Merge pull request #2100 from xianyi/develop

Merge develop in preparation of 0.3.6 release

Author: martin-frbg

.travis.yml

@@ -149,7 +149,7 @@ matrix:
 
     - &test-macos
       os: osx
-      osx_image: xcode8
+      osx_image: xcode10.1
       before_script:
         - COMMON_FLAGS="DYNAMIC_ARCH=1 TARGET=NEHALEM NUM_THREADS=32"
         - brew update
@@ -160,6 +160,7 @@ matrix:
         - BTYPE="BINARY=64 INTERFACE64=1"
 
     - <<: *test-macos
+      osx_image: xcode8.3
       env:
         - BTYPE="BINARY=32"
 

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 5)
+set(OpenBLAS_PATCH_VERSION 6)
 set(OpenBLAS_VERSION "${OpenBLAS_MAJOR_VERSION}.${OpenBLAS_MINOR_VERSION}.${OpenBLAS_PATCH_VERSION}")
 
 # Adhere to GNU filesystem layout conventions
@@ -42,6 +42,19 @@ endif()
 
 #######
 
+if(MSVC AND MSVC_STATIC_CRT)
+    set(CompilerFlags
+            CMAKE_CXX_FLAGS
+            CMAKE_CXX_FLAGS_DEBUG
+            CMAKE_CXX_FLAGS_RELEASE
+            CMAKE_C_FLAGS
+            CMAKE_C_FLAGS_DEBUG
+            CMAKE_C_FLAGS_RELEASE
+            )
+    foreach(CompilerFlag ${CompilerFlags})
+      string(REPLACE "/MD" "/MT" ${CompilerFlag} "${${CompilerFlag}}")
+    endforeach()
+endif()
 
 message(WARNING "CMake support is experimental. It does not yet support all build options and may not produce the same Makefiles that OpenBLAS ships with.")
 
@@ -62,10 +75,10 @@ endif ()
 
 set(SUBDIRS	${BLASDIRS})
 if (NOT NO_LAPACK)
-  list(APPEND SUBDIRS lapack)
   if(BUILD_RELAPACK)
     list(APPEND SUBDIRS relapack/src)
   endif()
+  list(APPEND SUBDIRS lapack)
 endif ()
 
 # set which float types we want to build for
@@ -134,7 +147,7 @@ endif ()
 
 # Only generate .def for dll on MSVC and always produce pdb files for debug and release
 if(MSVC)
-  if (${CMAKE_MAJOR_VERSION}.${CMAKE_MINOR_VERSION} LESS 3.4)
+  if (${CMAKE_MAJOR_VERSION}.${CMAKE_MINOR_VERSION} VERSION_LESS 3.4)
     set(OpenBLAS_DEF_FILE "${PROJECT_BINARY_DIR}/openblas.def")
   endif()
   set(CMAKE_C_FLAGS_RELEASE "${CMAKE_C_FLAGS_RELEASE} /Zi")
@@ -149,15 +162,9 @@ if (${DYNAMIC_ARCH})
   endforeach()
 endif ()
 
-# Only build shared libs for MSVC
-if (MSVC)
-  set(BUILD_SHARED_LIBS ON)
-endif()
-
-
 # add objects to the openblas lib
 add_library(${OpenBLAS_LIBNAME} ${LA_SOURCES} ${LAPACKE_SOURCES} ${RELA_SOURCES} ${TARGET_OBJS} ${OpenBLAS_DEF_FILE})
-target_include_directories(${OpenBLAS_LIBNAME} INTERFACE $<INSTALL_INTERFACE:include>)
+target_include_directories(${OpenBLAS_LIBNAME} INTERFACE $<INSTALL_INTERFACE:include/openblas${SUFFIX64}>)
 
