Merge branch 'develop' into issue5414

Author: martin-frbg

CMakeLists.txt

@@ -708,6 +708,39 @@ if(NOT NO_LAPACKE)
   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 ${CMAKE_INSTALL_INCLUDEDIR})
+  if (NOT (x${SYMBOLPREFIX}${SYMBOLSUFFIX} STREQUAL "x"))
+  message (STATUS "Generating lapacke.h in ${CMAKE_INSTALL_INCLUDEDIR}")
+  set(LAPACKE_H ${CMAKE_BINARY_DIR}/generated/lapacke.h)
+  file(READ ${CMAKE_CURRENT_SOURCE_DIR}/lapack-netlib/LAPACKE/include/lapacke.h LAPACKE_H_CONTENTS)
+  if (NOT ${SYMBOLPREFIX} STREQUAL "")
+	  string(REGEX REPLACE "(LAPACKE_*)" " ${SYMBOLPREFIX}\\1" LAPACKE_H_CONTENTS_NEW	"${LAPACKE_H_CONTENTS}")
+	  string(REPLACE "_ ${SYMBOLPREFIX}LAPACKE_H_"  "_LAPACKE_H_" LAPACKE_H_CONTENTS ${LAPACKE_H_CONTENTS_NEW})
+	  string(REPLACE "${SYMBOLPREFIX}LAPACKE_malloc"  "LAPACKE_malloc" LAPACKE_H_CONTENTS_NEW ${LAPACKE_H_CONTENTS})
+	  string(REPLACE "${SYMBOLPREFIX}LAPACKE_free"  "LAPACKE_free" LAPACKE_H_CONTENTS ${LAPACKE_H_CONTENTS_NEW})
+	  set(LAPACKE_H_CONTENTS_NEW ${LAPACKE_H_CONTENTS})
+  endif()
+  if (NOT ${SYMBOLSUFFIX} STREQUAL "")
+	  string(REGEX REPLACE "(${SYMBOLPREFIX}LAPACKE_[a-z1-9]*[^ (]*)" "\\1${SYMBOLSUFFIX}" LAPACKE_H_CONTENTS_NEW	"${LAPACKE_H_CONTENTS}")
+	  string(REPLACE "#define${SYMBOLSUFFIX}"  "#define" LAPACKE_H_CONTENTS ${LAPACKE_H_CONTENTS_NEW})
+	  string(REPLACE "LAPACKE_malloc${SYMBOLSUFFIX}"  "LAPACKE_malloc" LAPACKE_H_CONTENTS_NEW ${LAPACKE_H_CONTENTS})
+	  string(REPLACE "LAPACKE_free${SYMBOLSUFFIX}"  "LAPACKE_free" LAPACKE_H_CONTENTS ${LAPACKE_H_CONTENTS_NEW})
+	  set(LAPACKE_H_CONTENTS_NEW ${LAPACKE_H_CONTENTS})
+  endif()
+  file(WRITE ${LAPACKE_H} "${LAPACKE_H_CONTENTS_NEW}")
+  install (FILES ${LAPACKE_H} DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
+  message (STATUS "Generating lapack.h in ${CMAKE_INSTALL_INCLUDEDIR}")
+  set(LAPACK_H ${CMAKE_BINARY_DIR}/generated/lapack.h)
+  file(READ ${CMAKE_CURRENT_SOURCE_DIR}/lapack-netlib/LAPACKE/include/lapack.h LAPACK_H_CONTENTS)
+  if (NOT ${SYMBOLPREFIX} STREQUAL "")
+	  string(REGEX REPLACE "(LAPACK_[a-z1-9]*[ \(][.\)]*)" "${SYMBOLPREFIX}\\1" LAPACK_H_CONTENTS_NEW	"${LAPACK_H_CONTENTS}")
+	  set(LAPACK_H_CONTENTS ${LAPACK_H_CONTENTS_NEW})
+  endif()
+  if (NOT ${SYMBOLSUFFIX} STREQUAL "")
+	  string(REGEX REPLACE "(${SYMBOLPREFIX}LAPACK_[a-z1-9]*)([ \(].\)" "\\1${SYMBOLSUFFIX}\\2" LAPACK_H_CONTENTS_NEW	"${LAPACK_H_CONTENTS}")
+  endif()
+  file(WRITE ${LAPACK_H} "${LAPACK_H_CONTENTS_NEW}")
+  install (FILES ${LAPACK_H} DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
+endif()
 endif()
 
