Merge branch 'develop' into fbsd12

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 4.dev)
+set(OpenBLAS_PATCH_VERSION 5.dev)
 set(OpenBLAS_VERSION "${OpenBLAS_MAJOR_VERSION}.${OpenBLAS_MINOR_VERSION}.${OpenBLAS_PATCH_VERSION}")
 
 # Adhere to GNU filesystem layout conventions

Changelog.txt

@@ -1,4 +1,77 @@
 OpenBLAS ChangeLog
+====================================================================
+Version 0.3.4
+02-Dec-2018
+
+common:
+	* the new, experimental thread-local memory allocation had 
+	  inadvertently been left enabled for gmake builds in 0.3.3
+	  despite the announcement. It is now disabled by default, and
+	  single-threaded builds will keep using the old allocator even
+	  if the USE_TLS option is turned on.
+	* OpenBLAS will now provide enough buffer space for at least 50
+	  threads by default.
+	* The output of openblas_get_config() now contains the version
+	  number.
+	* A serious thread safety bug in GEMV operation with small M and
+	  large N size has been fixed.
+	* The code will now automatically call blas_thread_init after a
+	  fork if needed before handling a call to openblas_set_num_threads
+	* Accesses to parallelized level3 functions from multiple callers
+	  are now serialized to avoid thread races (unless using OpenMP).
+	  This should provide better performance than the known-threadsafe
+	  (but non-default) USE_SIMPLE_THREADED_LEVEL3 option.
+	* When building LAPACK with gfortran, -frecursive is now (again)
+	  enabled by default to ensure correct behaviour.
+        * The OpenBLAS version cblas.h now supports both CBLAS_ORDER and
+	  CBLAS_LAYOUT as the name of the matrix row/column order option.
+	* Externally set LDFLAGS are now passed through to the final compile/link
+	  steps to facilitate setting platform-specific linker flags.
+	* A potential race condition during the build of LAPACK (that would 
+	  usually manifest itself as a failure to build TESTING/MATGEN) has been 
+	  fixed.
+	* xHEMV has been changed to stay single-threaded for small input sizes
+	  where the overhead of multithreading exceeds any possible gains
+	* CSWAP and ZSWAP have been limited to a single thread except on ARMV8 or
+	  ThunderX hardware with sizable input.
+	* Linker flags for the PGI compiler have been updated
+	* Behaviour of AXPY with zero increments is now handled in the C interface,
+	  correcting the result on at least Intel Atom.
+	* The result matrix from calling SGELSS with an all-zero input matrix is 
+	  now zeroed completely.
+	  
+x86_64:
+	* Autodetection of AMD Ryzen2 has been fixed (again).
+        * CMAKE builds now support labeling of an INTERFACE64=1 build of
+	  the library with the _64 suffix.
+	* AVX512 version of DGEMM has been added and the AVX512 SGEMM kernel
+	  has been sped up by rewriting with C intrinsics
+	* Fixed compilation on RHEL5/CENTOS5 (issue with typename __WAIT_STATUS)
+	
+POWER:
+	* added support for building on AIX (with gcc and GNU tools from AIX Toolbox).
+	* CPU type detection has been implemented for AIX.
+	* CPU type detection has been fixed for NETBSD.
+	
+MIPS64:
+	* AXPY on LOONGSON3A has been corrected to pass "zero increment" utest.
+	* DSDOT on LOONGSON3A has been fixed.
+	* the SGEMM microkernel has been hardened against potential data loss.
+	
+ARMV8:
+	* DYNAMic_ARCH support is now available for 64bit ARM
+	* cross-compiling for ARMV8 under iOS now works.
+	* cpu-specific code has been rearranged to make better use of both
+	  hardware commonalities and model-specific compiler optimizations.
+	* XGENE1 has been removed as a TARGET, superseded by the improved generic
+	  ARMV8 support.
+	
+ARMV7:
+	* Older assembly mnemonics have been converted to UAL form to allow
+	  building with clang 7.0
+	* Cross compiling LAPACKE for Android has been fixed again (broken by
+	  update to LAPACK 3.7.0 some while ago).  
+	  
 ====================================================================
 Version 0.3.3
 31-Aug-2018

