Merge pull request #4 from xianyi/develop

Update branch

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 1.dev)
+set(OpenBLAS_PATCH_VERSION 4.dev)
 set(OpenBLAS_VERSION "${OpenBLAS_MAJOR_VERSION}.${OpenBLAS_MINOR_VERSION}.${OpenBLAS_PATCH_VERSION}")
 
 # Adhere to GNU filesystem layout conventions
@@ -150,6 +150,7 @@ 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>)
 
 # Android needs to explicitly link against libm
 if(ANDROID)
@@ -169,6 +170,7 @@ endif()
 # Set output for libopenblas
 set_target_properties( ${OpenBLAS_LIBNAME} PROPERTIES RUNTIME_OUTPUT_DIRECTORY ${PROJECT_BINARY_DIR}/lib)
 set_target_properties( ${OpenBLAS_LIBNAME} PROPERTIES LIBRARY_OUTPUT_NAME_DEBUG "${OpenBLAS_LIBNAME}_d")
+set_target_properties( ${OpenBLAS_LIBNAME} PROPERTIES EXPORT_NAME "OpenBLAS")
 
 foreach (OUTPUTCONFIG ${CMAKE_CONFIGURATION_TYPES})
   string( TOUPPER ${OUTPUTCONFIG} OUTPUTCONFIG )

Changelog.txt

@@ -1,4 +1,142 @@
 OpenBLAS ChangeLog
+====================================================================
+Version 0.3.3
+31-Aug-2018
+
+common:
+	* thread memory allocation has been switched back to the method
+	  used before version 0.3.1 due to unexpected problems caused by
+	  the new code under some circumstances. A new compile-time option
+	  USE_TLS has been added to enable the new code, and it is hoped
+	  that this can become the default again in the next version.
+	* LAPAck PR272 has been integrated, which fixes spurious errors
+	  in DSYEVR and related functions caused by missing conversion 
+	  from ILAENV to ILAENV_2STAGE in several _2stage routines.
+	* the cmake-generated OpenBLASConfig.cmake now uses correct case
+	  for the name of the library
+	* added support for Haiku OS  
+
+x86_64:
+	* added AVX512 implementations of SDOT, DDOT, SAXPY, DAXPY,
+	  DSCAL, DGEMVN and DSYMVL 
+	* added a workaround for a cygwin issue that prevented compilation
+	  of AVX512 code
+
+IBM Z:
+	* added autodetection of Z14
+	* fixed TRMM errors in the generic target
+
+====================================================================
+Version 0.3.2
+30-Jul-2018
+
+common:
+	* fixes for regressions caused by the rewrite of the thread
+	  initialization code in 0.3.1
+
+POWER:
+	* fixed cpu autodetection for the BSDs
+
+MIPS64:
+	* fixed utest errors in AXPY, DSDOT, ROT and SWAP
+
+x86_64:
+	* added autodetection of AMD Ryzen 2
+	* fixed build with older versions of MSVC
+
+====================================================================
+Version 0.3.1
+01-Jul-2018
+
+common:
+	* rewritten thread initialization code with significantly reduced overhead
+	* added CBLAS interfaces to the IxAMIN BLAS extension functions
+	* fixed the lapack-test target
+	* CMAKE builds now create an OpenBLASConfig.cmake file
+	* ZAXPY now uses a single thread for small input sizes
+	* the LAPACK code was updated from Reference-LAPACK/lapack#253
+          (fixing LAPACKE interfaces to Aasen's functions)
+
+POWER:
+	* corrected CROT and ZROT behaviour with zero INC_X
+
+ARMV7:
+	* corrected xDOT behaviour with zero INC_X or INC_Y
+
+x86_64:
+	* retired some older targets of DYNAMIC_ARCH builds to a new option DYNAMIC_OLDER,
+	  this affects PENRYN,DUNNINGTON,OPTERON,OPTERON_SSE3,BOBCAT,ATOM and NANO
+	  (which will still be supported via the slower PRESCOTT kernels when this option is not set)
+	* added an option DYNAMIC_LIST that (used in conjunction with DYNAMIC_ARCH) allows to 
+	  specify the list of x86_64 targets to include. Any target not on the list will be supported 
+	  by the Sandybridge or Nehalem kernels if available, or by Prescott.
+	* improved SWITCH_RATIO on Haswell for increased GEMM throughput
+	* added initial support for Intel Skylake X, including an AVX512 SGEMM kernel
+	* added autodetection of Intel Cannon Lake series as Skylake X
+	* added a default L2 cache size for hypervisors that return zero here (Chromebook)
+	* fixed a name clash with recent Windows10 headers that broke the build with (at least)
+	  recent mingw from MSYS2
+	* fixed a link error in mixed clang/gfortran builds with OpenMP
+	* updated the OSX deployment target to 10.8
+	* switched on parallel make for builds on MS Windows by default
+
+x86:
+	* fixed SSWAP and DSWAP behaviour with zero INC_X and INC_Y
+
+====================================================================
+Version 0.3.0
+23-May-2108
+
+common:
+	* fixed some more thread race and locking bugs
+	* added preliminary support for calling an OpenMP build of the library from multiple threads
+	* removed performance impact of thread locks added in 0.2.20 on OpenMP code
+	* general code cleanup 
+	* optimized DSDOT implementation
+	* improved thread distribution for GEMM
+	* corrected IMATCOPY/OMATCOPY implementation
+	* fixed out-of-bounds accesses in the multithreaded xBMV/xPMV and SYMV implementations
+	* cmake build improvements
+	* pkgconfig file now contains build options
+	* openblas_get_config() now reports USE_OPENMP and NUM_THREADS settings used for the build
+	* corrections and improvements for systems with more than 64 cpus
+	* LAPACK code updated to 3.8.0 including later fixes
+	* added ReLAPACK, a recursive implementation of several LAPACK functions
+	* Rewrote ROTMG to handle cases that the netlib code failed to address
+	* Disabled (broken) multithreading code for xTRMV
+	* corrected prototypes of complex CBLAS functions to make our cblas.h match the generally accepted standard
+	* shared memory access failures on startup are now handled more gracefully
+	* restored utests from earlier releases (and made them pass on all affected systems)
+
+SPARC:
+	* several fixes for cpu autodetection
+
+POWER:
+	* corrected vector register overwriting in several Power8 kernels
+	* optimized additional BLAS functions
+
+ARM:
+	* added support for CortexA53 and A72 
+	* added autodetection for ThunderX2T99
+	* made most optimized kernels the default for generic ARMv8 targets 
+
+x86_64:
+	* parallelized DDOT kernel for Haswell
+	* changed alignment directives in assembly kernels to boost performance on OSX
+	* fixed register handling in the GEMV microkernels (bug exposed by gcc7)
+	* added support for building on OpenBSD and Dragonfly 
+	* updated compiler options to work with Intel release 2018
+	* support fully optimized build with clang/flang on Microsoft Windows
+	* fixed building on AIX
+
+IBM Z:
+	* added optimized BLAS 1/2 functions
+
+MIPS:
+	* fixed cpu autodetection helper code
+	* added mips32 1004K cpu (Mediatek MT7621 and similar SoC)
+	* added mips64 I6500 cpu
+
 ====================================================================
 Version 0.2.20
 24-Jul-2017

