Merge pull request #4408 from OpenMathLib/develop

merge develop for 0.3.26 release

Author: martin-frbg

.cirrus.yml

@@ -29,7 +29,7 @@ task:
   - mkdir build
   - cd build
   - cmake -DTARGET=VORTEX -DCMAKE_C_COMPILER=clang -DBUILD_SHARED_LIBS=ON ..
-  - make
+  - make -j 4
 
 task:
   name: AppleM1/GCC/MAKE/OPENMP

.github/workflows/loongarch64.yml

@@ -16,13 +16,13 @@ jobs:
         include:
           - target: LOONGSONGENERIC
             triple:  loongarch64-unknown-linux-gnu
-            opts: NO_SHARED=1 TARGET=LOONGSONGENERIC
+            opts: NO_SHARED=1 DYNAMIC_ARCH=1 TARGET=LOONGSONGENERIC
           - target: LOONGSON3R5
             triple: loongarch64-unknown-linux-gnu
-            opts: NO_SHARED=1 TARGET=LOONGSON3R5
+            opts: NO_SHARED=1 DYNAMIC_ARCH=1 TARGET=LOONGSON3R5
           - target: LOONGSON2K1000
             triple: loongarch64-unknown-linux-gnu
-            opts: NO_SHARED=1 TARGET=LOONGSON2K1000
+            opts: NO_SHARED=1 DYNAMIC_ARCH=1 TARGET=LOONGSON2K1000
           - target: DYNAMIC_ARCH
             triple: loongarch64-unknown-linux-gnu
             opts: NO_SHARED=1 DYNAMIC_ARCH=1 TARGET=GENERIC
@@ -40,8 +40,9 @@ jobs:
 
       - name: Download and install loongarch64-toolchain
         run: |
-          wget https://github.com/loongson/build-tools/releases/download/2022.09.06/loongarch64-clfs-7.3-cross-tools-gcc-glibc.tar.xz
-          tar -xf loongarch64-clfs-7.3-cross-tools-gcc-glibc.tar.xz -C /opt
+          wget https://github.com/sunhaiyong1978/CLFS-for-LoongArch/releases/download/8.1/CLFS-loongarch64-8.1-x86_64-cross-tools-gcc-glibc.tar.xz
+          #wget https://github.com/loongson/build-tools/releases/download/2023.08.08/CLFS-loongarch64-8.1-x86_64-cross-tools-gcc-glibc.tar.xz
+          tar -xf CLFS-loongarch64-8.1-x86_64-cross-tools-gcc-glibc.tar.xz -C /opt
 
       - name: Set env
         run: |

CMakeLists.txt

@@ -8,7 +8,7 @@ project(OpenBLAS C ASM)
 
 set(OpenBLAS_MAJOR_VERSION 0)
 set(OpenBLAS_MINOR_VERSION 3)
-set(OpenBLAS_PATCH_VERSION 25)
+set(OpenBLAS_PATCH_VERSION 25.dev)
 
 set(OpenBLAS_VERSION "${OpenBLAS_MAJOR_VERSION}.${OpenBLAS_MINOR_VERSION}.${OpenBLAS_PATCH_VERSION}")
 
@@ -249,20 +249,21 @@ if (${CMAKE_SYSTEM_NAME} MATCHES "AIX|Android|Linux|FreeBSD|OpenBSD|NetBSD|Drago
   endif()
 endif()
 
