Merge pull request #3216 from xianyi/develop

Update from develop for 0.3.15 release

Author: martin-frbg

.travis.yml

@@ -224,12 +224,21 @@ matrix:
       before_script:
         - COMMON_FLAGS="DYNAMIC_ARCH=1 NUM_THREADS=32"
         - brew update
-        - brew install gcc@10
       script:
         - travis_wait 45 make QUIET_MAKE=1 $COMMON_FLAGS $BTYPE
       env:
-        - BTYPE="TARGET=NEHALEM BINARY=64 INTERFACE64=1 FC=gfortran-10"
-        
+        - BTYPE="TARGET=HASWELL USE_OPENMP=1 BINARY=64 INTERFACE64=1 CC=gcc-10 FC=gfortran-10"
+
+    - <<: *test-macos
+      osx_image: xcode12
+      before_script:
+        - COMMON_FLAGS="DYNAMIC_ARCH=1 NUM_THREADS=32"
+        - brew update
+      script:
+        - travis_wait 45 make QUIET_MAKE=1 $COMMON_FLAGS $BTYPE
+      env:
+        - BTYPE="TARGET=NEHALEM BINARY=64 INTERFACE64=1 FC=gfortran-10"      
+
   #  - <<: *test-macos
   #    osx_image: xcode10
   #    env:

CMakeLists.txt

@@ -6,7 +6,7 @@ cmake_minimum_required(VERSION 2.8.5)
 project(OpenBLAS C ASM)
 set(OpenBLAS_MAJOR_VERSION 0)
 set(OpenBLAS_MINOR_VERSION 3)
-set(OpenBLAS_PATCH_VERSION 14)
+set(OpenBLAS_PATCH_VERSION 14.dev)
 set(OpenBLAS_VERSION "${OpenBLAS_MAJOR_VERSION}.${OpenBLAS_MINOR_VERSION}.${OpenBLAS_PATCH_VERSION}")
 
 # Adhere to GNU filesystem layout conventions

Changelog.txt

@@ -1,4 +1,54 @@
 OpenBLAS ChangeLog
+====================================================================
+Version 0.3.15
+  2-May-2021
+
+common:
+ - imported improvements and bugfixes from Reference-LAPACK 3.9.1
+ - imported LAPACKE interface fixes from Reference-LAPACK PRs 534 + 537
+ - fixed a problem in the cpu detection of 0.3.14 that prevented cross-compilation
+ - fixed a sequence problem in the generation of softlinks to the library in GMAKE
+
+RISC V:
+ - fixed compilation on RISCV (missing entry in getarch)
+ - fixed a potential division by zero in CROTG and ZROTG
+
+POWER:
+ - fixed LAPACK testsuite failures seen with the NVIDIA HPC compiler
+ - improved CGEMM, DGEMM and ZGEMM performance on POWER10
+ - added an optimized ZGEMV kernel for POWER10
+ - fixed a potential division by zero in CROTG and ZROTG
+
+x86_64:
+ - added support for Intel Control-flow Enforcement Technology (CET)
+ - reverted the DOMATCOPY_RT code to the generic C version
+ - fixed a bug in the AVX512 SGEMM kernel introduced in 0.3.14
+ - fixed misapplication of -msse flag to non-SSE cpus in DYNAMIC_ARCH
+ - added support for compilation of the benchmarks on older OSX versions
+ - fix propagation of the NO_AVX512 option in CMAKE builds
+ - fix compilation of the AVX512 SGEMM kernel with clang-cl on Windows
+ - fixed compilation of the CTESTs with INTERFACE64=1 (random faults on OSX)
+ - corrected the Haswell DROT kernel to require AVX2/FMA3 rather than AVX512
+
+ARM:
+ - fixed a potential division by zero in CROTG and ZROTG
+ - fixed a potential overflow in IMATCOPY/ZIMATCOPY and the CTESTs
+
+ARM64:
+ - fixed spurious reads outside the array in the SGEMM tcopy macro
+ - fixed a potential division by zero in CROTG and ZROTG
+ - fixed a segmentation fault in DYNAMIC_ARCH builds (reappeared in 0.3.14)
+
+MIPS
+ - fixed a potential division by zero in CROTG and ZROTG
+ - fixed a potential overflow in IMATCOPY/ZIMATCOPY and the CTESTs
+
+MIPS64:
+ - fixed a potential division by zero in CROTG and ZROTG
+
+SPARC:
+ - fixed a potential division by zero in CROTG and ZROTG
+
 ====================================================================
 Version 0.3.14
  17-Mar-2021