 # Android needs to explicitly link against libm
 if(ANDROID)
@@ -166,7 +173,7 @@ endif()
 
 # Handle MSVC exports
 if(MSVC AND BUILD_SHARED_LIBS)
-  if (${CMAKE_MAJOR_VERSION}.${CMAKE_MINOR_VERSION} LESS 3.4)
+  if (${CMAKE_MAJOR_VERSION}.${CMAKE_MINOR_VERSION} VERSION_LESS 3.4)
     include("${PROJECT_SOURCE_DIR}/cmake/export.cmake")
   else()
     # Creates verbose .def file (51KB vs 18KB)
@@ -217,6 +224,14 @@ set_target_properties(${OpenBLAS_LIBNAME} PROPERTIES
   SOVERSION ${OpenBLAS_MAJOR_VERSION}
 )
 
+if (BUILD_SHARED_LIBS AND BUILD_RELAPACK)
+  if (NOT MSVC)
+    target_link_libraries(${OpenBLAS_LIBNAME} "-Wl,-allow-multiple-definition")
+  else()
+    target_link_libraries(${OpenBLAS_LIBNAME} "/FORCE:MULTIPLE")
+  endif()
+endif()
+
 if (BUILD_SHARED_LIBS AND NOT ${SYMBOLPREFIX}${SYMBOLSUFIX} STREQUAL "")
 if (NOT DEFINED ARCH)
   set(ARCH_IN "x86_64")
@@ -314,7 +329,7 @@ install (FILES ${OPENBLAS_CONFIG_H} DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
 if(NOT NOFORTRAN)
   message(STATUS "Generating f77blas.h in ${CMAKE_INSTALL_INCLUDEDIR}")
 
-  set(F77BLAS_H ${CMAKE_BINARY_DIR}/f77blas.h)
+  set(F77BLAS_H ${CMAKE_BINARY_DIR}/generated/f77blas.h)
   file(WRITE  ${F77BLAS_H} "#ifndef OPENBLAS_F77BLAS_H\n")
   file(APPEND ${F77BLAS_H} "#define OPENBLAS_F77BLAS_H\n")
   file(APPEND ${F77BLAS_H} "#include \"openblas_config.h\"\n")
@@ -327,10 +342,11 @@ endif()
 if(NOT NO_CBLAS)
 	message (STATUS "Generating cblas.h in ${CMAKE_INSTALL_INCLUDEDIR}")
 
+	set(CBLAS_H ${CMAKE_BINARY_DIR}/generated/cblas.h)
 	file(READ ${CMAKE_CURRENT_SOURCE_DIR}/cblas.h CBLAS_H_CONTENTS)
 	string(REPLACE "common" "openblas_config" CBLAS_H_CONTENTS_NEW "${CBLAS_H_CONTENTS}")
-	file(WRITE ${CMAKE_BINARY_DIR}/cblas.tmp "${CBLAS_H_CONTENTS_NEW}")
-	install (FILES ${CMAKE_BINARY_DIR}/cblas.tmp DESTINATION ${CMAKE_INSTALL_INCLUDEDIR} RENAME cblas.h)
+	file(WRITE ${CBLAS_H} "${CBLAS_H_CONTENTS_NEW}")
+	install (FILES ${CBLAS_H} DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
 endif()
 
 if(NOT NO_LAPACKE)