 # Install pkg-config files

CONTRIBUTORS.md

@@ -29,6 +29,9 @@
 * Annop Wongwathanarat <annop.wongwathanarat@arm.com>
   * Optimizations and other improvements targeting AArch64
 
+* Anna Mayne <anna.mayne@arm.com>
+  * Optimizations and other improvements targeting AArch64
+
 ## Previous Developers
 
 * Zaheer Chothia <zaheer.chothia@gmail.com>
@@ -267,3 +270,5 @@ In chronological order:
   * [2025-05-29] Optimise axpby kernel for RISCV64_ZVL256B
   * [2025-06-05] Optimise hbmv kernel for RISCV64_ZVL256B
 
+* Anna Mayne <anna.mayne@arm.com>
+  * [2025-11-19] Update thread throttling profile for SGEMV on NEOVERSEV1 and NEOVERSEV2

Jenkinsfile.pwr

@@ -1,16 +1,31 @@
 pipeline {
-    agent { 
-        docker {
-            image 'osuosl/ubuntu-ppc64le:18.04'
-        }
-    }
+    agent none
     stages {
-        stage('Build') {
+        stage('GCC build') {
+            agent {
+                docker {
+                    image 'osuosl/ubuntu-ppc64le:18.04' // gcc 7, gfortran 7
+                }
+            }
             steps {
+                checkout scm
                 sh 'sudo apt update'
                 sh 'sudo apt install gfortran -y'
                 sh 'make clean && make'
             }
         }
+        stage('Clang build') {
+            agent {
+                docker {
+                    image 'osuosl/ubuntu-ppc64le:20.04' // clang 10, gfortran 9
+                }
+            }
+            steps {
+                checkout scm
+                sh 'sudo apt update'
+                sh 'sudo apt install -y clang gfortran'
+                sh 'make clean && make CC=clang'
+            }
+        }
     }
 }

Makefile.install

@@ -93,9 +93,27 @@ endif
 
 ifneq ($(OSNAME), AIX)
 ifneq ($(NO_LAPACKE), 1)
+	@cp $(NETLIB_LAPACK_DIR)/LAPACKE/include/lapacke.h lapacke_h.tmp
+ifdef SYMBOLPREFIX
+	@sed 's/LAPACKE_[a-z1-9].[^() ]*/$(SYMBOLPREFIX)&/g' lapacke_h.tmp > lapacke.tmp2
+	@mv lapacke.tmp2 lapacke_h.tmp
+endif
+ifdef SYMBOLSUFFIX
+	@sed 's/LAPACKE_[a-z1-9].[^() ]*/&$(SYMBOLSUFFIX)/g' lapacke_h.tmp > lapacke.tmp2
+	@mv lapacke.tmp2 lapacke_h.tmp
+endif
+	@-install -m644 lapacke_h.tmp "$(DESTDIR)$(OPENBLAS_INCLUDE_DIR)/lapacke.h"
 	@echo Copying LAPACKE header files to $(DESTDIR)$(OPENBLAS_INCLUDE_DIR)
-	@-install -m644 $(NETLIB_LAPACK_DIR)/LAPACKE/include/lapack.h "$(DESTDIR)$(OPENBLAS_INCLUDE_DIR)/lapack.h"
-	@-install -m644 $(NETLIB_LAPACK_DIR)/LAPACKE/include/lapacke.h "$(DESTDIR)$(OPENBLAS_INCLUDE_DIR)/lapacke.h"
+	@cp $(NETLIB_LAPACK_DIR)/LAPACKE/include/lapack.h lapack_h.tmp
+ifdef SYMBOLPREFIX
+	@sed 's/LAPACK_[a-z1-9]*(\.\.\.)/$(SYMBOLPREFIX)&/g' lapack_h.tmp > lapack.tmp2
+	@mv lapack.tmp2 lapack_h.tmp
+endif
+ifdef SYMBOLSUFFIX
+	@sed 's/\(#define $(SYMBOLPREFIX)LAPACK_[a-z1-9].*\)\((...)\)/\1$(SYMBOLSUFFIX)\2/g' lapack_h.tmp > lapack.tmp2
+	@mv lapack.tmp2 lapack_h.tmp
+endif
+	@-install -m644 lapack_h.tmp "$(DESTDIR)$(OPENBLAS_INCLUDE_DIR)/lapack.h"
 	@-install -m644 $(NETLIB_LAPACK_DIR)/LAPACKE/include/lapacke_config.h "$(DESTDIR)$(OPENBLAS_INCLUDE_DIR)/lapacke_config.h"
 	@-install -m644 $(NETLIB_LAPACK_DIR)/LAPACKE/include/lapacke_mangling_with_flags.h.in "$(DESTDIR)$(OPENBLAS_INCLUDE_DIR)/lapacke_mangling.h"
 	@-install -m644 $(NETLIB_LAPACK_DIR)/LAPACKE/include/lapacke_utils.h "$(DESTDIR)$(OPENBLAS_INCLUDE_DIR)/lapacke_utils.h"

