Merge pull request #4 from xianyi/develop

rebase

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 12.dev)
+set(OpenBLAS_PATCH_VERSION 13.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 0.3.13
+ 12-Dec-2020
+ 
+ common:
+	* Added a generic bfloat16 SBGEMV kernel
+	* Fixed a potentially severe memory leak after fork in OpenMP builds
+	  that was introduced in 0.3.12
+	* Added detection of the Fujitsu Fortran compiler
+	* Added detection of the (e)gfortran compiler on OpenBSD
+	* Added support for overriding the default name of the library independently
+	  from symbol suffixing in the gmake builds (already supported in cmake)
+
+RISCV:
+	* Added a RISC V port optimized for C910V
+
+POWER:
+	* Added optimized POWER10 kernels for SAXPY, CAXPY, SDOT, DDOT and DGEMV_N
+	* Improved DGEMM performance on POWER10
+	* Improved STRSM and DTRSM performance on POWER9 and POWER10
+	* Fixed segmemtation faults in DYNAMIC_ARCH builds
+ 	* Fixed compilation with the PGI compiler
+
+x86:
+	* Fixed compilation of kernels that require SSE2 intrinsics since 0.3.12
+	
+x86_64:
+	* Added an optimized bfloat16 SBGEMV kernel for SkylakeX and Cooperlake
+	* Improved the performance of SASUM and DASUM kernels through parallelization
+	* Improved the performance of SROT and DROT kernels
+	* Improved the performance of multithreaded xSYRK
+	* Fixed OpenMP builds that use the LLVM Clang compiler together with GNU gfortran
+	  (where linking of both the LLVM libomp and GNU libgomp could lead to lockups or
+	  wrong results)
+	* Fixed miscompilations by old gcc 4.6
+	* Fixed misdetection of AVX2 capability in some Sandybridge cpus
+	* Fixed lockups in builds combining DYNAMIC_ARCH with TARGET=GENERIC on OpenBSD
+
+ARM64:
+	* Fixed segmemtation faults in DYNAMIC_ARCH builds
+
+MIPS:
+	* Improved kernels for Loongson 3R3 ("3A") and 3R4 ("3B") models, including MSA
+	* Fixed bugs in the MSA kernels for CGEMM, CTRMM, CGEMV and ZGEMV
+	* Added handling of zero increments in the MSA kernels for SSWAP and DSWAP
+	* Added DYNAMIC_ARCH support for MIPS64 (currently Loongson3R3/3R4 only)
+
+SPARC:
+	* Fixed building 32 and 64 bit SPARC kernels with the SolarisStudio compilers
+
 ====================================================================
 Version 0.3.12
  24-Oct-2020

Makefile.rule

@@ -3,7 +3,7 @@
 #
 
 # This library's version
-VERSION = 0.3.12.dev
+VERSION = 0.3.13.dev
 
 # If you set the suffix, the library name will be libopenblas_$(LIBNAMESUFFIX).a
 # and libopenblas_$(LIBNAMESUFFIX).so. Meanwhile, the soname in shared library

driver/others/dynamic_power.c

@@ -53,8 +53,10 @@ static gotoblas_t *get_coretype(void) {
 		return &gotoblas_POWER10;
 #endif
 	/* Fall back to the POWER9 implementation if the toolchain is too old or the MMA feature is not set */
+#if (!defined __GNUC__) || ( __GNUC__ >= 6)
 	if (__builtin_cpu_is("power10"))
 		return &gotoblas_POWER9;
+#endif	
 	return NULL;
 }
 