Changelog.txt

@@ -1,4 +1,82 @@
 OpenBLAS ChangeLog
+====================================================================
+Version 0.3.6
+29-Apr-2019
+
+common:
+	* the build tools now check that a given cpu TARGET is actually valid
+	* the build-time check of system features (c_check) has been made
+  	  less dependent on particular perl features (this should mainly
+  	  benefit building on Windows)
+	* several problem with the ReLAPACK integration were fixed,
+	  including INTERFACE64 support and building a shared library
+	* building with CMAKE on BSD systems was improved
+	* a non-absolute SUM function was added based on the
+  	  existing optimized code for ASUM
+	* CBLAS interfaces to the IxMIN and IxMAX functions were added
+	* a name clash between LAPACKE and BOOST headers was resolved
+	* CMAKE builds with OpenMP failed to include the appropriate getrf_parallel
+	  kernels
+	* a crash on thread (key) deletion with the USE_TLS=1 memory management
+	  option was fixed
+	* restored several earlier fixes, in particular for OpenMP performance,
+  	  building on BSD, and calling fork on CYGWIN, which had inadvertently
+  	  been dropped in the 0.3.3 rewrite of the memory management code.
+
+x86_64:
+	* the AVX512 DGEMM kernel has been disabled again due to unsolved problems
+	* building with old versions of MSVC was fixed
+	* it is now possible to build a static library on Windows with CMAKE
+	* accessing environment variables on CYGWIN at run time was fixed
+	* the CMAKE build system now recognizes 32bit userspace on 64bit hardware
+	* Intel "Denverton" atom and Hygon "Dhyana" zen CPUs are now autodetected
+	* building for DYNAMIC_ARCH with a DYNAMIC_LIST of targets is now supported
+  	  with CMAKE as well
+	* building for DYNAMIC_ARCH with GENERIC as the default target is now supported
+	* a buffer overflow in the SSE GEMM kernel for Intel Nano targets was fixed
+	* assembly bugs involving undeclared modification of input operands were fixed
+  	  in the AXPY, DOT, GEMV, GER, SCAL, SYMV and TRSM microkernels for Nehalem,
+	  Sandybridge, Haswell, Bulldozer and Piledriver. These would typically cause
+	  test failures or segfaults when compiled with recent versions of gcc from 8 onward.
+	* a similar bug was fixed in the blas_quickdivide code used to split workloads
+	  in most functions
+	* a bug in the IxMIN implementation for the GENERIC target made it return the result of IxMAX
+	* fixed building on SkylakeX systems when either the compiler or the (emulated) operating
+	  environment does not support AVX512
+	* improved GEMM performance on ZEN targets
+
+x86:
+	* build failures caused by the recently added checks for AVX512 were fixed
+	* an inline assembly bug involving undeclared modification of an input argument was
+  	  fixed in the blas_quickdivide code used to split workloads in most functions
+	* a bug in the IMIN implementation for the GENERIC target made it return the result of IMAX
+
+MIPS32:
+	* a bug in the IMIN implementation made it return the result of IMAX
+
+POWER:
+	* single precision BLAS1/2 functions have received optimized POWER8 kernels
+	* POWER9 is now a separate target, with an optimized DGEMM/DTRMM kernel
+	* building on PPC970 systems under OSX Leopard or Tiger is now supported
+	* out-of-bounds memory accesses in the gemm_beta microkernels were fixed
+	* building a shared library on AIX is now supported for POWER6
+	* DYNAMIC_ARCH support has been added for POWER6 and newer
+
+ARMv7:
+	* corrected xDOT behaviour with zero INC_X or INC_Y
+	* a bug in the IMIN implementation made it return the result of IMAX
+
+ARMv8:
+	* added support for HiSilicon TSV110 cpus
+	* the CMAKE build system now recognizes 32bit userspace on 64bit hardware
+	* cross-compilation with CMAKE now works again
+	* a bug in the IMIN implementation made it return the result of IMAX
+	* ARMV8 builds with the BINARY=32 option are now automatically handled as ARMV7
+
+IBM Z:
+	* optimized microkernels for single precicion BLAS1/2 functions have been added
+	  for both Z13 and Z14
+
 ====================================================================
 Version 0.3.5
 31-Dec-2018

Makefile

@@ -96,7 +96,7 @@ endif
 	@echo
 
 shared :
-ifndef NO_SHARED
+ifneq ($(NO_SHARED), 1)
 ifeq ($(OSNAME), $(filter $(OSNAME),Linux SunOS Android Haiku))
 	@$(MAKE) -C exports so
 	@ln -fs $(LIBSONAME) $(LIBPREFIX).so

Makefile.arm64

@@ -38,3 +38,8 @@ ifeq ($(CORE), THUNDERX2T99)
 CCOMMON_OPT += -march=armv8.1-a -mtune=thunderx2t99
 FCOMMON_OPT += -march=armv8.1-a -mtune=thunderx2t99
 endif
+
+ifeq ($(CORE), TSV110)
+CCOMMON_OPT += -march=armv8.2-a -mtune=tsv110
+FCOMMON_OPT += -march=armv8.2-a -mtune=tsv110
+endif

