Merge branch 'develop' into develop_arm_softfp

CMakeLists.txt

@@ -236,7 +236,11 @@ install(TARGETS ${OpenBLAS_LIBNAME}
 	DEPENDS ${CMAKE_CURRENT_SOURCE_DIR}/config.h
 	COMMAND ${GENCONFIG_BIN} ${CMAKE_CURRENT_SOURCE_DIR}/config.h ${CMAKE_CURRENT_SOURCE_DIR}/openblas_config_template.h > ${CMAKE_BINARY_DIR}/openblas_config.h
 	)
-	ADD_CUSTOM_TARGET(genconfig DEPENDS openblas_config.h)	
+
+	ADD_CUSTOM_TARGET(genconfig
+	ALL
+	DEPENDS openblas_config.h
+	)
 	add_dependencies(genconfig ${OpenBLAS_LIBNAME})
 
 	install (FILES ${CMAKE_BINARY_DIR}/openblas_config.h DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
@@ -244,6 +248,7 @@ install(TARGETS ${OpenBLAS_LIBNAME}
 	message(STATUS "Generating f77blas.h in ${CMAKE_INSTALL_INCLUDEDIR}")
 
 	ADD_CUSTOM_TARGET(genf77blas
+	ALL
 	COMMAND ${AWK} 'BEGIN{print \"\#ifndef OPENBLAS_F77BLAS_H\" \; print \"\#define OPENBLAS_F77BLAS_H\" \; print \"\#include \\"openblas_config.h\\" \"}; NF {print}; END{print \"\#endif\"}' ${CMAKE_CURRENT_SOURCE_DIR}/common_interface.h > ${CMAKE_BINARY_DIR}/f77blas.h
 	DEPENDS ${CMAKE_CURRENT_SOURCE_DIR}/config.h
 	)
@@ -255,11 +260,11 @@ if(NOT NO_CBLAS)
 	message (STATUS "Generating cblas.h in ${CMAKE_INSTALL_INCLUDEDIR}")
 
 	ADD_CUSTOM_TARGET(gencblas
+	ALL
 	COMMAND ${SED} 's/common/openblas_config/g' ${CMAKE_CURRENT_SOURCE_DIR}/cblas.h > "${CMAKE_BINARY_DIR}/cblas.tmp"
 	COMMAND cp "${CMAKE_BINARY_DIR}/cblas.tmp" "${CMAKE_BINARY_DIR}/cblas.h"
 	DEPENDS ${CMAKE_CURRENT_SOURCE_DIR}/cblas.h
 	)
-
 	add_dependencies(gencblas ${OpenBLAS_LIBNAME})
 
 	install (FILES ${CMAKE_BINARY_DIR}/cblas.h DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})

Makefile.arm64

@@ -20,6 +20,6 @@ FCOMMON_OPT += -mtune=thunderx -mcpu=thunderx
 endif
 
 ifeq ($(CORE), THUNDERX2T99)
-CCOMMON_OPT += -mtune=vulcan -mcpu=vulcan
-FCOMMON_OPT += -mtune=vulcan -mcpu=vulcan
+CCOMMON_OPT += -mtune=thunderx2t99 -mcpu=thunderx2t99
+FCOMMON_OPT += -mtune=thunderx2t99 -mcpu=thunderx2t99
 endif

Makefile.system

@@ -496,6 +496,13 @@ CCOMMON_OPT += -mfloat-abi=softfp
 FCOMMON_OPT += -mfloat-abi=softfp
 endif
 
+ifeq ($(OSNAME), Android)
+ifeq ($(ARM_SOFTFP_ABI), 1)
+EXTRALIB	+= -lm
+else
+EXTRALIB	+= -Wl,-lm_hard
+endif
+endif
 endif
 
 ifeq ($(ARCH), arm64)

cmake/c_check.cmake

@@ -91,3 +91,8 @@ file(WRITE ${TARGET_CONF}
   "#define __${BINARY}BIT__\t1\n"
   "#define FUNDERSCORE\t${FU}\n")
 
+if (${HOST_OS} STREQUAL "WINDOWSSTORE")
+  file(APPEND ${TARGET_CONF}
+    "#define OS_WINNT\t1\n")
+endif ()
+

cmake/os.cmake

@@ -77,7 +77,7 @@ if (CYGWIN)
   set(NO_EXPRECISION 1)
 endif ()
 
-if (NOT ${CMAKE_SYSTEM_NAME} STREQUAL "Windows" AND NOT ${CMAKE_SYSTEM_NAME} STREQUAL "Interix")
+if (NOT ${CMAKE_SYSTEM_NAME} STREQUAL "Windows" AND NOT ${CMAKE_SYSTEM_NAME} STREQUAL "Interix" AND NOT ${CMAKE_SYSTEM_NAME} STREQUAL "Android")
   if (SMP)
     set(EXTRALIB "${EXTRALIB} -lpthread")
   endif ()

cmake/prebuild.cmake

@@ -72,20 +72,26 @@ if (MSVC)
 set(GETARCH_FLAGS ${GETARCH_FLAGS} -DFORCE_GENERIC)
 endif()
 
+if ("${CMAKE_SYSTEM_NAME}" STREQUAL "WindowsStore")
+  # disable WindowsStore strict CRT checks
+  set(GETARCH_FLAGS ${GETARCH_FLAGS} -D_CRT_SECURE_NO_WARNINGS)
+endif ()
+
 set(GETARCH_DIR "${PROJECT_BINARY_DIR}/getarch_build")
 set(GETARCH_BIN "getarch${CMAKE_EXECUTABLE_SUFFIX}")
 file(MAKE_DIRECTORY ${GETARCH_DIR})
-try_compile(GETARCH_RESULT ${GETARCH_DIR}
-  SOURCES ${GETARCH_SRC}
-  COMPILE_DEFINITIONS ${EXFLAGS} ${GETARCH_FLAGS} -I${PROJECT_SOURCE_DIR}
-  OUTPUT_VARIABLE GETARCH_LOG
-  COPY_FILE ${PROJECT_BINARY_DIR}/${GETARCH_BIN}
-)
-
-if (NOT ${GETARCH_RESULT})
-  MESSAGE(FATAL_ERROR "Compiling getarch failed ${GETARCH_LOG}")
+if (NOT "${CMAKE_SYSTEM_NAME}" STREQUAL "WindowsStore")
+  try_compile(GETARCH_RESULT ${GETARCH_DIR}
+    SOURCES ${GETARCH_SRC}
+    COMPILE_DEFINITIONS ${EXFLAGS} ${GETARCH_FLAGS} -I${PROJECT_SOURCE_DIR}
+    OUTPUT_VARIABLE GETARCH_LOG
+    COPY_FILE ${PROJECT_BINARY_DIR}/${GETARCH_BIN}
+  )
+
+  if (NOT ${GETARCH_RESULT})
+    MESSAGE(FATAL_ERROR "Compiling getarch failed ${GETARCH_LOG}")
+  endif ()
 endif ()
-
 message(STATUS "Running getarch")
 