-if (APPLE AND DYNAMIC_ARCH AND BUILD_SHARED_LIBS)
+# Seems that this hack doesn't required since macOS 11 Big Sur
+if (APPLE AND BUILD_SHARED_LIBS AND CMAKE_HOST_SYSTEM_VERSION VERSION_LESS 20)
   set (CMAKE_C_USE_RESPONSE_FILE_FOR_OBJECTS 1)
   if (NOT NOFORTRAN)
   set (CMAKE_Fortran_USE_RESPONSE_FILE_FOR_OBJECTS 1)
   set (CMAKE_Fortran_CREATE_SHARED_LIBRARY
- "sh -c 'cat ${CMAKE_BINARY_DIR}/CMakeFiles/openblas_shared.dir/objects*.rsp | xargs -n 1024 ar -ru libopenblas.a && exit 0' "
- "sh -c 'ar -ru libopenblas.a ${CMAKE_BINARY_DIR}/driver/others/CMakeFiles/driver_others.dir/xerbla.c.o && exit 0' "
+ "sh -c 'cat ${CMAKE_BINARY_DIR}/CMakeFiles/openblas_shared.dir/objects*.rsp | xargs -n 1024 ${CMAKE_AR} -ru libopenblas.a && exit 0' "
+ "sh -c '${CMAKE_AR} -rs libopenblas.a ${CMAKE_BINARY_DIR}/driver/others/CMakeFiles/driver_others.dir/xerbla.c.o && exit 0' "
  "sh -c 'echo \"\" | ${CMAKE_Fortran_COMPILER} -o dummy.o -c -x f95-cpp-input - '"
  "sh -c '${CMAKE_Fortran_COMPILER} -fpic -shared -Wl,-all_load -Wl,-force_load,libopenblas.a -Wl,-noall_load dummy.o -o ${CMAKE_LIBRARY_OUTPUT_DIRECTORY}/libopenblas.${OpenBLAS_MAJOR_VERSION}.${OpenBLAS_MINOR_VERSION}.dylib'"
  "sh -c 'ls -l ${CMAKE_BINARY_DIR}/lib'")
   else ()
   set (CMAKE_C_CREATE_SHARED_LIBRARY
-   "sh -c 'cat ${CMAKE_BINARY_DIR}/CMakeFiles/openblas_shared.dir/objects*.rsp | xargs -n 1024 ar -ru libopenblas.a && exit 0' "
-   "sh -c 'ar -ru libopenblas.a ${CMAKE_BINARY_DIR}/driver/others/CMakeFiles/driver_others.dir/xerbla.c.o && exit 0' "
+   "sh -c 'cat ${CMAKE_BINARY_DIR}/CMakeFiles/openblas_shared.dir/objects*.rsp | xargs -n 1024 ${CMAKE_AR} -ru libopenblas.a && exit 0' "
+   "sh -c '${CMAKE_AR} -rs libopenblas.a ${CMAKE_BINARY_DIR}/driver/others/CMakeFiles/driver_others.dir/xerbla.c.o && exit 0' "
    "sh -c '${CMAKE_C_COMPILER} -fpic -shared -Wl,-all_load -Wl,-force_load,libopenblas.a -Wl,-noall_load -o ${CMAKE_LIBRARY_OUTPUT_DIRECTORY}/libopenblas.${OpenBLAS_MAJOR_VERSION}.${OpenBLAS_MINOR_VERSION}.dylib'")
   endif ()
 endif()
@@ -541,7 +542,7 @@ if(NOT NO_LAPACKE)
   ADD_CUSTOM_TARGET(genlapacke
   COMMAND ${CMAKE_COMMAND} -E copy ${CMAKE_CURRENT_SOURCE_DIR}/lapack-netlib/LAPACKE/include/lapacke_mangling_with_flags.h.in "${CMAKE_BINARY_DIR}/lapacke_mangling.h"
   )
-  install (FILES ${CMAKE_BINARY_DIR}/lapacke_mangling.h DESTINATION ${CMAKE_INSTALL_INCLUDEDIR}/openblas${SUFFIX64})
+  install (FILES ${CMAKE_BINARY_DIR}/lapacke_mangling.h DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
 endif()
 
 # Install pkg-config files

CONTRIBUTORS.md

@@ -216,3 +216,6 @@ In chronological order:
 
 * Pablo Romero <https://github.com/pablorcum>
   * [2022-08] Fix building from sources for QNX
+
+* Mark Seminatore <https://github.com/mseminatore>
+  * [2023-11-09] Improve Windows threading performance scaling

