Merge branch 'OpenMathLib:develop' into m3m_exprec

.github/workflows/dynamic_arch.yml

@@ -158,7 +158,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        msystem: [UCRT64, MINGW32, CLANG64, CLANG32]
+        msystem: [UCRT64, MINGW32, CLANG64]
         idx: [int32, int64]
         build-type: [Release]
         include:
@@ -174,14 +174,6 @@ jobs:
             idx: int32
             target-prefix: mingw-w64-clang-x86_64
             fc-pkg: fc
-            # Compiling with Flang 16 seems to cause test errors on machines
-            # with AVX512 instructions. Revisit after MSYS2 distributes Flang 17.
-            no-avx512-flags: -DNO_AVX512=1
-          - msystem: CLANG32
-            idx: int32
-            target-prefix: mingw-w64-clang-i686
-            fc-pkg: cc
-            c-lapack-flags: -DC_LAPACK=ON
           - msystem: UCRT64
             idx: int64
             idx64-flags: -DBINARY=64 -DINTERFACE64=1
@@ -192,9 +184,6 @@ jobs:
             idx64-flags: -DBINARY=64 -DINTERFACE64=1
             target-prefix: mingw-w64-clang-x86_64
             fc-pkg: fc
-            # Compiling with Flang 16 seems to cause test errors on machines
-            # with AVX512 instructions. Revisit after MSYS2 distributes Flang 17.
-            no-avx512-flags: -DNO_AVX512=1
           - msystem: UCRT64
             idx: int32
             target-prefix: mingw-w64-ucrt-x86_64
@@ -203,8 +192,6 @@ jobs:
         exclude:
           - msystem: MINGW32
             idx: int64
-          - msystem: CLANG32
-            idx: int64
 
     defaults:
       run:
@@ -280,8 +267,6 @@ jobs:
                 -DNUM_THREADS=64 \
                 -DTARGET=CORE2 \
                 ${{ matrix.idx64-flags }} \
-                ${{ matrix.c-lapack-flags }} \
-                ${{ matrix.no-avx512-flags }} \
                 -DCMAKE_C_COMPILER_LAUNCHER=ccache \
                 -DCMAKE_Fortran_COMPILER_LAUNCHER=ccache \
                 ..

CMakeLists.txt

@@ -4,6 +4,7 @@
 
 cmake_minimum_required(VERSION 3.16.0)
 
+set (CMAKE_ASM_SOURCE_FILE_EXTENSIONS "S")
 project(OpenBLAS C ASM)
 
 set(OpenBLAS_MAJOR_VERSION 0)

CONTRIBUTORS.md

@@ -229,3 +229,6 @@ In chronological order:
 
 * Christopher Daley <https://github.com/cdaley>
   * [2024-01-24] Optimize GEMV forwarding on ARM64 systems
+
+* Aniket P. Garade <https://github.com/garadeaniket>   Sushil Pratap Singh <https://github.com/SushilPratap04>  Juliya James <https://github.com/Juliya32> 
+  *  [2024-12-13] Optimized swap and rot  Level-1 BLAS routines with ARM SVE  

Makefile.arm64

@@ -351,4 +351,31 @@ endif
 
 endif
 
+else
+# NVIDIA HPC options necessary to enable SVE in the compiler
+ifeq ($(CORE), THUNDERX2T99)
+CCOMMON_OPT += -tp=thunderx2t99
+FCOMMON_OPT += -tp=thunderx2t99
+endif
+ifeq ($(CORE), NEOVERSEN1)
+CCOMMON_OPT += -tp=neoverse-n1
+FCOMMON_OPT += -tp=neoverse-n1
+endif
+ifeq ($(CORE), NEOVERSEV1)
+CCOMMON_OPT += -tp=neoverse-v1
+FCOMMON_OPT += -tp=neoverse-v1
+endif
+ifeq ($(CORE), NEOVERSEV2)
+CCOMMON_OPT += -tp=neoverse-v2
+FCOMMON_OPT += -tp=neoverse-v2
+endif
+ifeq ($(CORE), ARMV8SVE)
+CCOMMON_OPT += -tp=neoverse-v2
+FCOMMON_OPT += -tp=neoverse-v2
+endif
+ifeq ($(CORE), ARMV9SVE)
+CCOMMON_OPT += -tp=neoverse-v2
+FCOMMON_OPT += -tp=neoverse-v2
+endif
+
 endif