 # use the cmake binary w/ the -E param to run a shell command in a cross-platform way
@@ -101,15 +107,17 @@ ParseGetArchVars(${GETARCH_MAKE_OUT})
 set(GETARCH2_DIR "${PROJECT_BINARY_DIR}/getarch2_build")
 set(GETARCH2_BIN "getarch_2nd${CMAKE_EXECUTABLE_SUFFIX}")
 file(MAKE_DIRECTORY ${GETARCH2_DIR})
-try_compile(GETARCH2_RESULT ${GETARCH2_DIR}
-  SOURCES ${PROJECT_SOURCE_DIR}/getarch_2nd.c
-  COMPILE_DEFINITIONS ${EXFLAGS} ${GETARCH_FLAGS} ${GETARCH2_FLAGS} -I${PROJECT_SOURCE_DIR}
-  OUTPUT_VARIABLE GETARCH2_LOG
-  COPY_FILE ${PROJECT_BINARY_DIR}/${GETARCH2_BIN}
-)
-
-if (NOT ${GETARCH2_RESULT})
-  MESSAGE(FATAL_ERROR "Compiling getarch_2nd failed ${GETARCH2_LOG}")
+if (NOT "${CMAKE_SYSTEM_NAME}" STREQUAL "WindowsStore")
+  try_compile(GETARCH2_RESULT ${GETARCH2_DIR}
+    SOURCES ${PROJECT_SOURCE_DIR}/getarch_2nd.c
+    COMPILE_DEFINITIONS ${EXFLAGS} ${GETARCH_FLAGS} ${GETARCH2_FLAGS} -I${PROJECT_SOURCE_DIR}
+    OUTPUT_VARIABLE GETARCH2_LOG
+    COPY_FILE ${PROJECT_BINARY_DIR}/${GETARCH2_BIN}
+  )
+
+  if (NOT ${GETARCH2_RESULT})
+    MESSAGE(FATAL_ERROR "Compiling getarch_2nd failed ${GETARCH2_LOG}")
+  endif ()
 endif ()
 
 # use the cmake binary w/ the -E param to run a shell command in a cross-platform way
@@ -126,13 +134,15 @@ set(GEN_CONFIG_H_BIN "gen_config_h${CMAKE_EXECUTABLE_SUFFIX}")
 set(GEN_CONFIG_H_FLAGS "-DVERSION=\"${OpenBLAS_VERSION}\"")
 file(MAKE_DIRECTORY ${GEN_CONFIG_H_DIR})
 
-try_compile(GEN_CONFIG_H_RESULT ${GEN_CONFIG_H_DIR}
-  SOURCES ${PROJECT_SOURCE_DIR}/gen_config_h.c
-  COMPILE_DEFINITIONS ${EXFLAGS} ${GETARCH_FLAGS} ${GEN_CONFIG_H_FLAGS} -I${PROJECT_SOURCE_DIR}
-  OUTPUT_VARIABLE GEN_CONFIG_H_LOG
-  COPY_FILE ${PROJECT_BINARY_DIR}/${GEN_CONFIG_H_BIN}
-)
-
-if (NOT ${GEN_CONFIG_H_RESULT})
-  MESSAGE(FATAL_ERROR "Compiling gen_config_h failed ${GEN_CONFIG_H_LOG}")
-endif ()
+if (NOT "${CMAKE_SYSTEM_NAME}" STREQUAL "WindowsStore")
+  try_compile(GEN_CONFIG_H_RESULT ${GEN_CONFIG_H_DIR}
+    SOURCES ${PROJECT_SOURCE_DIR}/gen_config_h.c
+    COMPILE_DEFINITIONS ${EXFLAGS} ${GETARCH_FLAGS} ${GEN_CONFIG_H_FLAGS} -I${PROJECT_SOURCE_DIR}
+    OUTPUT_VARIABLE GEN_CONFIG_H_LOG
+    COPY_FILE ${PROJECT_BINARY_DIR}/${GEN_CONFIG_H_BIN}
+  )
+
+  if (NOT ${GEN_CONFIG_H_RESULT})
+    MESSAGE(FATAL_ERROR "Compiling gen_config_h failed ${GEN_CONFIG_H_LOG}")
+  endif ()
+endif ()

common.h

@@ -425,6 +425,10 @@ please https://github.com/xianyi/OpenBLAS/issues/246
 #endif
 
 #ifndef ASSEMBLER
+#ifdef OS_WINDOWSSTORE
+typedef char env_var_t[MAX_PATH];
+#define readenv(p, n) 0
+#else
 #ifdef OS_WINDOWS
 typedef char env_var_t[MAX_PATH];
 #define readenv(p, n) GetEnvironmentVariable((LPCTSTR)(n), (LPTSTR)(p), sizeof(p))
@@ -432,6 +436,7 @@ typedef char env_var_t[MAX_PATH];
 typedef char* env_var_t;
 #define readenv(p, n) ((p)=getenv(n))
 #endif
+#endif
 
 #if !defined(RPCC_DEFINED) && !defined(OS_WINDOWS)
 #ifdef _POSIX_MONOTONIC_CLOCK
@@ -654,7 +659,11 @@ static __inline void blas_unlock(volatile BLASULONG *address){
   *address = 0;
 }
 
-
+#ifdef OS_WINDOWSSTORE
+static __inline int readenv_atoi(char *env) {
+	return 0;
+}
+#else
 #ifdef OS_WINDOWS
 static __inline int readenv_atoi(char *env) {
   env_var_t p;
@@ -669,7 +678,7 @@ static __inline int readenv_atoi(char *env) {
 	return(0);
 }
 #endif
-
+#endif
 
 #if !defined(XDOUBLE) || !defined(QUAD_PRECISION)
 

driver/level3/syrk_thread.c

@@ -109,7 +109,7 @@ int CNAME(int mode, blas_arg_t *arg, BLASLONG *range_m, BLASLONG *range_n, int (
       if (nthreads - num_cpu > 1) {
 
 	di = (double)i;
-	width = ((BLASLONG)( sqrt(di * di + dnum) - di) + mask) & ~mask;
+	width = (BLASLONG)(( sqrt(di * di + dnum) - di + mask)/(mask+1)) * (mask+1);
 
 	if ((width <= 0) || (width > n_to - i)) width = n_to - i;
 
@@ -149,7 +149,7 @@ int CNAME(int mode, blas_arg_t *arg, BLASLONG *range_m, BLASLONG *range_n, int (
       if (nthreads - num_cpu > 1) {
 
 	di = (double)(arg -> n - i);
-	width = ((BLASLONG)(-sqrt(di * di + dnum) + di) + mask) & ~mask;
+	width = ((BLASLONG)((-sqrt(di * di + dnum) + di) + mask)/(mask+1)) * (mask+1);
 
 	if ((width <= 0) || (width > n_to - i)) width = n_to - i;
 

driver/others/CMakeLists.txt

@@ -12,6 +12,8 @@ if (SMP)
     set(BLAS_SERVER blas_server_omp.c)
   elseif (${CMAKE_SYSTEM_NAME} STREQUAL "Windows")
     set(BLAS_SERVER blas_server_win32.c)
+  elseif (${CMAKE_SYSTEM_NAME} STREQUAL "WindowsStore")
+    set(BLAS_SERVER blas_server_win32.c)
   endif ()
 
   if (NOT DEFINED BLAS_SERVER)

driver/others/blas_server_win32.c

@@ -443,8 +443,11 @@ int BLASFUNC(blas_thread_shutdown)(void){
     SetEvent(pool.killed);
 
     for(i = 0; i < blas_num_threads - 1; i++){
-     WaitForSingleObject(blas_threads[i], 5);  //INFINITE);
-	 TerminateThread(blas_threads[i],0);
+      WaitForSingleObject(blas_threads[i], 5);  //INFINITE);
+#ifndef OS_WINDOWSSTORE
+// TerminateThread is only available with WINAPI_DESKTOP and WINAPI_SYSTEM not WINAPI_APP in UWP
+      TerminateThread(blas_threads[i],0);
+#endif
     }
 
     blas_server_avail = 0;

driver/others/init.c

@@ -354,6 +354,24 @@ static int numa_check(void) {
   return common -> num_nodes;
 }
 