Makefile

@@ -97,7 +97,7 @@ endif
 
 shared :
 ifndef NO_SHARED
-ifeq ($(OSNAME), $(filter $(OSNAME),Linux SunOS Android))
+ifeq ($(OSNAME), $(filter $(OSNAME),Linux SunOS Android Haiku))
 	@$(MAKE) -C exports so
 	@ln -fs $(LIBSONAME) $(LIBPREFIX).so
 	@ln -fs $(LIBSONAME) $(LIBPREFIX).so.$(MAJOR_VERSION)
@@ -267,6 +267,8 @@ ifeq ($(F_COMPILER), GFORTRAN)
 ifdef SMP
 ifeq ($(OSNAME), WINNT)
 	-@echo "LOADER      = $(FC)" >> $(NETLIB_LAPACK_DIR)/make.inc
+else ifeq ($(OSNAME), Haiku)
+	-@echo "LOADER      = $(FC)" >> $(NETLIB_LAPACK_DIR)/make.inc
 else
 	-@echo "LOADER      = $(FC) -pthread" >> $(NETLIB_LAPACK_DIR)/make.inc
 endif

Makefile.install

@@ -66,7 +66,7 @@ endif
 #for install shared library
 ifndef NO_SHARED
 	@echo Copying the shared library to $(DESTDIR)$(OPENBLAS_LIBRARY_DIR)
-ifeq ($(OSNAME), $(filter $(OSNAME),Linux SunOS Android))
+ifeq ($(OSNAME), $(filter $(OSNAME),Linux SunOS Android Haiku))
 	@install -pm755 $(LIBSONAME) "$(DESTDIR)$(OPENBLAS_LIBRARY_DIR)"
 	@cd "$(DESTDIR)$(OPENBLAS_LIBRARY_DIR)" ; \
 	ln -fs $(LIBSONAME) $(LIBPREFIX).so ; \