Makefile.install

@@ -191,22 +191,29 @@ endif
 #Generating OpenBLASConfig.cmake
 	@echo Generating $(OPENBLAS_CMAKE_CONFIG) in $(DESTDIR)$(OPENBLAS_CMAKE_DIR)
 	@echo "SET(OpenBLAS_VERSION \"${VERSION}\")" > "$(DESTDIR)$(OPENBLAS_CMAKE_DIR)/$(OPENBLAS_CMAKE_CONFIG)"
-	@echo "SET(OpenBLAS_INCLUDE_DIRS ${OPENBLAS_INCLUDE_DIR})" >> "$(DESTDIR)$(OPENBLAS_CMAKE_DIR)/$(OPENBLAS_CMAKE_CONFIG)"
+	@echo "file(REAL_PATH \"../../..\" _OpenBLAS_ROOT_DIR BASE_DIRECTORY \$${CMAKE_CURRENT_LIST_DIR} )" > "$(DESTDIR)$(OPENBLAS_CMAKE_DIR)/$(OPENBLAS_CMAKE_CONFIG)"
+	@echo "SET(OpenBLAS_INCLUDE_DIRS \$${_OpenBLAS_ROOT_DIR}/include)" >> "$(DESTDIR)$(OPENBLAS_CMAKE_DIR)/$(OPENBLAS_CMAKE_CONFIG)"
 
 ifneq ($(NO_SHARED),1)
 #ifeq logical or
 ifeq ($(OSNAME), $(filter $(OSNAME),Linux FreeBSD NetBSD OpenBSD DragonFly))
-	@echo "SET(OpenBLAS_LIBRARIES ${OPENBLAS_LIBRARY_DIR}/$(LIBPREFIX)$(SYMBOLSUFFIX).so)" >> "$(DESTDIR)$(OPENBLAS_CMAKE_DIR)/$(OPENBLAS_CMAKE_CONFIG)"
+	@echo "SET(OpenBLAS_LIBRARIES \$${_OpenBLAS_ROOT_DIR}/lib/$(LIBPREFIX)$(SYMBOLSUFFIX).so)" >> "$(DESTDIR)$(OPENBLAS_CMAKE_DIR)/$(OPENBLAS_CMAKE_CONFIG)"
 endif
 ifeq ($(OSNAME), $(filter $(OSNAME),WINNT CYGWIN_NT))
-	@echo "SET(OpenBLAS_LIBRARIES ${OPENBLAS_BINARY_DIR}/$(LIBDLLNAME))" >> "$(DESTDIR)$(OPENBLAS_CMAKE_DIR)/$(OPENBLAS_CMAKE_CONFIG)"
+	@echo "SET(OpenBLAS_LIBRARIES \$${_OpenBLAS_ROOT_DIR}/bin/$(LIBDLLNAME))" >> "$(DESTDIR)$(OPENBLAS_CMAKE_DIR)/$(OPENBLAS_CMAKE_CONFIG)"
 endif
 ifeq ($(OSNAME), Darwin)
-	@echo "SET(OpenBLAS_LIBRARIES ${OPENBLAS_LIBRARY_DIR}/$(LIBPREFIX).dylib)" >> "$(DESTDIR)$(OPENBLAS_CMAKE_DIR)/$(OPENBLAS_CMAKE_CONFIG)"
+	@echo "SET(OpenBLAS_LIBRARIES \$${_OpenBLAS_ROOT_DIR}/lib/$(LIBPREFIX).dylib)" >> "$(DESTDIR)$(OPENBLAS_CMAKE_DIR)/$(OPENBLAS_CMAKE_CONFIG)"
+endif
+	@echo "add_library(OpenBLAS::OpenBLAS SHARED IMPORTED)"
+	@echo "target_include_directories(OpenBLAS::OpenBLAS INTERFACE \$${OpenBLAS_INCLUDE_DIRS})"
+ifeq ($(OSNAME), $(filter $(OSNAME),WINNT CYGWIN_NT))
+	@echo "set_property(TARGET OpenBLAS::OpenBLAS PROPERTY IMPORTED_LOCATION \$${OpenBLAS_LIBRARIES})"
+	@echo "set_property(TARGET OpenBLAS::OpenBLAS PROPERTY IMPORTED_IMPLIB \$${_OpenBLAS_ROOT_DIR}/lib/libopenblas.lib)"
 endif
 else
 #only static
-	@echo "SET(OpenBLAS_LIBRARIES ${OPENBLAS_LIBRARY_DIR}/$(LIBPREFIX).$(LIBSUFFIX))" >> "$(DESTDIR)$(OPENBLAS_CMAKE_DIR)/$(OPENBLAS_CMAKE_CONFIG)"
+	@echo "SET(OpenBLAS_LIBRARIES \$${_OpenBLAS_ROOT_DIR}/lib/$(LIBPREFIX).$(LIBSUFFIX))" >> "$(DESTDIR)$(OPENBLAS_CMAKE_DIR)/$(OPENBLAS_CMAKE_CONFIG)"
 endif
 #Generating OpenBLASConfigVersion.cmake
 	@echo Generating $(OPENBLAS_CMAKE_CONFIG_VERSION) in $(DESTDIR)$(OPENBLAS_CMAKE_DIR)