+#if defined(__GLIBC_PREREQ)
+#if !__GLIBC_PREREQ(2, 6)
+int sched_getcpu(void)
+{
+int cpu;
+FILE *fp = NULL;
+if ( (fp = fopen("/proc/self/stat", "r")) == NULL)
+   return -1;
+if ( fscanf( fp, "%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%*s%d", &cpu) != 1) {
+  fclose (fp);
+  return -1;
+  }
+  fclose (fp);
+  return(cpu);
+}
+#endif
+#endif
+
 static void numa_mapping(void) {
 
   int node, cpu, core;
@@ -808,7 +826,6 @@ void gotoblas_affinity_init(void) {
   common -> shmid = pshmid;
 
   if (common -> magic != SH_MAGIC) {
-
 #ifdef DEBUG
     fprintf(stderr, "Shared Memory Initialization.\n");
 #endif
@@ -830,7 +847,7 @@ void gotoblas_affinity_init(void) {
     if (common -> num_nodes > 1) numa_mapping();
 
     common -> final_num_procs = 0;
-    for(i = 0; i < common -> avail_count; i++) common -> final_num_procs += rcount(common -> avail[i]) + 1;   //Make the max cpu number.
+    for(i = 0; i < common -> avail_count; i++) common -> final_num_procs += rcount(common -> avail[i]) + 1;   //Make the max cpu number. 
 
     for (cpu = 0; cpu < common -> final_num_procs; cpu ++) common -> cpu_use[cpu] =  0;
 

kernel/power/casum_microk_power8.c

@@ -56,14 +56,14 @@ static float casum_kernel_16 (long n, float *x)
        "xxlxor		38, 38,	38	\n\t"
        "xxlxor		39, 39,	39	\n\t"
 
-       "lxvw4x		40, 0, %2	\n\t"
-       "lxvw4x		41, %8, %2	\n\t"
-       "lxvw4x		42, %9, %2	\n\t"
-       "lxvw4x		43, %10, %2	\n\t"
-       "lxvw4x		44, %11, %2	\n\t"
-       "lxvw4x		45, %12, %2	\n\t"
-       "lxvw4x		46, %13, %2	\n\t"
-       "lxvw4x		47, %14, %2	\n\t"
+       "lxvd2x		40, 0, %2	\n\t"
+       "lxvd2x		41, %8, %2	\n\t"
+       "lxvd2x		42, %9, %2	\n\t"
+       "lxvd2x		43, %10, %2	\n\t"
+       "lxvd2x		44, %11, %2	\n\t"
+       "lxvd2x		45, %12, %2	\n\t"
+       "lxvd2x		46, %13, %2	\n\t"
+       "lxvd2x		47, %14, %2	\n\t"
 
        "addi		%2, %2, 128	\n\t"
 
@@ -78,26 +78,26 @@ static float casum_kernel_16 (long n, float *x)
        "xvabssp		50, 42		\n\t"
        "xvabssp		51, 43		\n\t"
 
-       "lxvw4x		40, 0, %2	\n\t"
-       "lxvw4x		41, %8, %2	\n\t"
+       "lxvd2x		40, 0, %2	\n\t"
+       "lxvd2x		41, %8, %2	\n\t"
 
        "xvabssp		%x3, 44		\n\t"
        "xvabssp		%x4, 45		\n\t"
 
-       "lxvw4x		42, %9, %2	\n\t"
-       "lxvw4x		43, %10, %2	\n\t"
+       "lxvd2x		42, %9, %2	\n\t"
+       "lxvd2x		43, %10, %2	\n\t"
 
        "xvabssp		%x5, 46		\n\t"
        "xvabssp		%x6, 47		\n\t"
 
-       "lxvw4x		44, %11, %2	\n\t"
-       "lxvw4x		45, %12, %2	\n\t"
+       "lxvd2x		44, %11, %2	\n\t"
+       "lxvd2x		45, %12, %2	\n\t"
 
        "xvaddsp		32, 32, 48	\n\t"
        "xvaddsp		33, 33, 49	\n\t"
 
-       "lxvw4x		46, %13, %2	\n\t"
-       "lxvw4x		47, %14, %2	\n\t"
+       "lxvd2x		46, %13, %2	\n\t"
+       "lxvd2x		47, %14, %2	\n\t"
 
        "xvaddsp		34, 34, 50	\n\t"
        "xvaddsp		35, 35, 51	\n\t"

kernel/power/ccopy_microk_power8.c

@@ -39,25 +39,25 @@ static void ccopy_kernel_32 (long n, float *x, float *y)
 {
   __asm__
     (
-       "lxvw4x		32, 0, %2	\n\t"
-       "lxvw4x		33, %5, %2	\n\t"
-       "lxvw4x		34, %6, %2	\n\t"
-       "lxvw4x		35, %7, %2	\n\t"
-       "lxvw4x		36, %8, %2	\n\t"
-       "lxvw4x		37, %9, %2	\n\t"
-       "lxvw4x		38, %10, %2	\n\t"
-       "lxvw4x		39, %11, %2	\n\t"
+       "lxvd2x		32, 0, %2	\n\t"
+       "lxvd2x		33, %5, %2	\n\t"
+       "lxvd2x		34, %6, %2	\n\t"
+       "lxvd2x		35, %7, %2	\n\t"
+       "lxvd2x		36, %8, %2	\n\t"
+       "lxvd2x		37, %9, %2	\n\t"
+       "lxvd2x		38, %10, %2	\n\t"
+       "lxvd2x		39, %11, %2	\n\t"
 
        "addi		%2, %2, 128	\n\t"
 
-       "lxvw4x		40, 0, %2	\n\t"
-       "lxvw4x		41, %5, %2	\n\t"
-       "lxvw4x		42, %6, %2	\n\t"
-       "lxvw4x		43, %7, %2	\n\t"
-       "lxvw4x		44, %8, %2	\n\t"
-       "lxvw4x		45, %9, %2	\n\t"
-       "lxvw4x		46, %10, %2	\n\t"
-       "lxvw4x		47, %11, %2	\n\t"
+       "lxvd2x		40, 0, %2	\n\t"
+       "lxvd2x		41, %5, %2	\n\t"
+       "lxvd2x		42, %6, %2	\n\t"
+       "lxvd2x		43, %7, %2	\n\t"
+       "lxvd2x		44, %8, %2	\n\t"
+       "lxvd2x		45, %9, %2	\n\t"
+       "lxvd2x		46, %10, %2	\n\t"
+       "lxvd2x		47, %11, %2	\n\t"
 
