Merge branch 'develop' into benchmark-link-lib

Author: naoto-aoki-fy

.github/workflows/dynamic_arch.yml

@@ -593,7 +593,7 @@ jobs:
       - name: Build OpenBLAS
         run: |
           cd build
-          cmake --build . --target dgemm_thread_safety dgemm_thread_safety_mixed dgemv_thread_safety
+          cmake --build . --target dgemm_thread_safety dgemm_thread_safety_mixed dgemm_thread_safety_shutdown dgemv_thread_safety
 
       - name: Show ccache status
         continue-on-error: true
@@ -611,7 +611,7 @@ jobs:
         run: |
           cd build
           export PATH="$PWD/lib:$PATH"
-          OPENBLAS_NUM_THREADS=8 OMP_NUM_THREADS=16 ctest -R 'dgemm_thread_safety|dgemm_thread_safety_mixed|dgemv_thread_safety' --output-on-failure
+          OPENBLAS_NUM_THREADS=8 OMP_NUM_THREADS=16 ctest -R 'dgemm_thread_safety|dgemm_thread_safety_mixed|dgemm_thread_safety_shutdown|dgemv_thread_safety' --output-on-failure
 
 
   cross_build:

.github/workflows/loongarch64.yml

@@ -51,7 +51,7 @@ jobs:
       - name: Install APT deps
         run: |
           sudo apt-get update && \
-          sudo apt-get install autoconf automake autotools-dev ninja-build make ccache qemu-user-static \
+          sudo apt-get install autoconf automake autotools-dev ninja-build make ccache \
                                gcc-14-loongarch64-linux-gnu g++-14-loongarch64-linux-gnu gfortran-14-loongarch64-linux-gnu
 
       - name: Compilation cache
@@ -70,11 +70,18 @@ jobs:
           echo "compression = true" >> ~/.ccache/ccache.conf
           ccache -s
 
-      - name: Disable utest dsdot:dsdot_n_1
+      - name: Install QEMU 10.2.1
         run: |
-          echo -n > utest/test_dsdot.c
-          echo "Due to the current version of qemu causing utest cases to fail,"
-          echo "the utest dsdot:dsdot_n_1 have been temporarily disabled."
+          qemu_deb="${RUNNER_TEMP}/qemu-user_10.2.1+ds-1ubuntu3_amd64.deb"
+          qemu_root="${RUNNER_TEMP}/qemu-user-10.2.1"
+          wget --output-document="${qemu_deb}" \
+            https://archive.ubuntu.com/ubuntu/pool/universe/q/qemu/qemu-user_10.2.1+ds-1ubuntu3_amd64.deb
+          echo "e6bdaac070232a44b8657cad84ac0853d84b7167b55af82370af040291f55bd8  ${qemu_deb}" \
+            | sha256sum --check -
+          dpkg-deb --extract "${qemu_deb}" "${qemu_root}"
+          sudo install -m 0755 "${qemu_root}/usr/bin/qemu-loongarch64" \
+            /usr/local/bin/qemu-loongarch64-static
+          qemu-loongarch64-static --version
 
       - name: Build OpenBLAS
         run: |

.github/workflows/loongarch64_clang.yml

@@ -92,12 +92,6 @@ jobs:
           echo "compression = true" >> ~/.ccache/ccache.conf
           ccache -s
 
-      - name: Disable utest dsdot:dsdot_n_1
-        run: |
-          echo -n > utest/test_dsdot.c
-          echo "Due to the qemu versions 7.2 causing utest cases to fail,"
-          echo "the utest dsdot:dsdot_n_1 have been temporarily disabled."
-
       - name: Build OpenBLAS
         run: make CC='ccache clang --target=loongarch64-linux-gnu --sysroot=/opt/loongson-gnu-toolchain-8.3-x86_64-loongarch64-linux-gnu-rc1.3/loongarch64-linux-gnu/sysroot/ -static' FC='ccache loongarch64-linux-gnu-gfortran -static' HOSTCC='ccache clang' CROSS_SUFFIX=llvm-  NO_SHARED=1 ${{ matrix.opts }} -j$(nproc)
 