Makefile.system

@@ -1613,6 +1613,13 @@ NO_AFFINITY = 1
 endif
 endif
 
+ifeq ($(ARCH), POWER)
+ifeq ($(DEBUG), 1)
+CCOMMON_OPT := $(filter-out -O%, $(CCOMMON_OPT)) -O0
+FCOMMON_OPT := $(filter-out -O%, $(FCOMMON_OPT)) -O0
+endif
+endif
+
 ifdef NO_AFFINITY
 ifeq ($(NO_AFFINITY), 0)
 override undefine NO_AFFINITY

README.md

@@ -15,36 +15,44 @@ OSUOSL IBMZ-CI [![Build Status](http://ibmz-ci.osuosl.org/buildStatus/icon?job=O
 
 OpenBLAS is an optimized BLAS (Basic Linear Algebra Subprograms) library based on GotoBLAS2 1.13 BSD version.
 
-Please read the documentation in the OpenBLAS folder: <https://github.com/OpenMathLib/OpenBLAS/docs>.
+For more information about OpenBLAS, please see:
+
+- The documentation at [openmathlib.org/OpenBLAS/docs/](http://www.openmathlib.org/OpenBLAS/docs),
+- The home page at [openmathlib.org/OpenBLAS/](http://www.openmathlib.org/OpenBLAS).
 
 For a general introduction to the BLAS routines, please refer to the extensive documentation of their reference implementation hosted at netlib:
 <https://www.netlib.org/blas>. On that site you will likewise find documentation for the reference implementation of the higher-level library LAPACK - the **L**inear **A**lgebra **Pack**age that comes included with OpenBLAS. If you are looking for a general primer or refresher on Linear Algebra, the set of six
-20-minute lecture videos by Prof. Gilbert Strang on either MIT OpenCourseWare <https://ocw.mit.edu/resources/res-18-010-a-2020-vision-of-linear-algebra-spring-2020/> or Youtube <https://www.youtube.com/playlist?list=PLUl4u3cNGP61iQEFiWLE21EJCxwmWvvek> may be helpful.
+20-minute lecture videos by Prof. Gilbert Strang on either MIT OpenCourseWare [here](https://ocw.mit.edu/resources/res-18-010-a-2020-vision-of-linear-algebra-spring-2020/) or YouTube [here](https://www.youtube.com/playlist?list=PLUl4u3cNGP61iQEFiWLE21EJCxwmWvvek) may be helpful.
 