        "addi		%2, %2, 128	\n\t"
 
@@ -67,42 +67,42 @@ static void ccopy_kernel_32 (long n, float *x, float *y)
        ".p2align	5		\n"
      "1:				\n\t"
 
-       "stxvw4x		32, 0, %3	\n\t"
-       "stxvw4x		33, %5, %3	\n\t"
-       "lxvw4x		32, 0, %2	\n\t"
-       "lxvw4x		33, %5, %2	\n\t"
-       "stxvw4x		34, %6, %3	\n\t"
-       "stxvw4x		35, %7, %3	\n\t"
-       "lxvw4x		34, %6, %2	\n\t"
-       "lxvw4x		35, %7, %2	\n\t"
-       "stxvw4x		36, %8, %3	\n\t"
-       "stxvw4x		37, %9, %3	\n\t"
-       "lxvw4x		36, %8, %2	\n\t"
-       "lxvw4x		37, %9, %2	\n\t"
-       "stxvw4x		38, %10, %3	\n\t"
-       "stxvw4x		39, %11, %3	\n\t"
-       "lxvw4x		38, %10, %2	\n\t"
-       "lxvw4x		39, %11, %2	\n\t"
+       "stxvd2x		32, 0, %3	\n\t"
+       "stxvd2x		33, %5, %3	\n\t"
+       "lxvd2x		32, 0, %2	\n\t"
+       "lxvd2x		33, %5, %2	\n\t"
+       "stxvd2x		34, %6, %3	\n\t"
+       "stxvd2x		35, %7, %3	\n\t"
+       "lxvd2x		34, %6, %2	\n\t"
+       "lxvd2x		35, %7, %2	\n\t"
+       "stxvd2x		36, %8, %3	\n\t"
+       "stxvd2x		37, %9, %3	\n\t"
+       "lxvd2x		36, %8, %2	\n\t"
+       "lxvd2x		37, %9, %2	\n\t"
+       "stxvd2x		38, %10, %3	\n\t"
+       "stxvd2x		39, %11, %3	\n\t"
+       "lxvd2x		38, %10, %2	\n\t"
+       "lxvd2x		39, %11, %2	\n\t"
 
        "addi		%3, %3, 128	\n\t"
        "addi		%2, %2, 128	\n\t"
 
-       "stxvw4x		40, 0, %3	\n\t"
-       "stxvw4x		41, %5, %3	\n\t"
-       "lxvw4x		40, 0, %2	\n\t"
-       "lxvw4x		41, %5, %2	\n\t"
-       "stxvw4x		42, %6, %3	\n\t"
-       "stxvw4x		43, %7, %3	\n\t"
-       "lxvw4x		42, %6, %2	\n\t"
-       "lxvw4x		43, %7, %2	\n\t"
-       "stxvw4x		44, %8, %3	\n\t"
-       "stxvw4x		45, %9, %3	\n\t"
-       "lxvw4x		44, %8, %2	\n\t"
-       "lxvw4x		45, %9, %2	\n\t"
-       "stxvw4x		46, %10, %3	\n\t"
-       "stxvw4x		47, %11, %3	\n\t"
-       "lxvw4x		46, %10, %2	\n\t"
-       "lxvw4x		47, %11, %2	\n\t"
+       "stxvd2x		40, 0, %3	\n\t"
+       "stxvd2x		41, %5, %3	\n\t"
+       "lxvd2x		40, 0, %2	\n\t"
+       "lxvd2x		41, %5, %2	\n\t"
+       "stxvd2x		42, %6, %3	\n\t"
+       "stxvd2x		43, %7, %3	\n\t"
+       "lxvd2x		42, %6, %2	\n\t"
+       "lxvd2x		43, %7, %2	\n\t"
+       "stxvd2x		44, %8, %3	\n\t"
+       "stxvd2x		45, %9, %3	\n\t"
+       "lxvd2x		44, %8, %2	\n\t"
+       "lxvd2x		45, %9, %2	\n\t"
+       "stxvd2x		46, %10, %3	\n\t"
+       "stxvd2x		47, %11, %3	\n\t"
+       "lxvd2x		46, %10, %2	\n\t"
+       "lxvd2x		47, %11, %2	\n\t"
 
        "addi		%3, %3, 128	\n\t"
        "addi		%2, %2, 128	\n\t"
@@ -112,25 +112,25 @@ static void ccopy_kernel_32 (long n, float *x, float *y)
 
      "2:				\n\t"
 
-       "stxvw4x		32, 0, %3	\n\t"
-       "stxvw4x		33, %5, %3	\n\t"
-       "stxvw4x		34, %6, %3	\n\t"
-       "stxvw4x		35, %7, %3	\n\t"
-       "stxvw4x		36, %8, %3	\n\t"
-       "stxvw4x		37, %9, %3	\n\t"
-       "stxvw4x		38, %10, %3	\n\t"
-       "stxvw4x		39, %11, %3	\n\t"
+       "stxvd2x		32, 0, %3	\n\t"
+       "stxvd2x		33, %5, %3	\n\t"
+       "stxvd2x		34, %6, %3	\n\t"
+       "stxvd2x		35, %7, %3	\n\t"
+       "stxvd2x		36, %8, %3	\n\t"
+       "stxvd2x		37, %9, %3	\n\t"
+       "stxvd2x		38, %10, %3	\n\t"
+       "stxvd2x		39, %11, %3	\n\t"
 
        "addi		%3, %3, 128	\n\t"
 
-       "stxvw4x		40, 0, %3	\n\t"
-       "stxvw4x		41, %5, %3	\n\t"
-       "stxvw4x		42, %6, %3	\n\t"
-       "stxvw4x		43, %7, %3	\n\t"
-       "stxvw4x		44, %8, %3	\n\t"
-       "stxvw4x		45, %9, %3	\n\t"
-       "stxvw4x		46, %10, %3	\n\t"
-       "stxvw4x		47, %11, %3	\n"
+       "stxvd2x		40, 0, %3	\n\t"
+       "stxvd2x		41, %5, %3	\n\t"
+       "stxvd2x		42, %6, %3	\n\t"
+       "stxvd2x		43, %7, %3	\n\t"
+       "stxvd2x		44, %8, %3	\n\t"
+       "stxvd2x		45, %9, %3	\n\t"
+       "stxvd2x		46, %10, %3	\n\t"
+       "stxvd2x		47, %11, %3	\n"
 
      "#n=%1 x=%4=%2 y=%0=%3 o16=%5 o32=%6 o48=%7 o64=%8 o80=%9 o96=%10 o112=%11"
      :

kernel/power/cswap_microk_power8.c

@@ -42,91 +42,91 @@ static void cswap_kernel_32 (long n, float *x, float *y)
        ".p2align	5		\n"
      "1:				\n\t"
 