.github/workflows/mips64.yml

@@ -56,7 +56,7 @@ jobs:
         with:
           repository: qemu/qemu
           path: qemu
-          ref: ae35f033b874c627d81d51070187fbf55f0bf1a7
+          ref: 3e0bcba1ca7d6607ca49a988d165f052a3a53323
 
       - name: build qemu
         run: |

CONTRIBUTORS.md

@@ -283,4 +283,8 @@ hheei <hheei@users.noreply.github.com>
 * Aadityansha Verma <https://github.com/aadityansha06>
   * [2026-07-14] Add independent transpose support for C in GEADD (sgeadd/dgeadd/cgeadd/zgeadd).
 
+* Vincent Lovero <https://github.com/vlovero>
+  * [2026-08-11] ARM v9.2 SME GEMM kernels for Apple M
 
+* Hugo Meiland <hugo@meiland.nl>
+  * [2026-08-09] Add Cortex-A72 DGEMM 6x8 microkernel and blocking

Changelog.txt

@@ -158,6 +158,9 @@ x86_64:
 
 wasm:
  - added optimized kernels for STRSM and DTRSM
+ - added a 4x4 WASM SIMD128 GEMM microkernel for SGEMM/DGEMM with matching
+   4-wide packing and TRMM, using relaxed SIMD madd in the inner loop
+ - gated leftover WASM TRSM call counters behind OPENBLAS_WASM_TRSM_PROFILE
 
 ====================================================================
 Version 0.3.32

Makefile.arm64

@@ -59,7 +59,7 @@ endif
 endif
 
 ifeq ($(CORE), ARMV9SME)
-CCOMMON_OPT += -march=armv9-a+sve2+sme
+CCOMMON_OPT += -march=armv9-a+sve2+sme+sme-f64f64
 FCOMMON_OPT += -march=armv9-a+sve2
 ifdef OS_WINDOWS
 ifeq ($(C_COMPILER), CLANG)
@@ -310,7 +310,7 @@ endif
 
 ifeq ($(CORE), VORTEXM4)
 ifneq ($(C_COMPILER), GCC)
-CCOMMON_OPT += -march=armv8.4-a+sme
+CCOMMON_OPT += -march=armv8.4-a+sme+sme-f64f64
 #ifneq ($(APPLECLANG),1)
 #override LDFLAGS += -lclang_rt_builtins-aarch64
 #endif

Makefile.system

@@ -285,7 +285,7 @@ GEMM_GEMV_FORWARD = 1
 SBGEMM_GEMV_FORWARD = 1
 BGEMM_GEMV_FORWARD = 1
 endif
-ifeq ($(ARCH), riscv)
+ifeq ($(ARCH), riscv64)
 GEMM_GEMV_FORWARD = 1
 endif
 ifeq ($(ARCH), power)

Makefile.wasm

@@ -1 +1 @@
-CCOMMON_OPT += -msimd128
+CCOMMON_OPT += -msimd128 -mrelaxed-simd

azure-pipelines.yml

@@ -165,7 +165,7 @@ jobs:
 
 - job: OSX_OpenMP
   pool:
-     vmImage: 'macOS-14'
+     vmImage: 'macOS-15'
   steps:   
   - script: |
       brew update
@@ -175,7 +175,7 @@ jobs:
      
 - job: OSX_GCC_Nothreads
   pool:
-     vmImage: 'macOS-14'
+     vmImage: 'macOS-15'
   steps:   
   - script: |
       brew update
@@ -190,6 +190,7 @@ jobs:
       make CC=gcc-15 FC=gfortran-15
 
 - job: OSX_LLVM_flangnew
+  timeoutInMinutes: 100
   pool:
      vmImage: 'macOS-latest'
   variables:
@@ -203,6 +204,7 @@ jobs:
       make TARGET=NEHALEM CC=/usr/local/opt/llvm/bin/clang FC=/usr/local/opt/flang/bin/flang NO_SHARED=1
 