Makefile

@@ -167,7 +167,6 @@ ifeq ($(NO_SHARED), 1)
 	$(error OpenBLAS: neither static nor shared are enabled.)
 endif
 endif
-	@-ln -fs $(LIBNAME) $(LIBPREFIX).$(LIBSUFFIX)
 	@for d in $(SUBDIRS) ; \
 	do if test -d $$d; then \
 	  $(MAKE) -C $$d $(@F) || exit 1 ; \
@@ -196,6 +195,7 @@ endif
 ifdef USE_THREAD
 	@echo USE_THREAD=$(USE_THREAD) >>  Makefile.conf_last
 endif
+	@-ln -fs $(LIBNAME) $(LIBPREFIX).$(LIBSUFFIX)
 	@touch lib.grd
 
 prof : prof_blas prof_lapack

Makefile.rule

@@ -3,7 +3,7 @@
 #
 
 # This library's version
-VERSION = 0.3.14
+VERSION = 0.3.14.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.x86

@@ -1,10 +1,21 @@
 # COMPILER_PREFIX = mingw32-
 
+ifndef DYNAMIC_ARCH
+ADD_CPUFLAGS = 1
+else
+ifdef TARGET_CORE
+ADD_CPUFLAGS = 1
+endif
+endif
+
+ifdef ADD_CPUFLAGS
 ifdef HAVE_SSE
 CCOMMON_OPT += -msse
+ifneq ($(F_COMPILER), NAG)
 FCOMMON_OPT += -msse
 endif
-
+endif
+endif
 
 ifeq ($(OSNAME), Interix)
 ARFLAGS		= -m x86

Makefile.x86_64

@@ -8,6 +8,16 @@ endif
 endif
 endif
 
+
+ifndef DYNAMIC_ARCH
+ADD_CPUFLAGS = 1
+else
+ifdef TARGET_CORE
+ADD_CPUFLAGS = 1
+endif
+endif
+
+ifdef ADD_CPUFLAGS
 ifdef HAVE_SSE3
 CCOMMON_OPT += -msse3
 ifneq ($(F_COMPILER), NAG)
@@ -44,7 +54,6 @@ endif
 endif
 
 ifeq ($(CORE), SKYLAKEX)
-ifndef DYNAMIC_ARCH
 ifndef NO_AVX512
 CCOMMON_OPT += -march=skylake-avx512
 ifneq ($(F_COMPILER), NAG)
@@ -62,10 +71,8 @@ endif
 endif
 endif
 endif
-endif
 
 ifeq ($(CORE), COOPERLAKE)
-ifndef DYNAMIC_ARCH
 ifndef NO_AVX512
 ifeq ($(C_COMPILER), GCC)
 # cooperlake support was added in 10.1
@@ -88,7 +95,6 @@ endif
 endif
 endif
 endif
-endif
 
 ifdef HAVE_AVX2
 ifndef NO_AVX2
@@ -120,6 +126,7 @@ endif
 endif
 endif
 
+endif
 
 
 ifeq ($(OSNAME), Interix)

azure-pipelines.yml

@@ -4,7 +4,15 @@ trigger:
   branches:
     include:
       - develop
-
+resources:
+  containers:
+      - container: oneapi-hpckit
+        image: intel/oneapi-hpckit:latest
+        options: '-v /usr/bin/sudo:/usr/bin/sudo -v /usr/lib/sudo/libsudo_util.so.0:/usr/lib/sudo/libsudo_util.so.0 -v /usr/lib/sudo/sudoers.so:/usr/lib/sudo/sudoers.so'
+      - container: oneapi-basekit
+        image: intel/oneapi-basekit:latest
+        options: '-v /usr/bin/sudo:/usr/bin/sudo -v /usr/lib/sudo/libsudo_util.so.0:/usr/lib/sudo/libsudo_util.so.0 -v /usr/lib/sudo/sudoers.so:/usr/lib/sudo/sudoers.so'
+ 
 jobs:
 # manylinux1 is useful to test because the
 # standard Docker container uses an old version
@@ -68,4 +76,64 @@ jobs:
       dir
       openblas_utest.exe
   