Makefile.rule

@@ -3,7 +3,7 @@
 #
 
 # This library's version
-VERSION = 0.3.1.dev
+VERSION = 0.3.4.dev
 
 # If you set the suffix, the library name will be libopenblas_$(LIBNAMESUFFIX).a
 # and libopenblas_$(LIBNAMESUFFIX).so. Meanwhile, the soname in shared library
@@ -107,7 +107,13 @@ BUILD_LAPACK_DEPRECATED = 1
 # BUILD_RELAPACK = 1
 
 # If you want to use legacy threaded Level 3 implementation.
-# USE_SIMPLE_THREADED_LEVEL3 = 1
+USE_SIMPLE_THREADED_LEVEL3 = 1
+
+# If you want to use the new, still somewhat experimental code that uses
+# thread-local storage instead of a central memory buffer in memory.c
+# Note that if your system uses GLIBC, it needs to have at least glibc 2.21
+# for this to work.
+USE_TLS = 1
 
 # If you want to drive whole 64bit region by BLAS. Not all Fortran
 # compiler supports this. It's safe to keep comment it out if you

Makefile.system

@@ -1018,6 +1018,10 @@ ifdef USE_SIMPLE_THREADED_LEVEL3
 CCOMMON_OPT	+= -DUSE_SIMPLE_THREADED_LEVEL3
 endif
 
+ifdef USE_TLS
+CCOMMON_OPT += -DUSE_TLS
+endif
+
 ifndef SYMBOLPREFIX
 SYMBOLPREFIX =
 endif

Makefile.x86_64

@@ -12,6 +12,9 @@ ifeq ($(CORE), SKYLAKEX)
 ifndef NO_AVX512
 CCOMMON_OPT += -march=skylake-avx512
 FCOMMON_OPT += -march=skylake-avx512
+ifeq ($(OSNAME), CYGWIN_NT)
+CCOMMON_OPT += -fno-asynchronous-unwind-tables
+endif
 endif
 endif
 

README.md

@@ -110,6 +110,7 @@ Please read `GotoBLAS_01Readme.txt`.
 - **Intel Xeon 56xx (Westmere)**: Used GotoBLAS2 Nehalem codes.
 - **Intel Sandy Bridge**: Optimized Level-3 and Level-2 BLAS with AVX on x86-64.
 - **Intel Haswell**: Optimized Level-3 and Level-2 BLAS with AVX2 and FMA  on x86-64.
+- **Intel Skylake**: Optimized Level-3 and Level-2 BLAS with AVX512 and FMA  on x86-64.
 - **AMD Bobcat**: Used GotoBLAS2 Barcelona codes.
 - **AMD Bulldozer**: x86-64 ?GEMM FMA4 kernels. (Thanks to Werner Saar)
 - **AMD PILEDRIVER**: Uses Bulldozer codes with some optimizations.
@@ -200,6 +201,7 @@ Please see Changelog.txt to view the differences between OpenBLAS and GotoBLAS2
 * 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 Skyalke AVX512 kernels.
 * The number of CPUs/cores should 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`.

benchmark/gemv.c

@@ -122,7 +122,7 @@ int main(int argc, char *argv[]){
 
   FLOAT *a, *x, *y;
   FLOAT alpha[] = {1.0, 1.0};
-  FLOAT beta [] = {1.0, 1.0};
+  FLOAT beta [] = {1.0, 0.0};
   char trans='N';
   blasint m, i, j;
   blasint inc_x=1,inc_y=1;

c_check

@@ -64,6 +64,7 @@ $os = WINNT           if ($data =~ /OS_WINNT/);
 $os = CYGWIN_NT       if ($data =~ /OS_CYGWIN_NT/);
 $os = Interix         if ($data =~ /OS_INTERIX/);
 $os = Android         if ($data =~ /OS_ANDROID/);
+$os = Haiku           if ($data =~ /OS_HAIKU/);
 
 $architecture = x86    if ($data =~ /ARCH_X86/);
 $architecture = x86_64 if ($data =~ /ARCH_X86_64/);
@@ -223,14 +224,15 @@ $data =~ /globl\s([_\.]*)(.*)/;
 $need_fu      = $1;
 
 $cross = 0;
-$cross = 1 if ($os ne $hostos);
 
 if ($architecture ne $hostarch) {
     $cross = 1;
     $cross = 0 if (($hostarch eq "x86_64") && ($architecture eq "x86"));
     $cross = 0 if (($hostarch eq "mips64") && ($architecture eq "mips"));
 }
 