azure-pipelines.yml

@@ -155,7 +155,7 @@ jobs:
 
 - job: OSX_OpenMP
   pool:
-     vmImage: 'macOS-13'
+     vmImage: 'macOS-14'
   steps:   
   - script: |
       brew update
@@ -165,7 +165,7 @@ jobs:
      
 - job: OSX_GCC_Nothreads
   pool:
-     vmImage: 'macOS-13'
+     vmImage: 'macOS-14'
   steps:   
   - script: |
       brew update
@@ -221,7 +221,7 @@ jobs:
       
 - job: OSX_dynarch_cmake
   pool:
-     vmImage: 'macOS-13'
+     vmImage: 'macOS-14'
   variables:
      LD_LIBRARY_PATH: /usr/local/opt/llvm/lib
      LIBRARY_PATH: /usr/local/opt/llvm/lib
@@ -268,7 +268,7 @@ jobs:
  
 - job: OSX_NDK_ARMV7
   pool:
-     vmImage: 'macOS-13'
+     vmImage: 'macOS-14'
   steps:   
   - script: | 
       brew update
@@ -278,35 +278,37 @@ jobs:
 
 - job: OSX_IOS_ARMV8
   pool:
-     vmImage: 'macOS-13'
+     vmImage: 'macOS-14'
   variables:
-     CC: /Applications/Xcode_14.2.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/clang
-     CFLAGS: -O2 -Wno-macro-redefined -isysroot /Applications/Xcode_14.2.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS16.2.sdk -arch arm64 -miphoneos-version-min=10.0
+     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
   steps:
   - script: |
+     ls /Applications/Xcode_16.2.app/Contents/Developer/Platforms/
+     ls /Applications/Xcode_16.2.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-13'
+     vmImage: 'macOS-14'
   variables:
-     CC: /Applications/Xcode_14.2.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/clang
-     CFLAGS: -O2 -mno-thumb -Wno-macro-redefined -isysroot /Applications/Xcode_14.2.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS16.2.sdk -arch armv7 -miphoneos-version-min=5.1
+     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
   steps:
   - script: |
      make TARGET=ARMV7 DYNAMIC_ARCH=1 NUM_THREADS=32 HOSTCC=clang NOFORTRAN=1
 
 - job: OSX_xbuild_DYNAMIC_ARM64
   pool:
-     vmImage: 'macOS-13'
+     vmImage: 'macOS-14'
   variables:
-     CC: /Applications/Xcode_14.2.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/clang
-     CFLAGS: -O2 -Wno-macro-redefined -isysroot /Applications/Xcode_14.2.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX13.1.sdk -arch arm64
+     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
   steps:
   - script: |
-     ls /Applications/Xcode_14.2.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs
+     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_14.2.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/clang --version
+     /Applications/Xcode_16.2.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

common.h

@@ -362,18 +362,6 @@ typedef int blasint;
 #define MAX_CPU_NUMBER 2
 #endif
 
-#if defined(OS_SUNOS)
-#define YIELDING	thr_yield()
-#endif
-
-#if defined(OS_WINDOWS)
-#if defined(_MSC_VER) && !defined(__clang__)
-#define YIELDING    YieldProcessor()
-#else
-#define YIELDING	SwitchToThread()
-#endif
-#endif
-
 #if defined(ARMV7) || defined(ARMV6) || defined(ARMV8) || defined(ARMV5)
 #define YIELDING        __asm__ __volatile__ ("nop;nop;nop;nop;nop;nop;nop;nop; \n");
 #endif
@@ -398,14 +386,26 @@ typedef int blasint;
 #endif
 #endif
 