-       "lxvw4x		32, 0, %4	\n\t"
-       "lxvw4x		33, %5, %4	\n\t"
-       "lxvw4x		34, %6, %4	\n\t"
-       "lxvw4x		35, %7, %4	\n\t"
-       "lxvw4x		36, %8, %4	\n\t"
-       "lxvw4x		37, %9, %4	\n\t"
-       "lxvw4x		38, %10, %4	\n\t"
-       "lxvw4x		39, %11, %4	\n\t"
+       "lxvd2x		32, 0, %4	\n\t"
+       "lxvd2x		33, %5, %4	\n\t"
+       "lxvd2x		34, %6, %4	\n\t"
+       "lxvd2x		35, %7, %4	\n\t"
+       "lxvd2x		36, %8, %4	\n\t"
+       "lxvd2x		37, %9, %4	\n\t"
+       "lxvd2x		38, %10, %4	\n\t"
+       "lxvd2x		39, %11, %4	\n\t"
 
        "addi		%4, %4, 128	\n\t"
 
-       "lxvw4x		40, 0, %4	\n\t"
-       "lxvw4x		41, %5, %4	\n\t"
-       "lxvw4x		42, %6, %4	\n\t"
-       "lxvw4x		43, %7, %4	\n\t"
-       "lxvw4x		44, %8, %4	\n\t"
-       "lxvw4x		45, %9, %4	\n\t"
-       "lxvw4x		46, %10, %4	\n\t"
-       "lxvw4x		47, %11, %4	\n\t"
+       "lxvd2x		40, 0, %4	\n\t"
+       "lxvd2x		41, %5, %4	\n\t"
+       "lxvd2x		42, %6, %4	\n\t"
+       "lxvd2x		43, %7, %4	\n\t"
+       "lxvd2x		44, %8, %4	\n\t"
+       "lxvd2x		45, %9, %4	\n\t"
+       "lxvd2x		46, %10, %4	\n\t"
+       "lxvd2x		47, %11, %4	\n\t"
 
        "addi		%4, %4, -128	\n\t"
 
-       "lxvw4x		48, 0, %3	\n\t"
-       "lxvw4x		49, %5, %3	\n\t"
-       "lxvw4x		50, %6, %3	\n\t"
-       "lxvw4x		51, %7, %3	\n\t"
-       "lxvw4x		0, %8, %3	\n\t"
-       "lxvw4x		1, %9, %3	\n\t"
-       "lxvw4x		2, %10, %3	\n\t"
-       "lxvw4x		3, %11, %3	\n\t"
+       "lxvd2x		48, 0, %3	\n\t"
+       "lxvd2x		49, %5, %3	\n\t"
+       "lxvd2x		50, %6, %3	\n\t"
+       "lxvd2x		51, %7, %3	\n\t"
+       "lxvd2x		0, %8, %3	\n\t"
+       "lxvd2x		1, %9, %3	\n\t"
+       "lxvd2x		2, %10, %3	\n\t"
+       "lxvd2x		3, %11, %3	\n\t"
 
        "addi		%3, %3, 128	\n\t"
 
-       "lxvw4x		4, 0, %3	\n\t"
-       "lxvw4x		5, %5, %3	\n\t"
-       "lxvw4x		6, %6, %3	\n\t"
-       "lxvw4x		7, %7, %3	\n\t"
-       "lxvw4x		8, %8, %3	\n\t"
-       "lxvw4x		9, %9, %3	\n\t"
-       "lxvw4x		10, %10, %3	\n\t"
-       "lxvw4x		11, %11, %3	\n\t"
+       "lxvd2x		4, 0, %3	\n\t"
+       "lxvd2x		5, %5, %3	\n\t"
+       "lxvd2x		6, %6, %3	\n\t"
+       "lxvd2x		7, %7, %3	\n\t"
+       "lxvd2x		8, %8, %3	\n\t"
+       "lxvd2x		9, %9, %3	\n\t"
+       "lxvd2x		10, %10, %3	\n\t"
+       "lxvd2x		11, %11, %3	\n\t"
 
        "addi		%3, %3, -128	\n\t"
 
-       "stxvw4x		32, 0, %3	\n\t"
-       "stxvw4x		33, %5, %3	\n\t"
-       "stxvw4x		34, %6, %3	\n\t"
-       "stxvw4x		35, %7, %3	\n\t"
-       "stxvw4x		36, %8, %3	\n\t"
-       "stxvw4x		37, %9, %3	\n\t"
-       "stxvw4x		38, %10, %3	\n\t"
-       "stxvw4x		39, %11, %3	\n\t"
+       "stxvd2x		32, 0, %3	\n\t"
+       "stxvd2x		33, %5, %3	\n\t"
+       "stxvd2x		34, %6, %3	\n\t"
+       "stxvd2x		35, %7, %3	\n\t"
+       "stxvd2x		36, %8, %3	\n\t"
+       "stxvd2x		37, %9, %3	\n\t"
+       "stxvd2x		38, %10, %3	\n\t"
+       "stxvd2x		39, %11, %3	\n\t"
 
        "addi		%3, %3, 128	\n\t"
 
-       "stxvw4x		40, 0, %3	\n\t"
-       "stxvw4x		41, %5, %3	\n\t"
-       "stxvw4x		42, %6, %3	\n\t"
-       "stxvw4x		43, %7, %3	\n\t"
-       "stxvw4x		44, %8, %3	\n\t"
-       "stxvw4x		45, %9, %3	\n\t"
-       "stxvw4x		46, %10, %3	\n\t"
-       "stxvw4x		47, %11, %3	\n\t"
+       "stxvd2x		40, 0, %3	\n\t"
+       "stxvd2x		41, %5, %3	\n\t"
+       "stxvd2x		42, %6, %3	\n\t"
+       "stxvd2x		43, %7, %3	\n\t"
+       "stxvd2x		44, %8, %3	\n\t"
+       "stxvd2x		45, %9, %3	\n\t"
+       "stxvd2x		46, %10, %3	\n\t"
+       "stxvd2x		47, %11, %3	\n\t"
 
        "addi		%3, %3, 128	\n\t"
 
-       "stxvw4x		48, 0, %4	\n\t"
-       "stxvw4x		49, %5, %4	\n\t"
-       "stxvw4x		50, %6, %4	\n\t"
-       "stxvw4x		51, %7, %4	\n\t"
-       "stxvw4x		0, %8, %4	\n\t"
-       "stxvw4x		1, %9, %4	\n\t"
-       "stxvw4x		2, %10, %4	\n\t"
-       "stxvw4x		3, %11, %4	\n\t"
+       "stxvd2x		48, 0, %4	\n\t"
+       "stxvd2x		49, %5, %4	\n\t"
+       "stxvd2x		50, %6, %4	\n\t"
+       "stxvd2x		51, %7, %4	\n\t"
+       "stxvd2x		0, %8, %4	\n\t"
+       "stxvd2x		1, %9, %4	\n\t"
+       "stxvd2x		2, %10, %4	\n\t"
+       "stxvd2x		3, %11, %4	\n\t"
 
        "addi		%4, %4, 128	\n\t"
 
