Merge branch 'develop' into release-0.3.0

Author: martin-frbg

.github/workflows/dynamic_arch.yml

@@ -2,6 +2,9 @@ name: continuous build
 
 on: [push, pull_request]
 
+permissions:
+  contents: read # to fetch code (actions/checkout)
+
 jobs:
   build:
     runs-on: ${{ matrix.os }}
@@ -34,7 +37,7 @@ jobs:
       - name: Install Dependencies
         run: |
           if [ "$RUNNER_OS" == "Linux" ]; then
-            sudo apt-get install -y gfortran cmake ccache
+            sudo apt-get install -y gfortran cmake ccache libtinfo5
           elif [ "$RUNNER_OS" == "macOS" ]; then
             # It looks like "gfortran" isn't working correctly unless "gcc" is re-installed.
             brew reinstall gcc
@@ -150,6 +153,7 @@ jobs:
       matrix:
         msystem: [MINGW64, MINGW32, CLANG64]
         idx: [int32, int64]
+        build-type: [Release]
         include:
           - msystem: MINGW64
             idx: int32
@@ -173,6 +177,11 @@ jobs:
             idx64-flags: -DBINARY=64 -DINTERFACE64=1
             target-prefix: mingw-w64-clang-x86_64
             c-lapack-flags: -DC_LAPACK=ON
+          - msystem: MINGW64
+            idx: int32
+            target-prefix: mingw-w64-x86_64
+            fc-pkg: mingw-w64-x86_64-gcc-fortran
+            build-type: None
         exclude:
           - msystem: MINGW32
             idx: int64
@@ -215,11 +224,11 @@ jobs:
           path: C:/msys64/home/runneradmin/.ccache
           # We include the commit sha in the cache key, as new cache entries are
           # only created if there is no existing entry for the key yet.
-          key: ccache-msys2-${{ matrix.msystem }}-${{ matrix.idx }}-${{ github.ref }}-${{ github.sha }}
+          key: ccache-msys2-${{ matrix.msystem }}-${{ matrix.idx }}-${{ matrix.build-type }}-${{ github.ref }}-${{ github.sha }}
           # Restore a matching ccache cache entry. Prefer same branch.
           restore-keys: |
-            ccache-msys2-${{ matrix.msystem }}-${{ matrix.idx }}-${{ github.ref }}
-            ccache-msys2-${{ matrix.msystem }}-${{ matrix.idx }}
+            ccache-msys2-${{ matrix.msystem }}-${{ matrix.idx }}-${{ matrix.build-type }}-${{ github.ref }}
+            ccache-msys2-${{ matrix.msystem }}-${{ matrix.idx }}-${{ matrix.build-type }}
 
       - name: Configure ccache
         # Limit the maximum size and switch on compression to avoid exceeding the total disk or cache quota.
@@ -235,7 +244,8 @@ jobs:
       - name: Configure OpenBLAS
         run: |
           mkdir build && cd build
-          cmake -DBUILD_SHARED_LIBS=ON \
+          cmake -DCMAKE_BUILD_TYPE=${{ matrix.build-type }} \
+                -DBUILD_SHARED_LIBS=ON \
                 -DBUILD_STATIC_LIBS=ON \
                 -DDYNAMIC_ARCH=ON \
                 -DUSE_THREAD=ON \
@@ -257,3 +267,54 @@ jobs:
       - name: Run tests
         timeout-minutes: 60
         run: cd build && ctest
+
+
+  cross_build:
+    runs-on: ubuntu-22.04
+
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - target: mips64el
+            triple: mips64el-linux-gnuabi64
+            opts: DYNAMIC_ARCH=1 TARGET=GENERIC
+          - target: riscv64
+            triple: riscv64-linux-gnu
+            opts: TARGET=RISCV64_GENERIC
+          - target: mipsel
+            triple: mipsel-linux-gnu
+            opts: TARGET=MIPS1004K
+          - target: alpha
+            triple: alpha-linux-gnu
+            opts: TARGET=EV4
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v3
+
+      - name: Install Dependencies
+        run: |
+          sudo apt-get install -y ccache gcc-${{ matrix.triple }} gfortran-${{ matrix.triple }} libgomp1-${{ matrix.target }}-cross
+
+      - name: Compilation cache
+        uses: actions/cache@v3
+        with:
+          path: ~/.ccache
+          key: ccache-${{ runner.os }}-${{ matrix.target }}-${{ github.ref }}-${{ github.sha }}
+          restore-keys: |
+            ccache-${{ runner.os }}-${{ matrix.target }}-${{ github.ref }}
+            ccache-${{ runner.os }}-${{ matrix.target }}
+
+      - name: Configure ccache
+        run: |
+          # Limit the maximum size and switch on compression to avoid exceeding the total disk or cache quota (5 GB).
+          test -d ~/.ccache || mkdir -p ~/.ccache
+          echo "max_size = 300M" > ~/.ccache/ccache.conf
+          echo "compression = true" >> ~/.ccache/ccache.conf
+          ccache -s
+
+
+      - name: Build OpenBLAS
+        run: |
+          make -j$(nproc) HOSTCC="ccache gcc" CC="ccache ${{ matrix.triple }}-gcc" FC="ccache ${{ matrix.triple }}-gfortran" ARCH=${{ matrix.target }} ${{ matrix.opts }}