-
 #ifdef __EMSCRIPTEN__
 #define YIELDING
 #endif
 
+#if defined(_MSC_VER) && !defined(__clang__)
+#undef YIELDING // MSVC doesn't support assembly code
+#define YIELDING    	YieldProcessor()
+#endif
+
 #ifndef YIELDING
+#if defined(OS_SUNOS)
+#define YIELDING	thr_yield()
+
+#elif defined(OS_WINDOWS)
+#define YIELDING	SwitchToThread()
+
+#else // assume POSIX.1-2008
 #define YIELDING	sched_yield()
 #endif
+#endif
 
 /***
 To alloc job_t on heap or stack.
@@ -765,7 +765,7 @@ static __inline int readenv_atoi(char *env) {
 	return 0;
 }
 #else
-#ifdef OS_WINDOWS
+#if defined(OS_WINDOWS) && !defined(OS_CYGWIN_NT)
 static __inline int readenv_atoi(char *env) {
   env_var_t p;
   return readenv(p,env) ? 0 : atoi(p);

common_level3.h

@@ -110,6 +110,31 @@ void ssyrk_direct_alpha_betaLT(BLASLONG N, BLASLONG K,
 	float beta,
 	float * C, BLASLONG strideC);
 
+void ssyr2k_direct_alpha_betaUN(BLASLONG N, BLASLONG K,
+	float alpha,
+	float * A, BLASLONG strideA,
+	float * B, BLASLONG strideB,
+	float beta,
+	float * R, BLASLONG strideR);
+void ssyr2k_direct_alpha_betaUT(BLASLONG N, BLASLONG K,
+	float alpha,
+	float * A, BLASLONG strideA,
+	float * B, BLASLONG strideB,
+	float beta,
+	float * R, BLASLONG strideR);
+void ssyr2k_direct_alpha_betaLN(BLASLONG N, BLASLONG K,
+	float alpha,
+	float * A, BLASLONG strideA,
+	float * B, BLASLONG strideB,
+	float beta,
+	float * R, BLASLONG strideR);
+void ssyr2k_direct_alpha_betaLT(BLASLONG N, BLASLONG K,
+	float alpha,
+	float * A, BLASLONG strideA,
+	float * B, BLASLONG strideB,
+	float beta,
+	float * R, BLASLONG strideR);
+
 int sgemm_direct_performant(BLASLONG M, BLASLONG N, BLASLONG K);
 
 int shgemm_beta(BLASLONG, BLASLONG, BLASLONG, float,

common_param.h

@@ -269,6 +269,10 @@ int (*shgemv_t) (BLASLONG, BLASLONG, float, hfloat16 *, BLASLONG, hfloat16 *, BL
   void (*ssyrk_direct_alpha_betaUT) (BLASLONG, BLASLONG, float, float *, BLASLONG, float, float *, BLASLONG);
   void (*ssyrk_direct_alpha_betaLN) (BLASLONG, BLASLONG, float, float *, BLASLONG, float, float *, BLASLONG);
   void (*ssyrk_direct_alpha_betaLT) (BLASLONG, BLASLONG, float, float *, BLASLONG, float, float *, BLASLONG);
+  void (*ssyr2k_direct_alpha_betaUN) (BLASLONG, BLASLONG, float, float *, BLASLONG, float *, BLASLONG, float, float *, BLASLONG);
+  void (*ssyr2k_direct_alpha_betaUT) (BLASLONG, BLASLONG, float, float *, BLASLONG, float *, BLASLONG, float, float *, BLASLONG);
+  void (*ssyr2k_direct_alpha_betaLN) (BLASLONG, BLASLONG, float, float *, BLASLONG, float *, BLASLONG, float, float *, BLASLONG);
+  void (*ssyr2k_direct_alpha_betaLT) (BLASLONG, BLASLONG, float, float *, BLASLONG, float *, BLASLONG, float, float *, BLASLONG);
 #endif
 
 

common_s.h

@@ -60,6 +60,10 @@
 #define  SSYRK_DIRECT_ALPHA_BETA_UT         ssyrk_direct_alpha_betaUT
 #define  SSYRK_DIRECT_ALPHA_BETA_LN         ssyrk_direct_alpha_betaLN
 #define  SSYRK_DIRECT_ALPHA_BETA_LT         ssyrk_direct_alpha_betaLT
+#define  SSYR2K_DIRECT_ALPHA_BETA_UN         ssyr2k_direct_alpha_betaUN
+#define  SSYR2K_DIRECT_ALPHA_BETA_UT         ssyr2k_direct_alpha_betaUT
+#define  SSYR2K_DIRECT_ALPHA_BETA_LN         ssyr2k_direct_alpha_betaLN
+#define  SSYR2K_DIRECT_ALPHA_BETA_LT         ssyr2k_direct_alpha_betaLT
 