-       "stxvw4x		4, 0, %4	\n\t"
-       "stxvw4x		5, %5, %4	\n\t"
-       "stxvw4x		6, %6, %4	\n\t"
-       "stxvw4x		7, %7, %4	\n\t"
-       "stxvw4x		8, %8, %4	\n\t"
-       "stxvw4x		9, %9, %4	\n\t"
-       "stxvw4x		10, %10, %4	\n\t"
-       "stxvw4x		11, %11, %4	\n\t"
+       "stxvd2x		4, 0, %4	\n\t"
+       "stxvd2x		5, %5, %4	\n\t"
+       "stxvd2x		6, %6, %4	\n\t"
+       "stxvd2x		7, %7, %4	\n\t"
+       "stxvd2x		8, %8, %4	\n\t"
+       "stxvd2x		9, %9, %4	\n\t"
+       "stxvd2x		10, %10, %4	\n\t"
+       "stxvd2x		11, %11, %4	\n\t"
 
        "addi		%4, %4, 128	\n\t"
 

kernel/power/sasum_microk_power8.c

@@ -56,14 +56,14 @@ static float sasum_kernel_32 (long n, float *x)
        "xxlxor		38, 38,	38	\n\t"
        "xxlxor		39, 39,	39	\n\t"
 
-       "lxvw4x		40, 0, %2	\n\t"
-       "lxvw4x		41, %8, %2	\n\t"
-       "lxvw4x		42, %9, %2	\n\t"
-       "lxvw4x		43, %10, %2	\n\t"
-       "lxvw4x		44, %11, %2	\n\t"
-       "lxvw4x		45, %12, %2	\n\t"
-       "lxvw4x		46, %13, %2	\n\t"
-       "lxvw4x		47, %14, %2	\n\t"
+       "lxvd2x		40, 0, %2	\n\t"
+       "lxvd2x		41, %8, %2	\n\t"
+       "lxvd2x		42, %9, %2	\n\t"
+       "lxvd2x		43, %10, %2	\n\t"
+       "lxvd2x		44, %11, %2	\n\t"
+       "lxvd2x		45, %12, %2	\n\t"
+       "lxvd2x		46, %13, %2	\n\t"
+       "lxvd2x		47, %14, %2	\n\t"
 
        "addi		%2, %2, 128	\n\t"
 
@@ -78,26 +78,26 @@ static float sasum_kernel_32 (long n, float *x)
        "xvabssp		50, 42		\n\t"
        "xvabssp		51, 43		\n\t"
 
-       "lxvw4x		40, 0, %2	\n\t"
-       "lxvw4x		41, %8, %2	\n\t"
+       "lxvd2x		40, 0, %2	\n\t"
+       "lxvd2x		41, %8, %2	\n\t"
 
        "xvabssp		%x3, 44		\n\t"
        "xvabssp		%x4, 45		\n\t"
 
-       "lxvw4x		42, %9, %2	\n\t"
-       "lxvw4x		43, %10, %2	\n\t"
+       "lxvd2x		42, %9, %2	\n\t"
+       "lxvd2x		43, %10, %2	\n\t"
 
        "xvabssp		%x5, 46		\n\t"
        "xvabssp		%x6, 47		\n\t"
 
-       "lxvw4x		44, %11, %2	\n\t"
-       "lxvw4x		45, %12, %2	\n\t"
+       "lxvd2x		44, %11, %2	\n\t"
+       "lxvd2x		45, %12, %2	\n\t"
 
        "xvaddsp		32, 32, 48	\n\t"
        "xvaddsp		33, 33, 49	\n\t"
 
-       "lxvw4x		46, %13, %2	\n\t"
-       "lxvw4x		47, %14, %2	\n\t"
+       "lxvd2x		46, %13, %2	\n\t"
+       "lxvd2x		47, %14, %2	\n\t"
 
        "xvaddsp		34, 34, 50	\n\t"
        "xvaddsp		35, 35, 51	\n\t"

kernel/power/scopy_microk_power8.c

@@ -39,14 +39,14 @@ static void scopy_kernel_32 (long n, float *x, float *y)
 {
   __asm__
     (
-       "lxvw4x		40, 0, %2	\n\t"
-       "lxvw4x		41, %5, %2	\n\t"
-       "lxvw4x		42, %6, %2	\n\t"
-       "lxvw4x		43, %7, %2	\n\t"
-       "lxvw4x		44, %8, %2	\n\t"
-       "lxvw4x		45, %9, %2	\n\t"
-       "lxvw4x		46, %10, %2	\n\t"
-       "lxvw4x		47, %11, %2	\n\t"
+       "lxvd2x		40, 0, %2	\n\t"
+       "lxvd2x		41, %5, %2	\n\t"
+       "lxvd2x		42, %6, %2	\n\t"
+       "lxvd2x		43, %7, %2	\n\t"
+       "lxvd2x		44, %8, %2	\n\t"
+       "lxvd2x		45, %9, %2	\n\t"
+       "lxvd2x		46, %10, %2	\n\t"
+       "lxvd2x		47, %11, %2	\n\t"
 
        "addi		%2, %2, 128	\n\t"
 
@@ -56,22 +56,22 @@ static void scopy_kernel_32 (long n, float *x, float *y)
        ".p2align	5		\n"
      "1:				\n\t"
 
-       "stxvw4x		40, 0, %3	\n\t"
-       "stxvw4x		41, %5, %3	\n\t"
-       "lxvw4x		40, 0, %2	\n\t"
-       "lxvw4x		41, %5, %2	\n\t"
-       "stxvw4x		42, %6, %3	\n\t"
-       "stxvw4x		43, %7, %3	\n\t"
-       "lxvw4x		42, %6, %2	\n\t"
-       "lxvw4x		43, %7, %2	\n\t"
-       "stxvw4x		44, %8, %3	\n\t"
-       "stxvw4x		45, %9, %3	\n\t"
-       "lxvw4x		44, %8, %2	\n\t"
-       "lxvw4x		45, %9, %2	\n\t"
-       "stxvw4x		46, %10, %3	\n\t"
-       "stxvw4x		47, %11, %3	\n\t"
-       "lxvw4x		46, %10, %2	\n\t"
-       "lxvw4x		47, %11, %2	\n\t"
+       "stxvd2x		40, 0, %3	\n\t"
+       "stxvd2x		41, %5, %3	\n\t"
+       "lxvd2x		40, 0, %2	\n\t"
+       "lxvd2x		41, %5, %2	\n\t"
+       "stxvd2x		42, %6, %3	\n\t"
+       "stxvd2x		43, %7, %3	\n\t"
+       "lxvd2x		42, %6, %2	\n\t"
+       "lxvd2x		43, %7, %2	\n\t"
+       "stxvd2x		44, %8, %3	\n\t"
+       "stxvd2x		45, %9, %3	\n\t"
+       "lxvd2x		44, %8, %2	\n\t"
+       "lxvd2x		45, %9, %2	\n\t"
+       "stxvd2x		46, %10, %3	\n\t"
+       "stxvd2x		47, %11, %3	\n\t"
+       "lxvd2x		46, %10, %2	\n\t"
+       "lxvd2x		47, %11, %2	\n\t"
 
        "addi		%3, %3, 128	\n\t"
        "addi		%2, %2, 128	\n\t"
@@ -81,14 +81,14 @@ static void scopy_kernel_32 (long n, float *x, float *y)
 
      "2:				\n\t"
 