.github/workflows/mips64.yml

@@ -0,0 +1,117 @@
+name: mips64 qemu test
+
+on: [push, pull_request]
+
+permissions:
+  contents: read # to fetch code (actions/checkout)
+
+jobs:
+  TEST:
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - target: MIPS64_GENERIC
+            triple: mips64el-linux-gnuabi64
+            opts: NO_SHARED=1 TARGET=MIPS64_GENERIC
+          - target: SICORTEX
+            triple: mips64el-linux-gnuabi64
+            opts: NO_SHARED=1 TARGET=SICORTEX
+          - target: I6400
+            triple: mipsisa64r6el-linux-gnuabi64
+            opts: NO_SHARED=1 TARGET=I6400
+          - target: P6600
+            triple: mipsisa64r6el-linux-gnuabi64
+            opts: NO_SHARED=1 TARGET=P6600
+          - target: I6500
+            triple: mipsisa64r6el-linux-gnuabi64
+            opts: NO_SHARED=1 TARGET=I6500
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v3
+
+      - name: install build deps
+        run: |
+          sudo apt-get update
+          sudo apt-get install autoconf automake autotools-dev ninja-build make ccache \
+          gcc-${{ matrix.triple }} gfortran-${{ matrix.triple }} libgomp1-mips64el-cross
+
+      - name: checkout qemu
+        uses: actions/checkout@v3
+        with:
+          repository: qemu/qemu
+          path: qemu
+          ref: 79dfa177ae348bb5ab5f97c0915359b13d6186e2
+
+      - name: build qemu
+        run: |
+          cd qemu
+          ./configure --prefix=$GITHUB_WORKSPACE/qemu-install --target-list=mips64el-linux-user --disable-system
+          make -j$(nproc)
+          make install
+
+      - name: Compilation cache
+        uses: actions/cache@v3
+        with:
+          path: ~/.ccache
+          key: ccache-${{ runner.os }}-${{ matrix.target }}-${{ github.ref }}-${{ github.sha }}
+          restore-keys: |
+            ccache-${{ runner.os }}-${{ matrix.target }}-${{ github.ref }}
+            ccache-${{ runner.os }}-${{ matrix.target }}
+
+      - name: Configure ccache
+        run: |
+          test -d ~/.ccache || mkdir -p ~/.ccache
+          echo "max_size = 300M" > ~/.ccache/ccache.conf
+          echo "compression = true" >> ~/.ccache/ccache.conf
+          ccache -s
+
+      - name: build OpenBLAS
+        run: make CC='ccache ${{ matrix.triple }}-gcc -static' FC='ccache ${{ matrix.triple }}-gfortran -static' ${{ matrix.opts }} HOSTCC='ccache gcc' -j$(nproc)
+
+      - name: test
+        run: |
+          export PATH=$GITHUB_WORKSPACE/qemu-install/bin/:$PATH
+          qemu-mips64el ./utest/openblas_utest
+          OPENBLAS_NUM_THREADS=2 qemu-mips64el ./ctest/xscblat1
+          OPENBLAS_NUM_THREADS=2 qemu-mips64el ./ctest/xdcblat1
+          OPENBLAS_NUM_THREADS=2 qemu-mips64el ./ctest/xccblat1
+          OPENBLAS_NUM_THREADS=2 qemu-mips64el ./ctest/xzcblat1