Changelog.txt

@@ -1,4 +1,49 @@
 OpenBLAS ChangeLog
+====================================================================
+Version 0.3.26
+ 2-Jan-2024
+
+general:
+- improved the version of openblas.pc that is created by the CMAKE build
+- fixed a CMAKE-specific build problem on older versions of MacOS
+- worked around linking problems on old versions of MacOS
+- corrected installation location of the lapacke_mangling header in CMAKE builds
+- added type declarations for complex variables to the MSVC-specific parts of the LAPACK header
+- significantly sped up ?GESV for small problem sizes by introducing a lower bound for multithreading
+- imported additions and corrections from the Reference-LAPACK project:
+  - added new LAPACK functions for truncated QR with pivoting (Reference-LAPACK PRs 891&941)
+  - handle miscalculation of minimum work array size in corner cases (Reference-LAPACK PR 942)
+  - fixed use of uninitialized variables in ?GEDMD and improved inline documentation (PR 959)
+  - fixed use of uninitialized variables (and consequential failures) in ?BBCSD (PR 967)
+  - added tests for the recently introduced Dynamic Mode Decomposition functions (PR 736)
+  - fixed several memory leaks in the LAPACK testsuite (PR 953)
+  - fixed counting of testsuite results by the Python script (PR 954)
+   
+x86-64:
+- fixed computation of CASUM on SkylakeX and newer targets in the special
+  case that AVX512 is not supported by the compiler or operating environment
+- fixed potential undefined behaviour in the CASUM/ZASUM kernels for AVX512 targets
+- worked around a problem in the pre-AVX kernels for GEMV
+- sped up the thread management code on MS Windows
+
+arm64:
+- fixed building of the LAPACK testsuite with Xcode 15 on Apple M1 and newer
+- sped up the thread management code on MS Windows
+- sped up SGEMM and DGEMM on Neoverse V1 and N1
+- sped up ?DOT on SVE-capable targets
+- reduced the number of targets in DYNAMIC_ARCH builds by eliminating functionally equivalent ones
+- included support for Apple M1 and newer targets in DYNAMIC_ARCH builds
+
+power:
+- improved the SGEMM kernel for POWER10
+- fixed compilation with (very) old versions of gcc
+- fixed detection of old 32bit PPC targets in CMAKE-based builds
+- added autodetection of the POWERPC 7400 subtype
+- fixed CMAKE-based compilation for PPCG4 and PPC970 targets
+
+loongarch64:
+- added and improved optimized kernels for almost all BLAS functions
+
 ====================================================================
 Version 0.3.25
  12-Nov-2023

GotoBLAS_06WeirdPerformance.txt

@@ -11,7 +11,7 @@
    operation is finished.
 
 
-2. Simlar problem may happen under virtual machine. If supervisor
+2. Similar problem may happen under virtual machine. If supervisor
    allocates different cores for each scheduling, BLAS performnace
    will be bad. This is because BLAS also utilizes all cache,
    unexpected re-schedule for different core may result of heavy

Makefile.power

@@ -11,7 +11,19 @@ endif
 
 ifeq ($(CORE), POWER10)
 ifneq ($(C_COMPILER), PGI)
+ifeq ($(C_COMPILER), GCC)
+ifeq ($(GCCVERSIONGTEQ10), 1)
 CCOMMON_OPT += -Ofast -mcpu=power10 -mtune=power10 -mvsx -fno-fast-math
+else ifneq ($(GCCVERSIONGT4), 1)
+$(warning your compiler is too old to fully support POWER9, getting a newer version of gcc is recommended)
+CCOMMON_OPT += -Ofast -mcpu=power8 -mtune=power8 -mvsx -fno-fast-math
+else
+$(warning your compiler is too old to fully support POWER10, getting a newer version of gcc is recommended)
+CCOMMON_OPT += -Ofast -mcpu=power9 -mtune=power9 -mvsx -fno-fast-math
+endif
+else
+CCOMMON_OPT += -Ofast -mcpu=power10 -mtune=power10 -mvsx -fno-fast-math
+endif
 ifeq ($(F_COMPILER), IBM)
 FCOMMON_OPT += -O2 -qrecur -qnosave -qarch=pwr10 -qtune=pwr10 -qfloat=nomaf -qzerosize
 else

Makefile.rule

@@ -3,7 +3,7 @@
 #
 
 # This library's version
-VERSION = 0.3.25
+VERSION = 0.3.25.dev
 
 # If you set the suffix, the library name will be libopenblas_$(LIBNAMESUFFIX).a
 # and libopenblas_$(LIBNAMESUFFIX).so. Meanwhile, the soname in shared library

Makefile.system

@@ -407,6 +407,7 @@ XCVER = $(shell pkgutil --pkg-info=com.apple.pkg.CLTools_Executables |awk '/vers
 endif
 ifeq (x$(XCVER), x 15)
 CCOMMON_OPT += -Wl,-ld_classic
+FCOMMON_OPT += -Wl,-ld_classic
 endif
 endif
 