-      
+- job: OSX_OpenMP
+  pool:
+     vmImage: 'macOS-10.15'
+  steps:   
+  - script: |
+      brew update
+      make TARGET=CORE2 DYNAMIC_ARCH=1 USE_OPENMP=1 INTERFACE64=1 CC=gcc-10 FC=gfortran-10
+     
+- job: OSX_GCC_Nothreads
+  pool:
+     vmImage: 'macOS-10.15'
+  steps:   
+  - script: |
+      brew update
+      make USE_THREADS=0 CC=gcc-10 FC=gfortran-10
+     
+- job: OSX_OpenMP_Clang
+  pool:
+     vmImage: 'macOS-10.15'
+  variables:
+     LD_LIBRARY_PATH: /usr/local/opt/llvm/lib
+     LIBRARY_PATH: /usr/local/opt/llvm/lib
+  steps:   
+  - script: |
+      brew update
+      brew install llvm libomp
+      make TARGET=CORE2 USE_OPENMP=1 INTERFACE64=1 DYNAMIC_ARCH=1 CC=/usr/local/opt/llvm/bin/clang FC=gfortran-10
+
+- job: OSX_Ifort_Clang
+  pool:
+     vmImage: 'macOS-10.15'
+  variables:
+     LD_LIBRARY_PATH: /usr/local/opt/llvm/lib
+     MACOS_HPCKIT_URL: https://registrationcenter-download.intel.com/akdlm/irc_nas/17643/m_HPCKit_p_2021.2.0.2903_offline.dmg
+     LIBRARY_PATH: /usr/local/opt/llvm/lib
+     MACOS_FORTRAN_COMPONENTS: intel.oneapi.mac.ifort-compiler
+  steps:   
+  - script: |
+      brew update
+      brew install llvm libomp
+      sudo mkdir -p /opt/intel
+      sudo chown $USER /opt/intel
+    displayName: prepare for cache restore
+  - task: Cache@2
+    inputs:
+      path: /opt/intel/oneapi
+      key: '"install" | "$(MACOS_HPCKIT_URL)" | "$(MACOS_FORTRAN_COMPONENTS)"'
+      cacheHitVar: CACHE_RESTORED
+  - script: | 
+      curl --output webimage.dmg --url $(MACOS_HPCKIT_URL) --retry 5 --retry-delay 5
+      hdiutil attach webimage.dmg
+      sudo /Volumes/"$(basename "$(MACOS_HPCKIT_URL)" .dmg)"/bootstrapper.app/Contents/MacOS/bootstrapper -s --action install --components="$(MACOS_FORTRAN_COMPONENTS)" --eula=accept --continue-with-optional-error=yes --log-dir=.
+      installer_exit_code=$?
+      hdiutil detach /Volumes/"$(basename "$URL" .dmg)" -quiet
+      exit $installer_exit_code
+    displayName: install
+    condition: ne(variables.CACHE_RESTORED, 'true')
+  - script: | 
+      source /opt/intel/oneapi/setvars.sh
+      make CC=/usr/local/opt/llvm/bin/clang FC=ifort
+ 

benchmark/bench.h

@@ -3,6 +3,8 @@
 #include <time.h>
 #ifdef __CYGWIN32__
 #include <sys/time.h>
+#elif defined(__APPLE__)
+#include <mach/mach_time.h>
 #endif
 #include "common.h"
 