+          OPENBLAS_NUM_THREADS=2 qemu-mips64el ./ctest/xscblat2 < ./ctest/sin2
+          OPENBLAS_NUM_THREADS=2 qemu-mips64el ./ctest/xdcblat2 < ./ctest/din2
+          OPENBLAS_NUM_THREADS=2 qemu-mips64el ./ctest/xccblat2 < ./ctest/cin2
+          OPENBLAS_NUM_THREADS=2 qemu-mips64el ./ctest/xzcblat2 < ./ctest/zin2
+          OPENBLAS_NUM_THREADS=2 qemu-mips64el ./ctest/xscblat3 < ./ctest/sin3
+          OPENBLAS_NUM_THREADS=2 qemu-mips64el ./ctest/xdcblat3 < ./ctest/din3
+          OPENBLAS_NUM_THREADS=2 qemu-mips64el ./ctest/xccblat3 < ./ctest/cin3
+          OPENBLAS_NUM_THREADS=2 qemu-mips64el ./ctest/xzcblat3 < ./ctest/zin3
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-mips64el ./test/sblat1
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-mips64el ./test/dblat1
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-mips64el ./test/cblat1
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-mips64el ./test/zblat1
+          OPENBLAS_NUM_THREADS=2 qemu-mips64el ./test/sblat1
+          OPENBLAS_NUM_THREADS=2 qemu-mips64el ./test/dblat1
+          OPENBLAS_NUM_THREADS=2 qemu-mips64el ./test/cblat1
+          OPENBLAS_NUM_THREADS=2 qemu-mips64el ./test/zblat1
+          rm -f ./test/?BLAT2.SUMM
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-mips64el ./test/sblat2 < ./test/sblat2.dat
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-mips64el ./test/dblat2 < ./test/dblat2.dat
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-mips64el ./test/cblat2 < ./test/cblat2.dat
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-mips64el ./test/zblat2 < ./test/zblat2.dat
+          rm -f ./test/?BLAT2.SUMM
+          OPENBLAS_NUM_THREADS=2 qemu-mips64el ./test/sblat2 < ./test/sblat2.dat
+          OPENBLAS_NUM_THREADS=2 qemu-mips64el ./test/dblat2 < ./test/dblat2.dat
+          OPENBLAS_NUM_THREADS=2 qemu-mips64el ./test/cblat2 < ./test/cblat2.dat
+          OPENBLAS_NUM_THREADS=2 qemu-mips64el ./test/zblat2 < ./test/zblat2.dat
+          rm -f ./test/?BLAT3.SUMM
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-mips64el ./test/sblat3 < ./test/sblat3.dat
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-mips64el ./test/dblat3 < ./test/dblat3.dat
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-mips64el ./test/cblat3 < ./test/cblat3.dat
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-mips64el ./test/zblat3 < ./test/zblat3.dat
+          rm -f ./test/?BLAT3.SUMM
+          OPENBLAS_NUM_THREADS=2 qemu-mips64el ./test/sblat3 < ./test/sblat3.dat
+          OPENBLAS_NUM_THREADS=2 qemu-mips64el ./test/dblat3 < ./test/dblat3.dat
+          OPENBLAS_NUM_THREADS=2 qemu-mips64el ./test/cblat3 < ./test/cblat3.dat
+          OPENBLAS_NUM_THREADS=2 qemu-mips64el ./test/zblat3 < ./test/zblat3.dat