+$cross = 1 if ($os ne $hostos);
+
 $openmp = "" if $ENV{USE_OPENMP} != 1;
 
 $linker_L = "";

cblas.h

@@ -51,7 +51,8 @@ typedef enum CBLAS_TRANSPOSE {CblasNoTrans=111, CblasTrans=112, CblasConjTrans=1
 typedef enum CBLAS_UPLO      {CblasUpper=121, CblasLower=122} CBLAS_UPLO;
 typedef enum CBLAS_DIAG      {CblasNonUnit=131, CblasUnit=132} CBLAS_DIAG;
 typedef enum CBLAS_SIDE      {CblasLeft=141, CblasRight=142} CBLAS_SIDE;
-
+typedef CBLAS_ORDER CBLAS_LAYOUT;
+
 float  cblas_sdsdot(OPENBLAS_CONST blasint n, OPENBLAS_CONST float alpha, OPENBLAS_CONST float *x, OPENBLAS_CONST blasint incx, OPENBLAS_CONST float *y, OPENBLAS_CONST blasint incy);
 double cblas_dsdot (OPENBLAS_CONST blasint n, OPENBLAS_CONST float *x, OPENBLAS_CONST blasint incx, OPENBLAS_CONST float *y, OPENBLAS_CONST blasint incy);
 float  cblas_sdot(OPENBLAS_CONST blasint n, OPENBLAS_CONST float  *x, OPENBLAS_CONST blasint incx, OPENBLAS_CONST float  *y, OPENBLAS_CONST blasint incy);

cmake/prebuild.cmake

@@ -85,7 +85,7 @@ if (NOT NOFORTRAN)
 endif ()
 
 # Cannot run getarch on target if we are cross-compiling
-if (DEFINED CORE AND CMAKE_CROSSCOMPILING)
+if (DEFINED CORE AND CMAKE_CROSSCOMPILING AND NOT (${HOST_OS} STREQUAL "WINDOWSSTORE"))
   # Write to config as getarch would
 
   # TODO: Set up defines that getarch sets up based on every other target

cmake/system.cmake

@@ -214,6 +214,10 @@ if (CONSISTENT_FPCSR)
   set(CCOMMON_OPT "${CCOMMON_OPT} -DCONSISTENT_FPCSR")
 endif ()
 
+if (USE_TLS)
+  set(CCOMMON_OPT "${CCOMMON_OPT} -DUSE_TLS")
+endif ()
+
 # Only for development
 # set(CCOMMON_OPT "${CCOMMON_OPT} -DPARAMTEST")
 # set(CCOMMON_OPT "${CCOMMON_OPT} -DPREFETCHTEST")

cmake/system_check.cmake

@@ -68,7 +68,7 @@ endif()
 
 if (X86_64 OR X86)
   file(WRITE ${PROJECT_BINARY_DIR}/avx512.tmp "int main(void){ __asm__ volatile(\"vbroadcastss -4 * 4(%rsi), %zmm2\"); }")
-execute_process(COMMAND ${CMAKE_C_COMPILER} -v -o ${PROJECT_BINARY_DIR}/avx512.o -x c ${PROJECT_BINARY_DIR}/avx512.tmp RESULT_VARIABLE NO_AVX512)
+execute_process(COMMAND ${CMAKE_C_COMPILER} -march=skylake-avx512 -v -o ${PROJECT_BINARY_DIR}/avx512.o -x c ${PROJECT_BINARY_DIR}/avx512.tmp OUTPUT_QUIET ERROR_QUIET RESULT_VARIABLE NO_AVX512)
 if (NO_AVX512 EQUAL 1)
 set (CCOMMON_OPT "${CCOMMON_OPT} -DNO_AVX512")
 endif()

common.h

@@ -105,6 +105,10 @@ extern "C" {
 #endif
 #endif
 
+#ifdef OS_HAIKU
+#define NO_SYSV_IPC
+#endif
+
 #ifdef OS_WINDOWS
 #ifdef  ATOM
 #define GOTO_ATOM ATOM
@@ -253,8 +257,14 @@ typedef unsigned long BLASULONG;
 
 #ifdef USE64BITINT
 typedef BLASLONG blasint;
+#if defined(OS_WINDOWS) && defined(__64BIT__)
+#define blasabs(x) llabs(x)
+#else
+#define blasabs(x) labs(x)
+#endif
 #else
 typedef int blasint;
+#define blasabs(x) abs(x)
 #endif
 #else
 #ifdef USE64BITINT