-       "stxvw4x		40, 0, %3	\n\t"
-       "stxvw4x		41, %5, %3	\n\t"
-       "stxvw4x		42, %6, %3	\n\t"
-       "stxvw4x		43, %7, %3	\n\t"
-       "stxvw4x		44, %8, %3	\n\t"
-       "stxvw4x		45, %9, %3	\n\t"
-       "stxvw4x		46, %10, %3	\n\t"
-       "stxvw4x		47, %11, %3	\n"
+       "stxvd2x		40, 0, %3	\n\t"
+       "stxvd2x		41, %5, %3	\n\t"
+       "stxvd2x		42, %6, %3	\n\t"
+       "stxvd2x		43, %7, %3	\n\t"
+       "stxvd2x		44, %8, %3	\n\t"
+       "stxvd2x		45, %9, %3	\n\t"
+       "stxvd2x		46, %10, %3	\n\t"
+       "stxvd2x		47, %11, %3	\n"
 
      "#n=%1 x=%4=%2 y=%0=%3 o16=%5 o32=%6 o48=%7 o64=%8 o80=%9 o96=%10 o112=%11"
      :

kernel/power/sdot_microk_power8.c

@@ -57,22 +57,22 @@ static float sdot_kernel_16 (long n, float *x, float *y)
        "xxlxor		38, 38,	38	\n\t"
        "xxlxor		39, 39,	39	\n\t"
 
-       "lxvw4x		40, 0, %2	\n\t"
-       "lxvw4x		48, 0, %3	\n\t"
-       "lxvw4x		41, %10, %2	\n\t"
-       "lxvw4x		49, %10, %3	\n\t"
-       "lxvw4x		42, %11, %2	\n\t"
-       "lxvw4x		50, %11, %3	\n\t"
-       "lxvw4x		43, %12, %2	\n\t"
-       "lxvw4x		51, %12, %3	\n\t"
-       "lxvw4x		44, %13, %2	\n\t"
-       "lxvw4x		%x4, %13, %3	\n\t"
-       "lxvw4x		45, %14, %2	\n\t"
-       "lxvw4x		%x5, %14, %3	\n\t"
-       "lxvw4x		46, %15, %2	\n\t"
-       "lxvw4x		%x6, %15, %3	\n\t"
-       "lxvw4x		47, %16, %2	\n\t"
-       "lxvw4x		%x7, %16, %3	\n\t"
+       "lxvd2x		40, 0, %2	\n\t"
+       "lxvd2x		48, 0, %3	\n\t"
+       "lxvd2x		41, %10, %2	\n\t"
+       "lxvd2x		49, %10, %3	\n\t"
+       "lxvd2x		42, %11, %2	\n\t"
+       "lxvd2x		50, %11, %3	\n\t"
+       "lxvd2x		43, %12, %2	\n\t"
+       "lxvd2x		51, %12, %3	\n\t"
+       "lxvd2x		44, %13, %2	\n\t"
+       "lxvd2x		%x4, %13, %3	\n\t"
+       "lxvd2x		45, %14, %2	\n\t"
+       "lxvd2x		%x5, %14, %3	\n\t"
+       "lxvd2x		46, %15, %2	\n\t"
+       "lxvd2x		%x6, %15, %3	\n\t"
+       "lxvd2x		47, %16, %2	\n\t"
+       "lxvd2x		%x7, %16, %3	\n\t"
 
        "addi		%2, %2, 128	\n\t"
        "addi		%3, %3, 128	\n\t"
@@ -84,29 +84,29 @@ static float sdot_kernel_16 (long n, float *x, float *y)
      "1:				\n\t"
 
        "xvmaddasp	32, 40, 48	\n\t"
-       "lxvw4x		40, 0, %2	\n\t"
-       "lxvw4x		48, 0, %3	\n\t"
+       "lxvd2x		40, 0, %2	\n\t"
+       "lxvd2x		48, 0, %3	\n\t"
        "xvmaddasp	33, 41, 49	\n\t"
-       "lxvw4x		41, %10, %2	\n\t"
-       "lxvw4x		49, %10, %3	\n\t"
+       "lxvd2x		41, %10, %2	\n\t"
+       "lxvd2x		49, %10, %3	\n\t"
        "xvmaddasp	34, 42, 50	\n\t"
-       "lxvw4x		42, %11, %2	\n\t"
-       "lxvw4x		50, %11, %3	\n\t"
+       "lxvd2x		42, %11, %2	\n\t"
+       "lxvd2x		50, %11, %3	\n\t"
        "xvmaddasp	35, 43, 51	\n\t"
-       "lxvw4x		43, %12, %2	\n\t"
-       "lxvw4x		51, %12, %3	\n\t"
+       "lxvd2x		43, %12, %2	\n\t"
+       "lxvd2x		51, %12, %3	\n\t"
        "xvmaddasp	36, 44, %x4	\n\t"
-       "lxvw4x		44, %13, %2	\n\t"
-       "lxvw4x		%x4, %13, %3	\n\t"
+       "lxvd2x		44, %13, %2	\n\t"
+       "lxvd2x		%x4, %13, %3	\n\t"
        "xvmaddasp	37, 45, %x5	\n\t"
-       "lxvw4x		45, %14, %2	\n\t"
-       "lxvw4x		%x5, %14, %3	\n\t"
+       "lxvd2x		45, %14, %2	\n\t"
+       "lxvd2x		%x5, %14, %3	\n\t"
        "xvmaddasp	38, 46, %x6	\n\t"
-       "lxvw4x		46, %15, %2	\n\t"
-       "lxvw4x		%x6, %15, %3	\n\t"
+       "lxvd2x		46, %15, %2	\n\t"
+       "lxvd2x		%x6, %15, %3	\n\t"
        "xvmaddasp	39, 47, %x7	\n\t"
-       "lxvw4x		47, %16, %2	\n\t"
-       "lxvw4x		%x7, %16, %3	\n\t"
+       "lxvd2x		47, %16, %2	\n\t"
+       "lxvd2x		%x7, %16, %3	\n\t"
 
        "addi		%2, %2, 128	\n\t"
        "addi		%3, %3, 128	\n\t"

kernel/power/srot_microk_power8.c

@@ -57,15 +57,15 @@ static void srot_kernel_16 (long n, float *x, float *y, float c, float s)
        "xscvdpspn	37, %x14	\n\t"	// load s to all words
        "xxspltw		37, 37, 0	\n\t"
 
-       "lxvw4x		32, 0, %3	\n\t"	// load x
-       "lxvw4x		33, %15, %3	\n\t"
-       "lxvw4x		34, %16, %3	\n\t"
-       "lxvw4x		35, %17, %3	\n\t"
+       "lxvd2x		32, 0, %3	\n\t"	// load x
+       "lxvd2x		33, %15, %3	\n\t"
+       "lxvd2x		34, %16, %3	\n\t"
+       "lxvd2x		35, %17, %3	\n\t"
 
-       "lxvw4x		48, 0, %4	\n\t"	// load y
-       "lxvw4x		49, %15, %4	\n\t"
-       "lxvw4x		50, %16, %4	\n\t"
-       "lxvw4x		51, %17, %4	\n\t"
+       "lxvd2x		48, 0, %4	\n\t"	// load y
+       "lxvd2x		49, %15, %4	\n\t"
+       "lxvd2x		50, %16, %4	\n\t"
+       "lxvd2x		51, %17, %4	\n\t"
 