 ## Binary Packages
 
 We provide official binary packages for the following platform:
 
   * Windows x86/x86_64
 
-You can download them from [file hosting on sourceforge.net](https://sourceforge.net/projects/openblas/files/) or from the Releases section of the github project page, [https://github.com/OpenMathLib/OpenBLAS/releases](https://github.com/OpenMathLib/OpenBLAS/releases).
+You can download them from [file hosting on sourceforge.net](https://sourceforge.net/projects/openblas/files/) or from the [Releases section of the GitHub project page](https://github.com/OpenMathLib/OpenBLAS/releases).
+
+OpenBLAS is also packaged for many package managers - see [the installation section of the docs](http://www.openmathlib.org/OpenBLAS/docs/install/) for details.
 
 ## Installation from Source
 
-Download from project homepage, https://github.com/OpenMathLib/OpenBLAS/, or check out the code
-using Git from https://github.com/OpenMathLib/OpenBLAS.git. (If you want the most up to date version, be
-sure to use the develop branch - master is several years out of date due to a change of maintainership.)
-Buildtime parameters can be chosen in Makefile.rule, see there for a short description of each option.
-Most can also be given directly on the make or cmake command line.
+Obtain the source code from https://github.com/OpenMathLib/OpenBLAS/. Note that the default branch
+is `develop` (a `master` branch is still present, but far out of date).
+
+Build-time parameters can be chosen in `Makefile.rule`, see there for a short description of each option.
+Most options can also be given directly on the command line as parameters to your `make` or `cmake` invocation.
 
 ### Dependencies
 
 Building OpenBLAS requires the following to be installed:
 
 * GNU Make or CMake
-* A C compiler, e.g. GCC or Clang
+* A C compiler, e.g. GCC or Clang 
 * A Fortran compiler (optional, for LAPACK)
 
+In general, using a recent version of the compiler is strongly recommended.
+If a Fortran compiler is not available, it is possible to compile an older version of the included LAPACK
+that has been machine-translated to C.
 
 ### Normal compile
 
@@ -60,6 +68,9 @@ For building with `cmake`, the usual conventions apply, i.e. create a build dire
 OpenBLAS source directory or separate from it, and invoke `cmake` there with the path to the source tree and any 
 build options you plan to set.
 
+For more details, see the [Building from source](http://www.openmathlib.org/OpenBLAS/docs/install/#building-from-source)
+section in the docs.
+
 ### Cross compile
 
 Set `CC` and `FC` to point to the cross toolchains, and if you use `make`, also set `HOSTCC` to your host C compiler.
@@ -76,10 +87,12 @@ Examples:
   make CC="i686-w64-mingw32-gcc -Bstatic" FC="i686-w64-mingw32-gfortran -static-libgfortran" TARGET=HASWELL BINARY=32 CROSS=1 NUM_THREADS=20 CONSISTENT_FPCSR=1 HOSTCC=gcc
   ```
 
-You can find instructions for other cases both in the "Supported Systems" section below and in the docs folder. The .yml scripts included with the sources (which contain the
+You can find instructions for other cases both in the "Supported Systems" section below and in
+the [Building from source docs](http://www.openmathlib.org/OpenBLAS/docs/install).
+The `.yml` scripts included with the sources (which contain the
 build scripts for the "continuous integration" (CI) build tests automatically run on every proposed change to the sources) may also provide additional hints.
 
-When compiling for a more modern CPU TARGET of the same architecture, e.g. TARGET=SKYLAKEX on a HASWELL host, option "CROSS=1" can be used to suppress the automatic invocation of the tests at the end of the build.
+When compiling for a more modern CPU target of the same architecture, e.g. `TARGET=SKYLAKEX` on a `HASWELL` host, option `CROSS=1` can be used to suppress the automatic invocation of the tests at the end of the build.
 
 ### Debug version
 
@@ -325,11 +338,14 @@ Please see Changelog.txt.
 
 ## Troubleshooting
 
-* Please read the [FAQ](https://github.com/OpenMathLib/OpenBLAS/docs/faq,md) in the docs folder first.
+* Please read the [FAQ](http://www.openmathlib.org/OpenBLAS/docs/faq) section of the docs first.
 * Please use GCC version 4.6 and above to compile Sandy Bridge AVX kernels on Linux/MinGW/BSD.
 * Please use Clang version 3.1 and above to compile the library on Sandy Bridge microarchitecture.
   Clang 3.0 will generate the wrong AVX binary code.
-* Please use GCC version 6 or LLVM version 6 and above to compile Skylake AVX512 kernels.
+* Please use GCC version 6 or LLVM version 6 and above to compile Skylake/CooperLake AVX512 kernels
+* Please use LLVM version 18 and above (version 19 and above on Windows) if you plan to use
+  its new flang compiler for Fortran
+* Please use GCC version 11 and above to compile OpenBLAS on the POWER architecture
 * The number of CPUs/cores should be less than or equal to 256. On Linux `x86_64` (`amd64`),
   there is experimental support for up to 1024 CPUs/cores and 128 numa nodes if you build
   the library with `BIGNUMA=1`.
@@ -350,4 +366,4 @@ Please see Changelog.txt.
 
 ## Donation
 
-Please read [this wiki page](https://github.com/xianyi/OpenBLAS/wiki/Donation).
+Please see [the donations section](http://www.openmathlib.org/OpenBLAS/docs/about/#donations) in the docs.

benchmark/pybench/README.md

@@ -43,7 +43,17 @@ have all what it takes to build OpenBLAS from source, plus `python` and
 $ python -mpip install numpy meson ninja pytest pytest-benchmark
 ```
 