cmake/lapack.cmake

@@ -66,7 +66,7 @@ set(SLASRC
    slaqgb.f slaqge.f slaqp2.f slaqps.f slaqsb.f slaqsp.f slaqsy.f
    slaqr0.f slaqr1.f slaqr2.f slaqr3.f slaqr4.f slaqr5.f
    slaqtr.f slar1v.f slar2v.f ilaslr.f ilaslc.f
-   slarf.f  slarfb.f slarfg.f slarfgp.f slarft.f slarfx.f slarfy.f slargv.f
+   slarf.f  slarfb.f slarfb_gett.f slarfg.f slarfgp.f slarft.f slarfx.f slarfy.f slargv.f
    slarrv.f slartv.f
    slarz.f  slarzb.f slarzt.f slasy2.f
    slasyf.f slasyf_rook.f slasyf_rk.f slasyf_aa.f
@@ -112,14 +112,14 @@ set(SLASRC
    sgeqrt.f sgeqrt2.f sgeqrt3.f sgemqrt.f
    stpqrt.f stpqrt2.f stpmqrt.f stprfb.f
    sgelqt.f sgelqt3.f sgemlqt.f
-   sgetsls.f sgeqr.f slatsqr.f slamtsqr.f sgemqr.f
+   sgetsls.f sgetsqrhrt.f sgeqr.f slatsqr.f slamtsqr.f sgemqr.f
    sgelq.f slaswlq.f slamswlq.f sgemlq.f
    stplqt.f stplqt2.f stpmlqt.f
    ssytrd_2stage.f ssytrd_sy2sb.f ssytrd_sb2st.F ssb2st_kernels.f
    ssyevd_2stage.f ssyev_2stage.f ssyevx_2stage.f ssyevr_2stage.f
    ssbev_2stage.f ssbevx_2stage.f ssbevd_2stage.f ssygv_2stage.f
    sgesvdq.f slaorhr_col_getrfnp.f
-   slaorhr_col_getrfnp2.f sorgtsqr.f sorhr_col.f )
+   slaorhr_col_getrfnp2.f sorgtsqr.f sorgtsqr_row.f sorhr_col.f )
 
 set(SXLASRC sgesvxx.f sgerfsx.f sla_gerfsx_extended.f sla_geamv.f
    sla_gercond.f sla_gerpvgrw.f ssysvxx.f ssyrfsx.f
@@ -171,7 +171,7 @@ set(CLASRC
    claqhb.f claqhe.f claqhp.f claqp2.f claqps.f claqsb.f
    claqr0.f claqr1.f claqr2.f claqr3.f claqr4.f claqr5.f
    claqsp.f claqsy.f clar1v.f clar2v.f ilaclr.f ilaclc.f
-   clarf.f  clarfb.f clarfg.f clarfgp.f clarft.f
+   clarf.f  clarfb.f clarfb_gett.f clarfg.f clarfgp.f clarft.f
    clarfx.f clarfy.f clargv.f clarnv.f clarrv.f clartg.f clartv.f
    clarz.f  clarzb.f clarzt.f clascl.f claset.f clasr.f  classq.f
    clasyf.f clasyf_rook.f clasyf_rk.f clasyf_aa.f
@@ -209,14 +209,14 @@ set(CLASRC
    cgeqrt.f cgeqrt2.f cgeqrt3.f cgemqrt.f
    ctpqrt.f ctpqrt2.f ctpmqrt.f ctprfb.f
    cgelqt.f cgelqt3.f cgemlqt.f
-   cgetsls.f cgeqr.f clatsqr.f clamtsqr.f cgemqr.f
+   cgetsls.f cgetsqrhrt.f cgeqr.f clatsqr.f clamtsqr.f cgemqr.f
    cgelq.f claswlq.f clamswlq.f cgemlq.f
    ctplqt.f ctplqt2.f ctpmlqt.f
    chetrd_2stage.f chetrd_he2hb.f chetrd_hb2st.F chb2st_kernels.f
    cheevd_2stage.f cheev_2stage.f cheevx_2stage.f cheevr_2stage.f
    chbev_2stage.f chbevx_2stage.f chbevd_2stage.f chegv_2stage.f
    cgesvdq.f claunhr_col_getrfnp.f claunhr_col_getrfnp2.f 
-   cungtsqr.f cunhr_col.f )
+   cungtsqr.f cungtsqr_row.f cunhr_col.f )
 
