{buildroot}%{p_libdir}/lib%{pname}.a
# Fix source permissions (also applies to LAPACK)
find -name \*.f -exec chmod 644 {} +
# Dummy target for update-alternatives
install -d %{buildroot}/%{_sysconfdir}/alternatives
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
ln -sf %{_sysconfdir}/alternatives/libblas.so.3 %{buildroot}/%{_libdir}/libblas.so.3
ln -sf %{_sysconfdir}/alternatives/libcblas.so.3 %{buildroot}/%{_libdir}/libcblas.so.3
ln -sf %{_sysconfdir}/alternatives/liblapack.so.3 %{buildroot}/%{_libdir}/liblapack.so.3
ln -sf %{_sysconfdir}/alternatives/liblapacke.so.3 %{buildroot}/%{_libdir}/liblapacke.so.3
ln -sf %{_sysconfdir}/alternatives/openblas-default %{buildroot}/%{_libdir}/openblas-default
ln -s lib%{pname}.so.%{so_v} %{buildroot}%{p_libdir}/lib%{pname}.so
ln -s %{_libdir}/openblas-default %{buildroot}%{_sysconfdir}/alternatives/openblas-default
Accepting request 966356 from home:eeich:branches:science - Build PPC64LE libraries with the lastest gcc available to take advantage of instruction sets in later CPUs used in the CPU specific kernels (jsc#SLE-18143, bsc#1197721). For fortran use the stock compiler to avoid compatibility issues between different versions of libfortran. This is relevant for Leap/SLE only. It may be dropped once gcc < 10 is no longer supported. - Do the same for x86_64 on SLE to make sure Cooperlake support is built properly. - Remove: * Do-not-attempt-to-check-host-CPU-if-TARGET-is-set.patch * Create-independent-kernel-Makfile-configuration-when-building-DYNAMIC_ARCH.patch * For-DYNAMIC_ARCH-don-t-use-sbgemm_r-as-parameter.c-doesn-t-get-build.patch Instead, add from upstream: * Define-sbgemm_r-to-fix-DYNAMIC_ARCH-builds.patch * Remove-extraneous-and-wrong-definition-of-sbgemm_r-on-x86_64.patch * Fix-checks-for-AVX512-and-atomics.patch * Revert-AVX512-capability-check-from-PR-1980-moved-to-build.patch * Use-CC-and-full-command-line-instead-of-hard-coding-gcc-for-AVX512-checking.patch * Utilize-compiler-AVX512-capability-info-from-c_check-when-building-getarch.patch BLAS/LAPACK implementation (boo#1177260). 'openblas-common-devel' will provide the removed 'openblas-devel-headers' while the arch specific 'preferred' flavor will provide the removed 'openblas-devel'. OBS-URL: https://build.opensuse.org/request/show/966356 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=131
2022-04-01 09:59:26 +02:00
ln -s %{_sysconfdir}/alternatives/openblas-default/lib%{pname}.so.%{so_v} %{buildroot}%{_libdir}/lib%{pname}.so.%{so_v}
%if 0%{?build_devel}
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
ln -s lib%{pname}.so.%{so_v} %{buildroot}%{_libdir}/lib%{pname}.so
install -d %{buildroot}%{_libdir}/pkgconfig/
Accepting request 966356 from home:eeich:branches:science - Build PPC64LE libraries with the lastest gcc available to take advantage of instruction sets in later CPUs used in the CPU specific kernels (jsc#SLE-18143, bsc#1197721). For fortran use the stock compiler to avoid compatibility issues between different versions of libfortran. This is relevant for Leap/SLE only. It may be dropped once gcc < 10 is no longer supported. - Do the same for x86_64 on SLE to make sure Cooperlake support is built properly. - Remove: * Do-not-attempt-to-check-host-CPU-if-TARGET-is-set.patch * Create-independent-kernel-Makfile-configuration-when-building-DYNAMIC_ARCH.patch * For-DYNAMIC_ARCH-don-t-use-sbgemm_r-as-parameter.c-doesn-t-get-build.patch Instead, add from upstream: * Define-sbgemm_r-to-fix-DYNAMIC_ARCH-builds.patch * Remove-extraneous-and-wrong-definition-of-sbgemm_r-on-x86_64.patch * Fix-checks-for-AVX512-and-atomics.patch * Revert-AVX512-capability-check-from-PR-1980-moved-to-build.patch * Use-CC-and-full-command-line-instead-of-hard-coding-gcc-for-AVX512-checking.patch * Utilize-compiler-AVX512-capability-info-from-c_check-when-building-getarch.patch BLAS/LAPACK implementation (boo#1177260). 'openblas-common-devel' will provide the removed 'openblas-devel-headers' while the arch specific 'preferred' flavor will provide the removed 'openblas-devel'. OBS-URL: https://build.opensuse.org/request/show/966356 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=131