Makefile.rule

@@ -3,7 +3,7 @@
 #
 
 # This library's version
-VERSION = 0.3.4.dev
+VERSION = 0.3.5.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.system

@@ -1042,6 +1042,8 @@ ifdef USE_TLS
 CCOMMON_OPT += -DUSE_TLS
 endif
 
+CCOMMON_OPT += -DVERSION=\"$(VERSION)\"
+
 ifndef SYMBOLPREFIX
 SYMBOLPREFIX =
 endif

cmake/system.cmake

@@ -310,6 +310,8 @@ if (MIXED_MEMORY_ALLOCATION)
   set(CCOMMON_OPT "${CCOMMON_OPT} -DMIXED_MEMORY_ALLOCATION")
 endif ()
 
+set(CCOMMON_OPT "${CCOMMON_OPT} -DVERSION=\"\\\"${OpenBLAS_VERSION}\\\"\"")
+
 set(REVISION "-r${OpenBLAS_VERSION}")
 set(MAJOR_VERSION ${OpenBLAS_MAJOR_VERSION})
 

cmake/system_check.cmake

@@ -10,6 +10,16 @@ if (${HOST_OS} STREQUAL "WINDOWS")
   set(HOST_OS WINNT)
 endif ()
 
+if (${HOST_OS} STREQUAL "LINUX")
+# check if we're building natively on Android (TERMUX)
+    EXECUTE_PROCESS( COMMAND uname -o COMMAND tr -d '\n' OUTPUT_VARIABLE OPERATING_SYSTEM)
+      if(${OPERATING_SYSTEM} MATCHES "Android")
+        set(HOST_OS ANDROID)
+      endif(${OPERATING_SYSTEM} MATCHES "Android")
+endif()
+
+
+
 if(CMAKE_COMPILER_IS_GNUCC AND WIN32)
     execute_process(COMMAND ${CMAKE_C_COMPILER} -dumpmachine
               OUTPUT_VARIABLE OPENBLAS_GCC_TARGET_MACHINE

cpuid_power.c

@@ -175,9 +175,9 @@ int detect(void){
   return  CPUTYPE_PPC970;
 #endif
 
-#if defined(__FreeBSD__) || defined(__OpenBSD__) || defined(__DragonFly__)
+#if defined(__FreeBSD__) || defined(__OpenBSD__) || defined(__NetBSD__)
 int id;
-id = __asm __volatile("mfpvr %0" : "=r"(id));
+__asm __volatile("mfpvr %0" : "=r"(id));
 switch ( id >> 16 ) {
   case 0x4e: // POWER9
     return CPUTYPE_POWER8;

driver/others/memory.c

@@ -2586,7 +2586,7 @@ void *blas_memory_alloc(int procpos){
   printf("Alloc Start ...\n");
 #endif
 
-#if defined(WHEREAMI) && !defined(USE_OPENMP)
+/* #if defined(WHEREAMI) && !defined(USE_OPENMP)
 
   mypos = WhereAmI();
 
@@ -2596,12 +2596,12 @@ void *blas_memory_alloc(int procpos){
   do {
     if (!memory[position].used && (memory[position].pos == mypos)) {
       LOCK_COMMAND(&alloc_lock);
-/*      blas_lock(&memory[position].lock);*/
+//      blas_lock(&memory[position].lock);
 
       if (!memory[position].used) goto allocation;
 
       UNLOCK_COMMAND(&alloc_lock);
-/*      blas_unlock(&memory[position].lock);*/
+//      blas_unlock(&memory[position].lock);
     }
 
     position ++;
@@ -2609,7 +2609,7 @@ void *blas_memory_alloc(int procpos){
   } while (position < NUM_BUFFERS);
 