-The benchmark syntax is consistent with that of `pytest-benchmark` framework. The incantation to run the suite locally is `$ pytest benchmark/pybench/benchmarks/test_blas.py`.
+The Meson build system looks for the installed OpenBLAS using pkgconfig, so the openblas.pc created during the OpenBLAS build needs
+to be somewhere on the search path of pkgconfig or in a folder pointed to by the environment variable PKG_CONFIG_PATH.
+
+If you want to build the benchmark suite using flang (or flang-new) instead of gfortran for the Fortran parts, you currently need
+to edit the meson.build file and change the line `'fortran_std=legacy'` to `'fortran_std=none'` to work around an incompatibility
+between Meson and flang.
+
+If you are building and running the benchmark under MS Windows, it may be necessary to copy the generated openblas_wrap module from
+your build folder to the `benchmarks` folder.
+
+The benchmark syntax is consistent with that of `pytest-benchmark` framework. The incantation to run the suite locally is `$ pytest benchmark/pybench/benchmarks/bench_blas.py`.
 
 An ASV compatible benchmark suite is planned but currently not implemented.
 

c_check

@@ -6,6 +6,9 @@ hostarch=`uname -m | sed -e 's/i.86/x86/'`
 if [ "$hostos" = "AIX" ] || [ "$hostos" = "SunOS" ]; then
     hostarch=`uname -p`
 fi
+if [ "$hostarch" = "evbarm" ]; then
+    hostarch=`uname -p`
+fi
 case "$hostarch" in
     amd64) hostarch=x86_64 ;;
     arm*) [ "$hostarch" = "arm64" ] || hostarch='arm' ;;

cmake/f_check.cmake

@@ -45,13 +45,15 @@ if (NOT ONLY_CBLAS)
 
   # TODO: detect whether underscore needed, set #defines and BU appropriately - use try_compile
   # TODO: set FEXTRALIB flags a la f_check?
-
+  if (NOT (${CMAKE_SYSTEM_NAME} MATCHES "Windows" AND x${CMAKE_Fortran_COMPILER_ID} MATCHES "IntelLLVM"))
   set(BU "_")
   file(APPEND ${TARGET_CONF_TEMP}
     "#define BUNDERSCORE _\n"
     "#define NEEDBUNDERSCORE 1\n"
     "#define NEED2UNDERSCORES 0\n")
-
+  else ()
+	  set (FCOMMON_OPT "${FCOMMON_OPT} /fp:precise /recursive /names:lowercase /assume:nounderscore")
+  endif()
 else ()
 
   #When we only build CBLAS, we set NOFORTRAN=2

cmake/fc.cmake

@@ -269,6 +269,31 @@ if (${F_COMPILER} STREQUAL "CRAY")
   endif ()
 endif ()
 