 - job: OSX_OpenMP_Clang
+  timeoutInMinutes: 100
   pool:
      vmImage: 'macOS-latest'
   variables:
@@ -216,6 +218,7 @@ jobs:
       make TARGET=CORE2 USE_OPENMP=1 DYNAMIC_ARCH=1 DYNAMIC_LIST='NEHALEM HASWELL SKYLAKEX' CC=/usr/local/opt/llvm/bin/clang NOFORTRAN=1
 
 - job: OSX_OpenMP_Clang_cmake
+  timeoutInMinutes: 100
   pool:
      vmImage: 'macOS-latest'
   variables:
@@ -233,8 +236,9 @@ jobs:
       ctest
       
 - job: OSX_dynarch_cmake
+  timeoutInMinutes: 100
   pool:
-     vmImage: 'macOS-14'
+     vmImage: 'macOS-15'
   variables:
      LD_LIBRARY_PATH: /usr/local/opt/llvm/lib
      LIBRARY_PATH: /usr/local/opt/llvm/lib
@@ -283,7 +287,7 @@ jobs:
  
 - job: OSX_NDK_ARMV7
   pool:
-     vmImage: 'macOS-14'
+     vmImage: 'macOS-15'
   steps:   
   - script: | 
       brew update
@@ -293,19 +297,19 @@ jobs:
 
 - job: OSX_IOS_ARMV8
   pool:
-     vmImage: 'macOS-14'
+     vmImage: 'macOS-15'
   variables:
-     CC: /Applications/Xcode_16.2.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/clang
-     CFLAGS: -O2 -Wno-macro-redefined -isysroot /Applications/Xcode_16.2.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS18.2.sdk -arch arm64 -miphoneos-version-min=10.0
+     CC: /Applications/Xcode_16.4.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/clang
+     CFLAGS: -O2 -Wno-macro-redefined -isysroot /Applications/Xcode_16.4.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS18.5.sdk -arch arm64 -miphoneos-version-min=10.0
   steps:
   - script: |
-     ls /Applications/Xcode_16.2.app/Contents/Developer/Platforms/
-     ls /Applications/Xcode_16.2.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs
+     ls /Applications/Xcode_16.4.app/Contents/Developer/Platforms/
+     ls /Applications/Xcode_16.4.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs
      make TARGET=ARMV8 DYNAMIC_ARCH=1 NUM_THREADS=32 HOSTCC=clang NOFORTRAN=1
 
 - job: OSX_IOS_ARMV7
   pool:
-     vmImage: 'macOS-14'
+     vmImage: 'macOS-15'
   variables:
      CC: /Applications/Xcode_16.2.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/clang
      CFLAGS: -O2 -mno-thumb -Wno-macro-redefined -isysroot /Applications/Xcode_16.2.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS18.2.sdk -arch armv7 -miphoneos-version-min=5.1
@@ -315,15 +319,15 @@ jobs:
 
 - job: OSX_xbuild_DYNAMIC_ARM64
   pool:
-     vmImage: 'macOS-14'
+     vmImage: 'macOS-15'
   variables:
-     CC: /Applications/Xcode_16.2.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/clang
-     CFLAGS: -O2 -Wno-macro-redefined -isysroot /Applications/Xcode_16.2.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX15.2.sdk -arch arm64
+     CC: /Applications/Xcode_16.4.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/clang
+     CFLAGS: -O2 -Wno-macro-redefined -isysroot /Applications/Xcode_16.4.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX15.5.sdk -arch arm64
   steps:
   - script: |
-     ls /Applications/Xcode_16.2.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs
-     /Applications/Xcode_12.2.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/clang -arch arm64 --print-supported-cpus
-     /Applications/Xcode_16.2.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/clang --version
+     ls /Applications/Xcode_16.4.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs
+     /Applications/Xcode_16.4.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/clang -arch arm64 --print-supported-cpus
+     /Applications/Xcode_16.4.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/clang --version
      make TARGET=ARMV8 DYNAMIC_ARCH=1 NUM_THREADS=32 HOSTCC=clang NOFORTRAN=1
 
 - job: ALPINE_MUSL