.github/workflows/nightly-Homebrew-build.yml

@@ -17,6 +17,10 @@ on:
 # it only makes sense to test if this file has been changed
 
 name: Nightly-Homebrew-Build
+
+permissions:
+  contents: read # to fetch code (actions/checkout)
+
 jobs:
   build-OpenBLAS-with-Homebrew:
     runs-on: macos-latest
@@ -28,6 +32,8 @@ jobs:
       HOMEBREW_NO_AUTO_UPDATE: "ON"
       HOMEBREW_NO_BOTTLE_SOURCE_FALLBACK: "ON"
       HOMEBREW_NO_INSTALL_CLEANUP: "ON"
+      HOMEBREW_NO_INSTALLED_DEPENDENTS_CHECK: "ON"
+      HOMEBREW_NO_INSTALL_FROM_API: "ON"      
 
     steps:
       - name: Random delay for cron job

.travis.yml

@@ -30,7 +30,7 @@ matrix:
       before_script: &common-before
         - COMMON_FLAGS="DYNAMIC_ARCH=1 TARGET=POWER8 NUM_THREADS=32"
       script:
-        - travis_wait 20 make QUIET_MAKE=1 $COMMON_FLAGS $BTYPE
+        - travis_wait 50 make QUIET_MAKE=1 $COMMON_FLAGS $BTYPE
         - make -C test $COMMON_FLAGS $BTYPE
         - make -C ctest $COMMON_FLAGS $BTYPE
         - make -C utest $COMMON_FLAGS $BTYPE
@@ -104,7 +104,7 @@ matrix:
         - sudo apt-get update
         - sudo apt-get install gcc-9 gfortran-9 -y
       script:
-        - travis_wait 20 make QUIET_MAKE=1  BINARY=64 USE_OPENMP=1 CC=gcc-9 FC=gfortran-9
+        - travis_wait 50 make QUIET_MAKE=1  BINARY=64 USE_OPENMP=1 CC=gcc-9 FC=gfortran-9
         - make -C test $COMMON_FLAGS $BTYPE
         - make -C ctest $COMMON_FLAGS $BTYPE
         - make -C utest $COMMON_FLAGS $BTYPE 
@@ -121,7 +121,7 @@ matrix:
         - sudo apt-get update
         - sudo apt-get install gcc-9 gfortran-9 -y
       script:
-        - travis_wait 20 make QUIET_MAKE=1  BUILD_BFLOAT16=1 BINARY=64 USE_OPENMP=1 CC=gcc-9 FC=gfortran-9
+        - travis_wait 50 make QUIET_MAKE=1  BUILD_BFLOAT16=1 BINARY=64 USE_OPENMP=1 CC=gcc-9 FC=gfortran-9
         - make -C test $COMMON_FLAGS $BTYPE
         - make -C ctest $COMMON_FLAGS $BTYPE
         - make -C utest $COMMON_FLAGS $BTYPE 
@@ -285,6 +285,25 @@ matrix:
             - gfortran
       script:
         - travis_wait 45 make && make lapack-test
+      env:
+        - TARGET_BOX=NEOVERSE_N1 
+  
+    - &test-neon1-gcc8
+      os: linux
+      arch: arm64
+      dist: focal
+      group: edge
+      virt: lxd
+      compiler: gcc
+      addons:
+        apt:
+          packages:
+            - gcc-8
+            - gfortran-8
+      script:
+        - travis_wait 45 make QUIET_MAKE=1 CC=gcc-8 FC=gfortran-8 DYNAMIC_ARCH=1 
+      env:
+        - TARGET_BOX=NEOVERSE_N1-GCC8
 
 # whitelist
 branches:

CMakeLists.txt

@@ -8,7 +8,7 @@ project(OpenBLAS C ASM)
 
 set(OpenBLAS_MAJOR_VERSION 0)
 set(OpenBLAS_MINOR_VERSION 3)
-set(OpenBLAS_PATCH_VERSION 21)
+set(OpenBLAS_PATCH_VERSION 22)
 
 set(OpenBLAS_VERSION "${OpenBLAS_MAJOR_VERSION}.${OpenBLAS_MINOR_VERSION}.${OpenBLAS_PATCH_VERSION}")
 
@@ -36,6 +36,8 @@ option(USE_LOCKING "Use locks even in single-threaded builds to make them callab
 
 option(USE_PERL "Use the older PERL scripts for build preparation instead of universal shell scripts" OFF)
 