kernel/power/ccopy_microk_power10.c

@@ -0,0 +1,115 @@
+/***************************************************************************
+Copyright (c) 2020, The OpenBLAS Project
+All rights reserved.
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are
+met:
+1. Redistributions of source code must retain the above copyright
+notice, this list of conditions and the following disclaimer.
+2. Redistributions in binary form must reproduce the above copyright
+notice, this list of conditions and the following disclaimer in
+the documentation and/or other materials provided with the
+distribution.
+3. Neither the name of the OpenBLAS project nor the names of
+its contributors may be used to endorse or promote products
+derived from this software without specific prior written permission.
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ARE DISCLAIMED. IN NO EVENT SHALL THE OPENBLAS PROJECT OR CONTRIBUTORS BE
+LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE
+USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*****************************************************************************/
+
+#define HAVE_KERNEL 1
+
+static void copy_kernel (BLASLONG n, FLOAT *x, FLOAT *y)
+{
+  __asm__
+    (
+       "lxvp		32, 0(%2)	\n\t"
+       "lxvp		34, 32(%2)	\n\t"
+       "lxvp		36, 64(%2)	\n\t"
+       "lxvp		38, 96(%2)	\n\t"
+       "lxvp		40, 128(%2)	\n\t"
+       "lxvp		42, 160(%2)	\n\t"
+       "lxvp		44, 192(%2)	\n\t"
+       "lxvp		46, 224(%2)	\n\t"
+
+       "addi		%2, %2, 256	\n\t"
+       "addic.		%1, %1, -32	\n\t"
+       "ble		two%=		\n\t"
+
+       ".align	5		\n"
+     "one%=:				\n\t"
+
+       "stxv		33, 0(%3)	\n\t"
+       "stxv		32, 16(%3)	\n\t"
+       "stxv		35, 32(%3)	\n\t"
+       "stxv		34, 48(%3)	\n\t"
+       "stxv		37, 64(%3)	\n\t"
+       "stxv		36, 80(%3)	\n\t"
+       "stxv		39, 96(%3)	\n\t"
+       "stxv		38, 112(%3)	\n\t"
+       "lxvp		32, 0(%2)	\n\t"
+       "lxvp		34, 32(%2)	\n\t"
+       "lxvp		36, 64(%2)	\n\t"
+       "lxvp		38, 96(%2)	\n\t"
+
+       "stxv		41, 128(%3)	\n\t"
+       "stxv		40, 144(%3)	\n\t"
+       "stxv		43, 160(%3)	\n\t"
+       "stxv		42, 176(%3)	\n\t"
+       "stxv		45, 192(%3)	\n\t"
+       "stxv		44, 208(%3)	\n\t"
+       "stxv		47, 224(%3)	\n\t"
+       "stxv		46, 240(%3)	\n\t"
+       "lxvp		40, 128(%2)	\n\t"
+       "lxvp		42, 160(%2)	\n\t"
+       "lxvp		44, 192(%2)	\n\t"
+       "lxvp		46, 224(%2)	\n\t"
+
+
+       "addi		%3, %3, 256	\n\t"
+       "addi		%2, %2, 256	\n\t"
+
+       "addic.		%1, %1, -32	\n\t"
+       "bgt		one%=		\n"
+
+     "two%=:				\n\t"
+
+       "stxv		33, 0(%3)	\n\t"
+       "stxv		32, 16(%3)	\n\t"
+       "stxv		35, 32(%3)	\n\t"
+       "stxv		34, 48(%3)	\n\t"
+       "stxv		37, 64(%3)	\n\t"
+       "stxv		36, 80(%3)	\n\t"
+       "stxv		39, 96(%3)	\n\t"
+       "stxv		38, 112(%3)	\n\t"
+       "stxv		41, 128(%3)	\n\t"
+       "stxv		40, 144(%3)	\n\t"
+       "stxv		43, 160(%3)	\n\t"
+       "stxv		42, 176(%3)	\n\t"
+       "stxv		45, 192(%3)	\n\t"
+       "stxv		44, 208(%3)	\n\t"
+       "stxv		47, 224(%3)	\n\t"
+       "stxv		46, 240(%3)	\n\t"
+
+     "#n=%1 x=%4=%2 y=%0=%3"
+     :
+       "=m" (*y),
+       "+r" (n),	// 1
+       "+b" (x),	// 2
+       "+b" (y) 	// 3
+     :
+       "m" (*x)
+     :
+       "cr0",
+       "vs32","vs33","vs34","vs35","vs36","vs37","vs38","vs39",
+       "vs40","vs41","vs42","vs43","vs44","vs45","vs46","vs47"
+     );
+}

kernel/power/ccopy_power10.c

@@ -28,7 +28,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #include "common.h"
 
 #if defined(__VEC__) || defined(__ALTIVEC__)
-#include "copy_microk_power10.c"
+#include "ccopy_microk_power10.c"
 #endif
 
 #ifndef HAVE_KERNEL
@@ -86,7 +86,7 @@ int CNAME(BLASLONG n, FLOAT *x, BLASLONG inc_x, FLOAT *y, BLASLONG inc_y)
 	if ( (inc_x == 1) && (inc_y == 1 ))
 	{
 
-		BLASLONG n1 = n & -64;
+		BLASLONG n1 = n & -32;
 		if ( n1 > 0 )
 		{
 			copy_kernel(n1, x, y);

kernel/power/copy_microk_power10.c

@@ -62,38 +62,39 @@ static void copy_kernel (BLASLONG n, FLOAT *x, FLOAT *y)
      "one%=:				\n\t"
 