 set(CXLASRC cgesvxx.f cgerfsx.f cla_gerfsx_extended.f cla_geamv.f
    cla_gercond_c.f cla_gercond_x.f cla_gerpvgrw.f
@@ -253,7 +253,7 @@ set(DLASRC
    dlaqgb.f dlaqge.f dlaqp2.f dlaqps.f dlaqsb.f dlaqsp.f dlaqsy.f
    dlaqr0.f dlaqr1.f dlaqr2.f dlaqr3.f dlaqr4.f dlaqr5.f
    dlaqtr.f dlar1v.f dlar2v.f iladlr.f iladlc.f
-   dlarf.f  dlarfb.f dlarfg.f dlarfgp.f dlarft.f dlarfx.f dlarfy.f
+   dlarf.f  dlarfb.f dlarfb_gett.f dlarfg.f dlarfgp.f dlarft.f dlarfx.f dlarfy.f
    dlargv.f dlarrv.f dlartv.f
    dlarz.f  dlarzb.f dlarzt.f dlasy2.f
    dlasyf.f dlasyf_rook.f dlasyf_rk.f dlasyf_aa.f
@@ -300,14 +300,14 @@ set(DLASRC
    dgeqrt.f dgeqrt2.f dgeqrt3.f dgemqrt.f
    dtpqrt.f dtpqrt2.f dtpmqrt.f dtprfb.f
    dgelqt.f dgelqt3.f dgemlqt.f
-   dgetsls.f dgeqr.f dlatsqr.f dlamtsqr.f dgemqr.f
+   dgetsls.f dgetsqrhrt.f dgeqr.f dlatsqr.f dlamtsqr.f dgemqr.f
    dgelq.f dlaswlq.f dlamswlq.f dgemlq.f
    dtplqt.f dtplqt2.f dtpmlqt.f
    dsytrd_2stage.f dsytrd_sy2sb.f dsytrd_sb2st.F dsb2st_kernels.f
    dsyevd_2stage.f dsyev_2stage.f dsyevx_2stage.f dsyevr_2stage.f
    dsbev_2stage.f dsbevx_2stage.f dsbevd_2stage.f dsygv_2stage.f
    dcombssq.f dgesvdq.f dlaorhr_col_getrfnp.f
-   dlaorhr_col_getrfnp2.f dorgtsqr.f dorhr_col.f )
+   dlaorhr_col_getrfnp2.f dorgtsqr.f dorgtsqr_row.f dorhr_col.f )
 
 set(DXLASRC dgesvxx.f dgerfsx.f dla_gerfsx_extended.f dla_geamv.f
    dla_gercond.f dla_gerpvgrw.f dsysvxx.f dsyrfsx.f
@@ -360,7 +360,7 @@ set(ZLASRC
    zlaqhb.f zlaqhe.f zlaqhp.f zlaqp2.f zlaqps.f zlaqsb.f
    zlaqr0.f zlaqr1.f zlaqr2.f zlaqr3.f zlaqr4.f zlaqr5.f
    zlaqsp.f zlaqsy.f zlar1v.f zlar2v.f ilazlr.f ilazlc.f
-   zlarcm.f zlarf.f  zlarfb.f
+   zlarcm.f zlarf.f  zlarfb.f zlarfb_gett.f
    zlarfg.f zlarfgp.f zlarft.f
    zlarfx.f zlarfy.f zlargv.f zlarnv.f zlarrv.f zlartg.f zlartv.f
    zlarz.f  zlarzb.f zlarzt.f zlascl.f zlaset.f zlasr.f
@@ -402,13 +402,13 @@ set(ZLASRC
    ztpqrt.f ztpqrt2.f ztpmqrt.f ztprfb.f
    ztplqt.f ztplqt2.f ztpmlqt.f
    zgelqt.f zgelqt3.f zgemlqt.f
-   zgetsls.f zgeqr.f zlatsqr.f zlamtsqr.f zgemqr.f
+   zgetsls.f zgetsqrhrt.f zgeqr.f zlatsqr.f zlamtsqr.f zgemqr.f
    zgelq.f zlaswlq.f zlamswlq.f zgemlq.f
    zhetrd_2stage.f zhetrd_he2hb.f zhetrd_hb2st.F zhb2st_kernels.f
    zheevd_2stage.f zheev_2stage.f zheevx_2stage.f zheevr_2stage.f
    zhbev_2stage.f zhbevx_2stage.f zhbevd_2stage.f zhegv_2stage.f
    zgesvdq.f zlaunhr_col_getrfnp.f zlaunhr_col_getrfnp2.f
-   zungtsqr.f zunhr_col.f)
+   zungtsqr.f zungtsqr_row.f zunhr_col.f)
 
 set(ZXLASRC zgesvxx.f zgerfsx.f zla_gerfsx_extended.f zla_geamv.f
    zla_gercond_c.f zla_gercond_x.f zla_gerpvgrw.f zsysvxx.f zsyrfsx.f