2022-04-01 09:59:26 +02:00
ln -s %{_sysconfdir}/alternatives/openblas-default/pkgconfig/openblas.pc %{buildroot}%{_libdir}/pkgconfig/
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
install -d %{buildroot}/%{_libdir}/cmake
Accepting request 966356 from home:eeich:branches:science - Build PPC64LE libraries with the lastest gcc available to take advantage of instruction sets in later CPUs used in the CPU specific kernels (jsc#SLE-18143, bsc#1197721). For fortran use the stock compiler to avoid compatibility issues between different versions of libfortran. This is relevant for Leap/SLE only. It may be dropped once gcc < 10 is no longer supported. - Do the same for x86_64 on SLE to make sure Cooperlake support is built properly. - Remove: * Do-not-attempt-to-check-host-CPU-if-TARGET-is-set.patch * Create-independent-kernel-Makfile-configuration-when-building-DYNAMIC_ARCH.patch * For-DYNAMIC_ARCH-don-t-use-sbgemm_r-as-parameter.c-doesn-t-get-build.patch Instead, add from upstream: * Define-sbgemm_r-to-fix-DYNAMIC_ARCH-builds.patch * Remove-extraneous-and-wrong-definition-of-sbgemm_r-on-x86_64.patch * Fix-checks-for-AVX512-and-atomics.patch * Revert-AVX512-capability-check-from-PR-1980-moved-to-build.patch * Use-CC-and-full-command-line-instead-of-hard-coding-gcc-for-AVX512-checking.patch * Utilize-compiler-AVX512-capability-info-from-c_check-when-building-getarch.patch BLAS/LAPACK implementation (boo#1177260). 'openblas-common-devel' will provide the removed 'openblas-devel-headers' while the arch specific 'preferred' flavor will provide the removed 'openblas-devel'. OBS-URL: https://build.opensuse.org/request/show/966356 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=131
2022-04-01 09:59:26 +02:00
ln -s %{_sysconfdir}/alternatives/openblas-default/cmake/openblas %{buildroot}/%{_libdir}/cmake/
%endif
%else # with hpc
# HPC module file
%hpc_write_modules_files
#%%Module1.0#####################################################################
proc ModulesHelp { } {
puts stderr " "
puts stderr "This module loads the %{pname} library built with the %{compiler_family} compiler toolchain."
puts stderr "\nVersion %{version}\n"
}
module-whatis "Name: %{hpc_upcase %pname} built with %{compiler_family} toolchain"
module-whatis "Version: %{version}"
module-whatis "Category: runtime library"
module-whatis "Description: %{dep_summary}"
module-whatis "%{url}"
set version %{version}
prepend-path LD_LIBRARY_PATH %{p_libdir}
setenv %{hpc_upcase %pname}_DIR %{hpc_prefix}
if {[file isdirectory %{hpc_includedir}]} {
prepend-path LIBRARY_PATH %{p_libdir}
prepend-path CPATH %{p_includedir}
prepend-path C_INCLUDE_PATH %{p_includedir}
prepend-path CPLUS_INCLUDE_PATH %{p_includedir}
prepend-path INCLUDE %{p_includedir}
%hpc_modulefile_add_pkgconfig_path
setenv %{hpc_upcase %pname}_DIR %{hpc_prefix}
setenv %{hpc_upcase %pname}_LIB %{p_libdir}
setenv %{hpc_upcase %pname}_INC %{p_includedir}
}
family "openblas"
EOF
%{hpc_write_pkgconfig -l %{pname}}
%endif # with hpc
%if %{without hpc}
# Ensure directory used in older versions are replaced by symlink properly
%pre -n %{pname}-common-devel
d=%{_libdir}/cmake/openblas
[ -d $d -a ! -L $d -a "$(rpm -q --qf '%%{NAME}' -f $d 2>/dev/null)" = "openblas-devel" ] \
&& { n=$(mktemp -d $(dirname $d)/tmpd-XXXXX); mv $d $n; rm -rf $n; } || true
%post -n lib%{name}%{so_v}
%{_sbindir}/update-alternatives --install \
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
%{_libdir}/openblas-default openblas-default %{p_libdir} %openblas_so_prio
for lib in libblas.so.3 libcblas.so.3 liblapack.so.3 liblapacke.so.3; do
%{_sbindir}/update-alternatives --install \
%{_libdir}/${lib} ${lib} %{_libdir}/lib%{pname}.so.%{so_v} 20
done
/sbin/ldconfig
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
%postun -n lib%{name}%{so_v}
if [ ! -f %{p_libdir}/lib%{pname}.so.%{so_v} ]; then
for lib in libblas.so.3 libcblas.so.3 liblapack.so.3 liblapacke.so.3; do
%{_sbindir}/update-alternatives --remove ${lib} %{_libdir}/lib%{pname}.so.%{so_v}
done
fi
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
if [ ! -d %{p_libdir} ]; then
%{_sbindir}/update-alternatives --remove openblas-default %{p_libdir}
fi
/sbin/ldconfig
%posttrans -n lib%{name}%{so_v}
if [ "$1" = 0 ] ; then
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
if [ ! -d %{_libdir}/openblas-default ] ; then
%{_sbindir}/update-alternatives --auto openblas-default
fi
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
for lib in libblas.so.3 libcblas.so.3 liblapack.so.3 liblapacke.so.3; do
if ! [ -f %{_libdir}/${lib} ] ; then
%{_sbindir}/update-alternatives --auto ${lib}
fi
done
fi
%else
%postun -n lib%{name}
%hpc_module_delete_if_default
%endif
%files -n lib%{name}%{?so_v}
%defattr(-,root,root,-)
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
%{p_libdir}/lib%{pname}.so.0
%if %{without hpc}
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
%dir %{p_libdir}
%{_libdir}/openblas-default
%{_libdir}/lib%{pname}.so.%{so_v}
%ghost %{_libdir}/libblas.so.3
%ghost %{_libdir}/libcblas.so.3
%ghost %{_libdir}/liblapack.so.3
%ghost %{_libdir}/liblapacke.so.3
%ghost %{_sysconfdir}/alternatives/openblas-default
%ghost %{_sysconfdir}/alternatives/libblas.so.3
%ghost %{_sysconfdir}/alternatives/libcblas.so.3
%ghost %{_sysconfdir}/alternatives/liblapack.so.3
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
%ghost %{_sysconfdir}/alternatives/liblapacke.so.3
%else
%hpc_dirs
%{p_libdir}/libopenblas*r*.so
%hpc_modules_files
%endif
%files -n lib%{name}-devel
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
%{p_libdir}/lib%{pname}.so
%{p_cmakedir}/
%if %{with hpc}
%license LICENSE
%doc Changelog.txt GotoBLAS* README.md README.HPC.SUSE
%hpc_pkgconfig_file
%{p_includedir}/
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
%else
%dir %{p_libdir}/cmake
%dir %{p_libdir}/pkgconfig
%{p_libdir}/pkgconfig
%endif
%files devel-static
%{p_libdir}/libopenblas*.a
%if 0%{?build_devel}
%files -n %{pname}-common-devel
%license LICENSE
%doc Changelog.txt GotoBLAS* README.md README.SUSE
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
%{_libdir}/lib%{pname}.so
%{p_includedir}/
Accepting request 965613 from home:eeich:branches:science_alt - Update to v0.3.20: * general: some code cleanup, with added casts etc. fixed obtaining the cpu count with OpenMP and OMP_PROC_BIND unset fixed pivot index calculation by ?LASWP for negative increments other than one fixed input argument check in LAPACK ? GEQRT2 improved the check for a Fortran compiler in CMAKE builds disabled building OpenBLAS' optimized versions of LAPACK complex SPMV, SPR,SYMV,SYR with NO_LAPACK=1 fixed building of LAPACK on certain distributed filesystems with parallel gmake fixed building the shared library on MacOS with classic flang (v0.3.19) reverted unsafe TRSV/ZRSV optimizations introduced in 0.3.16 fixed a potential thread race in the thread buffer reallocation routines that were introduced in 0.3.18 fixed miscounting of thread pool size on Linux with OMP_PROC_BIND=TRUE fixed CBLAS interfaces for CSROT/ZSROT and CROTG/ZROTG made automatic library suffix for CMAKE builds with INTERFACE64 available to CBLAS-only builds (v0.3.18) when the build-time number of preconfigured threads is exceeded at runtime (by an external program calling BLAS functions from a larger number of threads), OpenBLAS will now allocate an auxiliary control structure for up to 512 additional threads instead of aborting added support for Loongson's LoongArch64 cpu architecture fixed building OpenBLAS with CMAKE and -DBUILD_BFLOAT16=ON added support for building OpenBLAS as a CMAKE subproject OBS-URL: https://build.opensuse.org/request/show/965613 OBS-URL: https://build.opensuse.org/package/show/science/openblas?expand=0&rev=129
2022-03-30 11:28:45 +02:00
%{_libdir}/pkgconfig/openblas.pc
%dir %{_libdir}/cmake
%{_libdir}/cmake/openblas
%endif
%changelog