+option(NO_WARMUP "Do not run a benchmark on each startup just to find the best location for the memory buffer" ON)
+
 if(${CMAKE_SYSTEM_NAME} MATCHES "Linux")
   option(NO_AFFINITY "Disable support for CPU affinity masks to avoid binding processes from e.g. R or numpy/scipy to a single core" ON)
 else()
@@ -212,10 +214,10 @@ if(NOT NO_LAPACKE)
   add_library(LAPACKE OBJECT ${LAPACKE_SOURCES})
   list(APPEND TARGET_OBJS "$<TARGET_OBJECTS:LAPACKE>")
 endif()
-if(BUILD_RELAPACK)
-  add_library(RELAPACK OBJECT ${RELA_SOURCES})
-  list(APPEND TARGET_OBJS "$<TARGET_OBJECTS:RELAPACK>")
-endif()
+#if(BUILD_RELAPACK)
+#  add_library(RELAPACK OBJECT ${RELA_SOURCES})
+#  list(APPEND TARGET_OBJS "$<TARGET_OBJECTS:RELAPACK>")
+#endif()
 set(OpenBLAS_LIBS "")
 if(BUILD_STATIC_LIBS)
   add_library(${OpenBLAS_LIBNAME}_static STATIC ${TARGET_OBJS} ${OpenBLAS_DEF_FILE})
@@ -236,7 +238,7 @@ endif()
 set_target_properties(${OpenBLAS_LIBS} PROPERTIES OUTPUT_NAME ${OpenBLAS_LIBNAME})
 
 # Android needs to explicitly link against libm
-if(ANDROID)
+if (${CMAKE_SYSTEM_NAME} MATCHES "AIX|Android|Linux|FreeBSD|OpenBSD|NetBSD|DragonFly|Darwin")
   if(BUILD_STATIC_LIBS)
     target_link_libraries(${OpenBLAS_LIBNAME}_static m)
   endif()
@@ -396,7 +398,7 @@ if (BUILD_SHARED_LIBS AND NOT ${SYMBOLPREFIX}${SYMBOLSUFFIX} STREQUAL "")
     message(STATUS "adding suffix ${SYMBOLSUFFIX} to names of exported symbols in ${OpenBLAS_LIBNAME}")
   endif()
 
-  if (NOT DEFINED USE_PERL)
+  if (NOT USE_PERL)
   add_custom_command(TARGET ${OpenBLAS_LIBNAME}_shared POST_BUILD
     COMMAND  ${PROJECT_SOURCE_DIR}/exports/gensymbol "objcopy" "${ARCH}" "${BU}" "${EXPRECISION_IN}" "${NO_CBLAS_IN}" "${NO_LAPACK_IN}" "${NO_LAPACKE_IN}" "${NEED2UNDERSCORES_IN}" "${ONLY_CBLAS_IN}" \"${SYMBOLPREFIX}\" \"${SYMBOLSUFFIX}\" "${BUILD_LAPACK_DEPRECATED}" "${BUILD_BFLOAT16}" "${BUILD_SINGLE}" "${BUILD_DOUBLE}" "${BUILD_COMPLEX}" "${BUILD_COMPLEX16}" > ${PROJECT_BINARY_DIR}/objcopy.def
     COMMAND objcopy -v --redefine-syms ${PROJECT_BINARY_DIR}/objcopy.def  ${PROJECT_BINARY_DIR}/lib/lib${OpenBLAS_LIBNAME}.so

CONTRIBUTORS.md

@@ -211,4 +211,5 @@ In chronological order:
 * PLCT Lab, Institute of Software Chinese Academy of Sciences
   * [2022-03] Support RISC-V Vector Intrinisc 1.0 version.
 
-  
+* Pablo Romero <https://github.com/pablorcum>
+  * [2022-08] Fix building from sources for QNX

GotoBLAS_01Readme.txt

@@ -80,7 +80,7 @@
                       SUN
                       Fujitsu
 
-4. Suported precision
+4. Supported precision
 
  Now x86/x86_64 version support 80bit FP precision in addition to
 normal double presicion and single precision. Currently only