 #define	SGEMM_ONCOPY		sgemm_oncopy
 #define	SGEMM_OTCOPY		sgemm_otcopy
@@ -240,6 +244,10 @@
 #define  SSYRK_DIRECT_ALPHA_BETA_UT     gotoblas -> ssyrk_direct_alpha_betaUT
 #define  SSYRK_DIRECT_ALPHA_BETA_LN     gotoblas -> ssyrk_direct_alpha_betaLN
 #define  SSYRK_DIRECT_ALPHA_BETA_LT     gotoblas -> ssyrk_direct_alpha_betaLT
+#define  SSYR2K_DIRECT_ALPHA_BETA_UN     gotoblas -> ssyr2k_direct_alpha_betaUN
+#define  SSYR2K_DIRECT_ALPHA_BETA_UT     gotoblas -> ssyr2k_direct_alpha_betaUT
+#define  SSYR2K_DIRECT_ALPHA_BETA_LN     gotoblas -> ssyr2k_direct_alpha_betaLN
+#define  SSYR2K_DIRECT_ALPHA_BETA_LT     gotoblas -> ssyr2k_direct_alpha_betaLT
 #endif
 
 #define	SGEMM_ONCOPY		gotoblas -> sgemm_oncopy

docs/install.md

@@ -217,30 +217,34 @@ in this section, since the process for each is quite different.
 For Visual Studio, you can use CMake to generate Visual Studio solution files;
 note that you will need at least CMake 3.11 for linking to work correctly).
 
-Note that you need a Fortran compiler if you plan to build and use the LAPACK
-functions included with OpenBLAS. The sections below describe using either
+Note that you need a Fortran compiler if you plan to build and use the latest version 
+of the LAPACK functions included with OpenBLAS. (If you do not have a Fortran compiler
+installed, you can build an older version of the LAPACK sources that has been converted
+to C - but its performance will likely be slower and accuracy may be poorer too.) 
+The sections below describe using either
 `flang` as an add-on to clang/LLVM or `gfortran` as part of MinGW for this
 purpose. If you want to use the Intel Fortran compiler (`ifort` or `ifx`) for
 this, be sure to also use the Intel C compiler (`icc` or `icx`) for building
 the C parts, as the ABI imposed by `ifort` is incompatible with MSVC
 
 A fully-optimized OpenBLAS that can be statically or dynamically linked to your
 application can currently be built for the 64-bit architecture with the LLVM
-compiler infrastructure. We're going to use [Miniconda3](https://docs.anaconda.com/miniconda/)
+compiler infrastructure. We're going to use [Miniforge3] the pre-configured
+and more versatile alternative to [Miniconda](https://docs.anaconda.com/miniconda/)
 to grab all of the tools we need, since some of them are in an experimental
 status. Before you begin, you'll need to have Microsoft Visual Studio 2015 or
 newer installed.
 
-1. Install Miniconda3 for 64-bit Windows using `winget install --id Anaconda.Miniconda3`,
-   or easily download from [conda.io](https://docs.conda.io/en/latest/miniconda.html).
-2. Open the "Anaconda Command Prompt" now available in the Start Menu, or at `%USERPROFILE%\miniconda3\shell\condabin\conda-hook.ps1`.
+1. Install Miniforge for 64-bit Windows with the latest version of the installer Miniforge3-Windows-x86_64.exe
+   available on [github.com](https://github.com/conda-forge/miniforge/releases/)
+2. Open the "Miniforge Command Prompt" now available in the Start Menu, or at `%USERPROFILE%\miniforge3\shell\condabin\conda-hook.ps1`.
 3. In that command prompt window, use `cd` to change to the directory where you want to build OpenBLAS.
 4. Now install all of the tools we need:
    ```
    conda update -n base conda
-   conda config --add channels conda-forge
-   conda install -y cmake flang clangdev perl libflang ninja
+   conda install -y cmake flang_win-64 clangdev perl libflang ninja
    ```
+   (if you want to build with OpenMP support, add `llvm-openmp` and `llvm-openmp-fortran`)
 5.  Still in the Anaconda Command Prompt window, activate the 64-bit MSVC environment with `vcvarsall x64`.
     On Windows 11 with Visual Studio 2022, this would be done by invoking: