gemv.c:229:31: error: 'GEMM_MULTITHREAD_THRESHOLD' undeclared (first use in this function)
229 | if ( 1L * m * n < 115200L * GEMM_MULTITHREAD_THRESHOLD )
On some amd64 CPUs, this change broke py-scipy during configure,
and octave at runtime. In the end, octave kills the package
builders.
This reverts commit c45681ef50.
Approved by: portmgr (antoine)
PR: 273219
MFH: 2023Q4
Commit b7f05445c0 has added WWW entries to port Makefiles based on
WWW: lines in pkg-descr files.
This commit removes the WWW: lines of moved-over URLs from these
pkg-descr files.
Approved by: portmgr (tcberner)
It has been common practice to have one or more URLs at the end of the
ports' pkg-descr files, one per line and prefixed with "WWW:". These
URLs should point at a project website or other relevant resources.
Access to these URLs required processing of the pkg-descr files, and
they have often become stale over time. If more than one such URL was
present in a pkg-descr file, only the first one was tarnsfered into
the port INDEX, but for many ports only the last line did contain the
port specific URL to further information.
There have been several proposals to make a project URL available as
a macro in the ports' Makefiles, over time.
This commit implements such a proposal and moves one of the WWW: entries
of each pkg-descr file into the respective port's Makefile. A heuristic
attempts to identify the most relevant URL in case there is more than
one WWW: entry in some pkg-descr file. URLs that are not moved into the
Makefile are prefixed with "See also:" instead of "WWW:" in the pkg-descr
files in order to preserve them.
There are 1256 ports that had no WWW: entries in pkg-descr files. These
ports will not be touched in this commit.
The portlint port has been adjusted to expect a WWW entry in each port
Makefile, and to flag any remaining "WWW:" lines in pkg-descr files as
deprecated.
Approved by: portmgr (tcberner)
- Upgrade math/blas, math/cblas, math/lapack, math/lapacke and math/xlapack
to 3.9.1;
Latest release notes at <http://www.netlib.org/lapack/lapack-3.9.1.html>
- Chase this upgrade in biology/treekin;
- Add a test target;
- Remove a conflict with math/openblas (PR 244296);
- Fix the build with Gcc10 (PR 247485).
PR: 247542
Approved by: expr-run by antoine@
OpenMP is not available on these arches:
In file included from ../common.h:847:
../common_thread.h:43:10: fatal error: 'omp.h' file not found
Approved by: portmgr (build fix blanket)
The removed check was simply a safety belt. Improve option's description to underline that turning it off semantically equals to building with march=native.
PR: 245926
Approved by: Eijiro Shibusawa <phd_kimberlite@yahoo.co.jp> (maintainer)
Changelog:
common:
Improved thread locking behaviour in blas_server and parallel getrf
Imported bugfix 394 from LAPACK (spurious reference to "XERBL"
due to overlong lines)
Imported bugfix 403 from LAPACK (compile option "recursive" required
for correctness with Intel and PGI)
Imported bugfix 408 from LAPACK (wrong scaling in ZHEEQUB)
Imported bugfix 411 from LAPACK (infinite loop in LARGV/LARTG/LARTGP)
Fixed mismatches between BUFFERSIZE and GEMM_UNROLL parameters that
could lead to crashes at large matrix sizes
Restored internal soname in dynamic libraries on FreeBSD and Dragonfly
Added API (openblas_setaffinity) to set thread affinity
programmatically on Linux
Added initial infrastructure for half-precision floating point
(bfloat16) support with a generic implementation of SHGEMM
Added CMAKE build system support for building the cblas_Xgemm3m
functions
Fixed CMAKE support for building in a path with embedded spaces
Fixed CMAKE (non)handling of NO_EXPRECISION and MAX_STACK_ALLOC
Fixed GCC version detection in the Makefiles
Allowed overriding the names of AR, AS and LD in Makefile builds
POWER:
fixed big-endian POWER8 ELFv2 builds on FreeBSD
Fixed GCC version checks and DYNAMIC_ARCH builds on POWER9
Fixed CMAKE build support for POWER9
fixed a potential race condition in the thread buffer allocation
Worked around LAPACK test failures on PPC G4
MIPS:
fixed a potential race condition in the thread buffer allocation
Added support for MIPS 24K/24KE family based on P5600 kernels
MIPS64:
fixed a potential race condition in the thread buffer allocation
Added TARGET=GENERIC
ARMV7:
fixed a race condition in the thread buffer allocation
ARMV8:
Fixed a race condition in the thread buffer allocation
Fixed zero initialisation in the assembly for SGEMM and DGEMM BETA
Improved performance of the ThunderX2 DAXPY kernel
Added an optimized SGEMM kernel for Cortex A53
Fixed Makefile support for INTERFACE64 (8-byte integer)
x86_64:
Fixed a syntax error in the CMAKE setup for SkylakeX
Improved performance of STRSM on Haswell, SkylakeX and Ryzen
Improved SGEMM performance on SGEMM for workloads with ldc a
multiple of 1024
Improved DGEMM performance on Skylake X
Fixed unwanted AVX512-dependency of SGEMM in DYNAMIC_ARCH
builds created on SkylakeX
Removed data alignment requirement in the SSE2 copy kernels
that could cause spurious crashes
Added a workaround for an optimizer bug in AppleClang 11.0.3
Fixed LAPACK-TEST failures with Intel Fortran
Fixed compilation and LAPACK test results with recent Flang
and AMD AOCC
Fixed DYNAMIC_ARCH builds with CMAKE on OS X
Fixed missing exports of cblas_i?amin, cblas_i?min, cblas_i?max,
cblas_?sum, cblas_?gemm3m in the shared library on OS X
Fixed reporting of cpu name in DYNAMIC_ARCH builds (would sometimes
show the name of an older generation chip supported by the same kernels)
IBM Z:
Improved performance of SGEMM/STRMM and DGEMM/DTRMM on Z14
PR: 249120
Approved by: phd_kimberlite@yahoo.co.jp (maintainer)
Merge upstream commit to fix build on powerpc.
Use GCC from ports to build it.
Since this is mostly about G3 and G4 machines that users have and OpenBLAS supports only G4 of those, target G4.
Due to some (still unknown) regression on head, this builds on 12.1, but fails tests on head.
Change the default to OPENMP=ON
For example, erkale (science/erkale) was breaking:
OpenBLAS Warning : Detect OpenMP Loop and this application may hang. Please rebuild the library with USE_OPENMP=1 option.
Approved by: portmgr (unbreak)
On powerpc64 head, libopenblas_ppc970p-r0.3.7.* weren't added to the package, even though the package itself built succesfully. This problem doesn't exist on 12.
Lack of libopenblas_ppc970p-r0.3.7.* made many packages fail.
Add DYNAMIC_ARCH option everywhere since it will cause the post-install target dependent on this option to run and include libopenblas_ppc970p-r0.3.7.* in the package.
This is only to get the necessary libs in the package, the option only works on ppc64le, which we don't support.
Bump PORTREVISION because of package change.
This is further fallout from r523749 and r524642
While here, fix order of PORTREVISION. There is no PORTREVISION bump
because this a build fix.
PR: 231371
PR: 243739
Reported by: many
* Fix build on i386 [1]
* Fix science/code_saturne build with new openblas [2]
* Avoid installing private headers [3]
* Prevent build from optimizing for host by correcting build confg [4]
* Bump portrevision of dependent ports [5]
This is correcting issues from r523749 [1][2][4] and r515970 [3]
PR: 231371
Reported by: build cluster [1]
Reported by: Dima Pasechnik <dimpase+freebsd@gmail.com> [2]
Reported by: many [5]
Reviewed by: mat, bapt
Approved by: implicit, since this is a build fix