        "stxvp		32, 0(%3)	\n\t"
-       "lxvp		32, 0(%2)	\n\t"
        "stxvp		34, 32(%3)	\n\t"
-       "lxvp		34, 32(%2)	\n\t"
        "stxvp		36, 64(%3)	\n\t"
-       "lxvp		36, 64(%2)	\n\t"
        "stxvp		38, 96(%3)	\n\t"
+       "lxvp		32, 0(%2)	\n\t"
+       "lxvp		34, 32(%2)	\n\t"
+       "lxvp		36, 64(%2)	\n\t"
        "lxvp		38, 96(%2)	\n\t"
 
        "stxvp		40, 128(%3)	\n\t"
-       "lxvp		40, 128(%2)	\n\t"
        "stxvp		42, 160(%3)	\n\t"
-       "lxvp		42, 160(%2)	\n\t"
        "stxvp		44, 192(%3)	\n\t"
-       "lxvp		44, 192(%2)	\n\t"
        "stxvp		46, 224(%3)	\n\t"
+       "lxvp		40, 128(%2)	\n\t"
+       "lxvp		42, 160(%2)	\n\t"
+       "lxvp		44, 192(%2)	\n\t"
        "lxvp		46, 224(%2)	\n\t"
 
        "stxvp		48, 256(%3)	\n\t"
-       "lxvp		48, 256(%2)	\n\t"
        "stxvp		50, 288(%3)	\n\t"
-       "lxvp		50, 288(%2)	\n\t"
        "stxvp		52, 320(%3)	\n\t"
-       "lxvp		52, 320(%2)	\n\t"
        "stxvp		54, 352(%3)	\n\t"
+       "lxvp		48, 256(%2)	\n\t"
+       "lxvp		50, 288(%2)	\n\t"
+       "lxvp		52, 320(%2)	\n\t"
        "lxvp		54, 352(%2)	\n\t"
+
        "stxvp		56, 384(%3)	\n\t"
-       "lxvp		56, 384(%2)	\n\t"
        "stxvp		58, 416(%3)	\n\t"
-       "lxvp		58, 416(%2)	\n\t"
        "stxvp		60, 448(%3)	\n\t"
-       "lxvp		60, 448(%2)	\n\t"
        "stxvp		62, 480(%3)	\n\t"
+       "lxvp		56, 384(%2)	\n\t"
+       "lxvp		58, 416(%2)	\n\t"
+       "lxvp		60, 448(%2)	\n\t"
        "lxvp		62, 480(%2)	\n\t"
 
        "addi		%3, %3, 512	\n\t"

kernel/power/dcopy_power10.c

@@ -85,12 +85,18 @@ int CNAME(BLASLONG n, FLOAT *x, BLASLONG inc_x, FLOAT *y, BLASLONG inc_y)
 
 	if ( (inc_x == 1) && (inc_y == 1 ))
 	{
-
-		BLASLONG n1 = n & -64;
-		if ( n1 > 0 )
+		if ( n >= 64 )
+		{
+			BLASLONG align = ((32 - ((uintptr_t)y & (uintptr_t)0x1F)) >> 3) & 0x3;
+			for (i = 0; i < align; i++) {
+				y[i] = x[i] ;
+			}
+		}
+		BLASLONG n1 = (n-i) & -64;
+		if ( n1 )
 		{
-			copy_kernel(n1, x, y);
-			i=n1;
+			copy_kernel(n1, &x[i], &y[i]);
+			i += n1;
 		}
 
 		while(i < n)

kernel/power/scopy_power10.c

@@ -86,11 +86,18 @@ int CNAME(BLASLONG n, FLOAT *x, BLASLONG inc_x, FLOAT *y, BLASLONG inc_y)
 	if ( (inc_x == 1) && (inc_y == 1 ))
 	{
 
-		BLASLONG n1 = n & -128;
-		if ( n1 > 0 )
+		if ( n >= 128 )
 		{
-			copy_kernel (n1, x, y);
-			i=n1;
+			BLASLONG align = ((32 - ((uintptr_t)y & (uintptr_t)0x1F)) >> 2) & 0x7;
+			for (i = 0; i < align; i++) {
+				y[i] = x[i] ;
+			}
+		}
+		BLASLONG n1 = (n-i) & -128;
+		if ( n1 )
+		{
+			copy_kernel(n1, &x[i], &y[i]);
+			i += n1;
 		}
 
 		while(i < n)