 
-#endif
+#endif */
 
   position = 0;
 

driver/others/openblas_get_config.c

@@ -42,6 +42,9 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #endif
 
 static char* openblas_config_str=""
+"OpenBLAS "
+ VERSION
+" "
 #ifdef USE64BITINT
   " USE64BITINT "
 #endif

kernel/mips64/KERNEL

@@ -7,6 +7,7 @@ ZROTKERNEL = ../mips/zrot.c
 CSWAPKERNEL = ../mips/zswap.c
 ZSWAPKERNEL = ../mips/zswap.c
                                                                                         
+                                                                                                                                          
 ifndef SNRM2KERNEL
 SNRM2KERNEL = snrm2.S
 endif

kernel/mips64/KERNEL.LOONGSON3A

@@ -63,6 +63,7 @@ ZTRSMKERNEL_LT	= ../generic/trsm_kernel_LT.c
 ZTRSMKERNEL_RN	= ../generic/trsm_kernel_RN.c
 ZTRSMKERNEL_RT	= ../generic/trsm_kernel_RT.c
 
+DSDOTKERNEL     = ../mips/dot.c
 
 
 

kernel/mips64/sgemm_kernel_8x4_ps.S

@@ -146,11 +146,11 @@
 	sd	$21,  40($sp)
 	sd	$22,  48($sp)
 
-	ST	$f24, 56($sp)
+	sdc1	$f24, 56($sp)
-	ST	$f25, 64($sp)
+	sdc1	$f25, 64($sp)
-	ST	$f26, 72($sp)
+	sdc1	$f26, 72($sp)
-	ST	$f27, 80($sp)
+	sdc1	$f27, 80($sp)
-	ST	$f28, 88($sp)
+	sdc1	$f28, 88($sp)
 
 #if defined(TRMMKERNEL)
 	sd	$23,  96($sp)
@@ -161,10 +161,10 @@
 #endif
 
 #ifndef __64BIT__
-	ST	$f20,120($sp)
+	sdc1	$f20,120($sp)
-	ST	$f21,128($sp)
+	sdc1	$f21,128($sp)
-	ST	$f22,136($sp)
+	sdc1	$f22,136($sp)
-	ST	$f23,144($sp)
+	sdc1	$f23,144($sp)
 #endif
 
 	.align	4
@@ -7766,11 +7766,11 @@
 	ld	$21,  40($sp)
 	ld	$22,  48($sp)
 
-	LD	$f24, 56($sp)
+	ldc1	$f24, 56($sp)
-	LD	$f25, 64($sp)
+	ldc1	$f25, 64($sp)
-	LD	$f26, 72($sp)
+	ldc1	$f26, 72($sp)
-	LD	$f27, 80($sp)
+	ldc1	$f27, 80($sp)
-	LD	$f28, 88($sp)
+	ldc1	$f28, 88($sp)
 
 #if defined(TRMMKERNEL)
 	ld	$23,  96($sp)
@@ -7779,10 +7779,10 @@
 #endif
 
 #ifndef __64BIT__
-	LD	$f20,120($sp)
+	ldc1	$f20,120($sp)
-	LD	$f21,128($sp)
+	ldc1	$f21,128($sp)
-	LD	$f22,136($sp)
+	ldc1	$f22,136($sp)
-	LD	$f23,144($sp)
+	ldc1	$f23,144($sp)
 #endif
 
 	daddiu	$sp,$sp,STACKSIZE

kernel/x86_64/sgemm_beta_skylakex.c

@@ -56,7 +56,7 @@ int CNAME(BLASLONG m, BLASLONG n, BLASLONG dummy1, FLOAT beta,
   }
 
   if (n == 0 || m == 0)
-	return;
+	return 0;
 
   c_offset = c;