Makefile.install

@@ -58,14 +58,14 @@ ifndef NO_LAPACKE
 endif
 
 #for install static library
-ifndef NO_STATIC
+ifneq ($(NO_STATIC),1)
 	@echo Copying the static library to $(DESTDIR)$(OPENBLAS_LIBRARY_DIR)
 	@install -pm644 $(LIBNAME) "$(DESTDIR)$(OPENBLAS_LIBRARY_DIR)"
 	@cd "$(DESTDIR)$(OPENBLAS_LIBRARY_DIR)" ; \
 	ln -fs $(LIBNAME) $(LIBPREFIX).$(LIBSUFFIX)
 endif
 #for install shared library
-ifndef NO_SHARED
+ifneq ($(NO_SHARED),1)
 	@echo Copying the shared library to $(DESTDIR)$(OPENBLAS_LIBRARY_DIR)
 ifeq ($(OSNAME), $(filter $(OSNAME),Linux SunOS Android Haiku))
 	@install -pm755 $(LIBSONAME) "$(DESTDIR)$(OPENBLAS_LIBRARY_DIR)"
@@ -106,14 +106,14 @@ ifndef NO_LAPACKE
 endif
 
 #for install static library
-ifndef NO_STATIC
+ifneq ($(NO_STATIC),1)
 	@echo Copying the static library to $(DESTDIR)$(OPENBLAS_LIBRARY_DIR)
 	@installbsd -c -m 644 $(LIBNAME) "$(DESTDIR)$(OPENBLAS_LIBRARY_DIR)"
 	@cd "$(DESTDIR)$(OPENBLAS_LIBRARY_DIR)" ; \
 	ln -fs $(LIBNAME) $(LIBPREFIX).$(LIBSUFFIX)
 endif
 #for install shared library
-ifndef NO_SHARED
+ifneq ($(NO_SHARED),1)
 	@echo Copying the shared library to $(DESTDIR)$(OPENBLAS_LIBRARY_DIR)
 	@installbsd -c -m 755 $(LIBSONAME) "$(DESTDIR)$(OPENBLAS_LIBRARY_DIR)"
 	@cd "$(DESTDIR)$(OPENBLAS_LIBRARY_DIR)" ; \
@@ -138,7 +138,7 @@ endif
 	@echo "SET(OpenBLAS_VERSION \"${VERSION}\")" > "$(DESTDIR)$(OPENBLAS_CMAKE_DIR)/$(OPENBLAS_CMAKE_CONFIG)"
 	@echo "SET(OpenBLAS_INCLUDE_DIRS ${OPENBLAS_INCLUDE_DIR})" >> "$(DESTDIR)$(OPENBLAS_CMAKE_DIR)/$(OPENBLAS_CMAKE_CONFIG)"
 
-ifndef NO_SHARED
+ifneq ($(NO_SHARED),1)
 #ifeq logical or
 ifeq ($(OSNAME), $(filter $(OSNAME),Linux FreeBSD NetBSD OpenBSD DragonFly))
 	@echo "SET(OpenBLAS_LIBRARIES ${OPENBLAS_LIBRARY_DIR}/$(LIBPREFIX).so)" >> "$(DESTDIR)$(OPENBLAS_CMAKE_DIR)/$(OPENBLAS_CMAKE_CONFIG)"

Makefile.power

@@ -9,7 +9,15 @@ else
 USE_OPENMP = 1
 endif
 
-
+ifeq ($(CORE), POWER9)
+ifeq ($(USE_OPENMP), 1)
+COMMON_OPT += -Ofast -mcpu=power9 -mtune=power9 -mvsx -malign-power -DUSE_OPENMP -fno-fast-math -fopenmp
+FCOMMON_OPT += -O2 -frecursive -mcpu=power9 -mtune=power9 -malign-power -DUSE_OPENMP -fno-fast-math -fopenmp
+else
+COMMON_OPT += -Ofast -mcpu=power9 -mtune=power9 -mvsx -malign-power -fno-fast-math
+FCOMMON_OPT += -O2 -frecursive -mcpu=power9 -mtune=power9 -malign-power -fno-fast-math
+endif
+endif
 
 ifeq ($(CORE), POWER8)
 ifeq ($(USE_OPENMP), 1)