@@ -676,16 +677,12 @@ ifeq ($(ARCH), arm64)
 DYNAMIC_CORE =  ARMV8
 DYNAMIC_CORE += CORTEXA53
 DYNAMIC_CORE += CORTEXA57
-DYNAMIC_CORE += CORTEXA72
-DYNAMIC_CORE += CORTEXA73
 DYNAMIC_CORE += NEOVERSEN1
 ifneq ($(NO_SVE), 1)
 DYNAMIC_CORE += NEOVERSEV1
 DYNAMIC_CORE += NEOVERSEN2
 DYNAMIC_CORE += ARMV8SVE
 endif
-DYNAMIC_CORE += CORTEXA55
-DYNAMIC_CORE += FALKOR
 DYNAMIC_CORE += THUNDERX
 DYNAMIC_CORE += THUNDERX2T99
 DYNAMIC_CORE += TSV110

README.md

@@ -196,20 +196,22 @@ Please read `GotoBLAS_01Readme.txt` for older CPU models already supported by th
   ```sh
   make HOSTCC=gcc TARGET=C910V CC=riscv64-unknown-linux-gnu-gcc FC=riscv64-unknown-linux-gnu-gfortran
   ```
-  (also known to work on C906)
+  (also known to work on C906 as long as you use only single-precision functions - its instruction set support appears to be incomplete in double precision)
 
 ### Support for multiple targets in a single library
 
 OpenBLAS can be built for multiple targets with runtime detection of the target cpu by specifiying `DYNAMIC_ARCH=1` in Makefile.rule, on the gmake command line or as `-DDYNAMIC_ARCH=TRUE` in cmake.
 
-For **x86_64**, the list of targets this activates contains Prescott, Core2, Nehalem, Barcelona, Sandybridge, Bulldozer, Piledriver, Steamroller, Excavator, Haswell, Zen, SkylakeX. For cpu generations not included in this list, the corresponding older model is used. If you also specify `DYNAMIC_OLDER=1`, specific support for Penryn, Dunnington, Opteron, Opteron/SSE3, Bobcat, Atom and Nano is added. Finally there is an option `DYNAMIC_LIST` that allows to specify an individual list of targets to include instead of the default.
+For **x86_64**, the list of targets this activates contains Prescott, Core2, Nehalem, Barcelona, Sandybridge, Bulldozer, Piledriver, Steamroller, Excavator, Haswell, Zen, SkylakeX, Cooper Lake, Sapphire Rapids. For cpu generations not included in this list, the corresponding older model is used. If you also specify `DYNAMIC_OLDER=1`, specific support for Penryn, Dunnington, Opteron, Opteron/SSE3, Bobcat, Atom and Nano is added. Finally there is an option `DYNAMIC_LIST` that allows to specify an individual list of targets to include instead of the default.
 
 `DYNAMIC_ARCH` is also supported on **x86**, where it translates to Katmai, Coppermine, Northwood, Prescott, Banias,
 Core2, Penryn, Dunnington, Nehalem, Athlon, Opteron, Opteron_SSE3, Barcelona, Bobcat, Atom and Nano.
 
-On **ARMV8**, it enables support for CortexA53, CortexA57, CortexA72, CortexA73, Falkor, ThunderX, ThunderX2T99, TSV110 as well as generic ARMV8 cpus.
+On **ARMV8**, it enables support for CortexA53, CortexA57, CortexA72, CortexA73, Falkor, ThunderX, ThunderX2T99, TSV110 as well as generic ARMV8 cpus. If compiler support for SVE is available at build time, support for NeoverseN2, NeoverseV1 as well as generic ArmV8SVE targets is also enabled.
 
-For **POWER**, the list encompasses POWER6, POWER8 and POWER9, on **ZARCH** it comprises Z13 and Z14.
+For **POWER**, the list encompasses POWER6, POWER8 and POWER9. POWER10 is additionally available if a sufficiently recent compiler is used for the build.
+
+on **ZARCH** it comprises Z13 and Z14 as well as generic zarch support.
 
 The `TARGET` option can be used in conjunction with `DYNAMIC_ARCH=1` to specify which cpu model should be assumed for all the
 common code in the library, usually you will want to set this to the oldest model you expect to encounter.