+if (${F_COMPILER} STREQUAL "NAGFOR")
+  set(CCOMMON_OPT "${CCOMMON_OPT} -DF_INTERFACE_NAG")
+  if (INTERFACE64)
+    set(FCOMMON_OPT "${FCOMMON_OPT} -i8")
+  endif ()
+  # Options from Makefile.system
+  # -dcfuns: Enable non-standard double precision complex intrinsic functions
+  # -ieee=full: enables all IEEE arithmetic facilities including non-stop arithmetic.
+  # -w=obs: Suppress warning messages about obsolescent features
+  # -thread_safe: Compile code for safe execution in a multi-threaded environment.
+  # -recursive: Specifies that procedures are RECURSIVE by default.
+  set(FCOMMON_OPT "${FCOMMON_OPT} -dcfuns -recursive -ieee=full -w=obs -thread_safe")
+  # Options from Reference-LAPACK
+  # Suppress compiler banner and summary
+  set(FCOMMON_OPT "${FCOMMON_OPT} -quiet")
+  # Disable other common warnings
+  # -w=x77: Suppress warning messages about Fortran 77 features
+  # -w=ques: Suppress warning messages about questionable usage
+  # -w=unused: Suppress warning messages about unused variables
+  set(FCOMMON_OPT "${FCOMMON_OPT} -w=x77 -w=ques -w=unused")
+  if (USE_OPENMP)
+    set(FCOMMON_OPT "${FCOMMON_OPT} -openmp")
+  endif ()
+endif ()
+
 # from the root Makefile - this is for lapack-netlib to compile the correct secnd file.
 if (${F_COMPILER} STREQUAL "GFORTRAN")
   set(TIMER "INT_ETIME")

cmake/lapack.cmake

@@ -1018,7 +1018,12 @@ foreach (LA_FILE ${LA_GEN_SRC})
 endforeach ()
 
 if (NOT C_LAPACK)
-  set_source_files_properties(${LA_SOURCES} PROPERTIES COMPILE_FLAGS "${LAPACK_FFLAGS}")
+  # The below line is duplicating Fortran flags but NAG has a few flags
+  # that cannot be specified twice. It's possible this is not needed for
+  # any compiler, but for safety, we only turn off for NAG
+  if (NOT ${F_COMPILER} STREQUAL "NAGFOR")
+    set_source_files_properties(${LA_SOURCES} PROPERTIES COMPILE_FLAGS "${LAPACK_FFLAGS}")
+  endif ()
   if (${F_COMPILER} STREQUAL "GFORTRAN")
     set_source_files_properties(${LA_SOURCES} PROPERTIES COMPILE_FLAGS "${LAPACK_FFLAGS} -fno-tree-vectorize")
   endif()

cmake/prebuild.cmake

@@ -58,7 +58,7 @@ set(TARGET_CONF_TEMP "${PROJECT_BINARY_DIR}/${TARGET_CONF}.tmp")
 
 # c_check
 set(FU "")
-if (APPLE OR (MSVC AND NOT ${CMAKE_C_COMPILER_ID} MATCHES "Clang"))
+if (APPLE OR (MSVC AND NOT (${CMAKE_C_COMPILER_ID} MATCHES "Clang" OR ${CMAKE_C_COMPILER_ID} MATCHES "IntelLLVM")))
   set(FU "_")
 endif()
 if(MINGW AND NOT MINGW64)
@@ -1433,7 +1433,9 @@ else(NOT CMAKE_CROSSCOMPILING)
     message(STATUS "MSVC")
     set(GETARCH_FLAGS ${GETARCH_FLAGS} -DFORCE_GENERIC)
   else()
-    list(APPEND GETARCH_SRC ${PROJECT_SOURCE_DIR}/cpuid.S)
+    if ("${CMAKE_SYSTEM_NAME}" STREQUAL "Darwin")
+      list(APPEND GETARCH_SRC ${PROJECT_SOURCE_DIR}/cpuid.S)
+    endif()
     if (DEFINED TARGET_CORE)
     set(GETARCH_FLAGS ${GETARCH_FLAGS} -DFORCE_${TARGET_CORE})
   endif ()