        "addi		%3, %3, 64	\n\t"
        "addi		%4, %4, 64	\n\t"
@@ -89,26 +89,26 @@ static void srot_kernel_16 (long n, float *x, float *y, float c, float s)
        "xvmulsp		44, 32, 37	\n\t"	// s * x
        "xvmulsp		45, 33, 37	\n\t"
 
-       "lxvw4x		32, 0, %3	\n\t"	// load x
-       "lxvw4x		33, %15, %3	\n\t"
+       "lxvd2x		32, 0, %3	\n\t"	// load x
+       "lxvd2x		33, %15, %3	\n\t"
 
        "xvmulsp		46, 34, 37	\n\t"
        "xvmulsp		47, 35, 37	\n\t"
 
-       "lxvw4x		34, %16, %3	\n\t"
-       "lxvw4x		35, %17, %3	\n\t"
+       "lxvd2x		34, %16, %3	\n\t"
+       "lxvd2x		35, %17, %3	\n\t"
 
        "xvmulsp		%x9, 48, 37	\n\t"	// s * y
        "xvmulsp		%x10, 49, 37	\n\t"
 
-       "lxvw4x		48, 0, %4	\n\t"	// load y
-       "lxvw4x		49, %15, %4	\n\t"
+       "lxvd2x		48, 0, %4	\n\t"	// load y
+       "lxvd2x		49, %15, %4	\n\t"
 
        "xvmulsp		%x11, 50, 37	\n\t"
        "xvmulsp		%x12, 51, 37	\n\t"
 
-       "lxvw4x		50, %16, %4	\n\t"
-       "lxvw4x		51, %17, %4	\n\t"
+       "lxvd2x		50, %16, %4	\n\t"
+       "lxvd2x		51, %17, %4	\n\t"
 
        "xvaddsp		40, 40, %x9	\n\t"	// c * x + s * y
        "xvaddsp		41, 41, %x10	\n\t"	// c * x + s * y
@@ -124,15 +124,15 @@ static void srot_kernel_16 (long n, float *x, float *y, float c, float s)
        "xvsubsp		%x7, %x7, 46	\n\t"	// c * y - s * x
        "xvsubsp		%x8, %x8, 47	\n\t"	// c * y - s * x
 
-       "stxvw4x		40, 0, %3	\n\t"	// store x
-       "stxvw4x		41, %15, %3	\n\t"
-       "stxvw4x		42, %16, %3	\n\t"
-       "stxvw4x		43, %17, %3	\n\t"
+       "stxvd2x		40, 0, %3	\n\t"	// store x
+       "stxvd2x		41, %15, %3	\n\t"
+       "stxvd2x		42, %16, %3	\n\t"
+       "stxvd2x		43, %17, %3	\n\t"
 
-       "stxvw4x		%x5, 0, %4	\n\t"	// store y
-       "stxvw4x		%x6, %15, %4	\n\t"
-       "stxvw4x		%x7, %16, %4	\n\t"
-       "stxvw4x		%x8, %17, %4	\n\t"
+       "stxvd2x		%x5, 0, %4	\n\t"	// store y
+       "stxvd2x		%x6, %15, %4	\n\t"
+       "stxvd2x		%x7, %16, %4	\n\t"
+       "stxvd2x		%x8, %17, %4	\n\t"
 
        "addi		%3, %3, 128	\n\t"
        "addi		%4, %4, 128	\n\t"
@@ -175,15 +175,15 @@ static void srot_kernel_16 (long n, float *x, float *y, float c, float s)
        "xvsubsp		%x7, %x7, 46	\n\t"	// c * y - s * x
        "xvsubsp		%x8, %x8, 47	\n\t"	// c * y - s * x
 
-       "stxvw4x		40, 0, %3	\n\t"	// store x
-       "stxvw4x		41, %15, %3	\n\t"
-       "stxvw4x		42, %16, %3	\n\t"
-       "stxvw4x		43, %17, %3	\n\t"
+       "stxvd2x		40, 0, %3	\n\t"	// store x
+       "stxvd2x		41, %15, %3	\n\t"
+       "stxvd2x		42, %16, %3	\n\t"
+       "stxvd2x		43, %17, %3	\n\t"
 
-       "stxvw4x		%x5, 0, %4	\n\t"	// store y
-       "stxvw4x		%x6, %15, %4	\n\t"
-       "stxvw4x		%x7, %16, %4	\n\t"
-       "stxvw4x		%x8, %17, %4	\n"
+       "stxvd2x		%x5, 0, %4	\n\t"	// store y
+       "stxvd2x		%x6, %15, %4	\n\t"
+       "stxvd2x		%x7, %16, %4	\n\t"
+       "stxvd2x		%x8, %17, %4	\n"
 
      "#n=%2 x=%0=%3 y=%1=%4 c=%13 s=%14 o16=%15 o32=%16 o48=%17\n"
      "#t0=%x5 t1=%x6 t2=%x7 t3=%x8 t4=%x9 t5=%x10 t6=%x11 t7=%x12"

kernel/power/sscal_microk_power8.c

@@ -44,14 +44,14 @@ static void sscal_kernel_16 (long n, float *x, float alpha)
        "xscvdpspn	%x3, %x3	\n\t"
        "xxspltw		%x3, %x3, 0	\n\t"
 
-       "lxvw4x		32, 0, %2	\n\t"
-       "lxvw4x		33, %4, %2	\n\t"
-       "lxvw4x		34, %5, %2	\n\t"
-       "lxvw4x		35, %6, %2	\n\t"
-       "lxvw4x		36, %7, %2	\n\t"
-       "lxvw4x		37, %8, %2	\n\t"
-       "lxvw4x		38, %9, %2	\n\t"
-       "lxvw4x		39, %10, %2	\n\t"
+       "lxvd2x		32, 0, %2	\n\t"
+       "lxvd2x		33, %4, %2	\n\t"
+       "lxvd2x		34, %5, %2	\n\t"
+       "lxvd2x		35, %6, %2	\n\t"
+       "lxvd2x		36, %7, %2	\n\t"
+       "lxvd2x		37, %8, %2	\n\t"
+       "lxvd2x		38, %9, %2	\n\t"
+       "lxvd2x		39, %10, %2	\n\t"
 
        "addi		%2, %2, 128	\n\t"
 
@@ -63,31 +63,31 @@ static void sscal_kernel_16 (long n, float *x, float alpha)
 
        "xvmulsp		40, 32, %x3	\n\t"
        "xvmulsp		41, 33, %x3	\n\t"
-       "lxvw4x		32, 0, %2	\n\t"
-       "lxvw4x		33, %4, %2	\n\t"
+       "lxvd2x		32, 0, %2	\n\t"
+       "lxvd2x		33, %4, %2	\n\t"
        "xvmulsp		42, 34, %x3	\n\t"
        "xvmulsp		43, 35, %x3	\n\t"
-       "lxvw4x		34, %5, %2	\n\t"
-       "lxvw4x		35, %6, %2	\n\t"
+       "lxvd2x		34, %5, %2	\n\t"
+       "lxvd2x		35, %6, %2	\n\t"
        "xvmulsp		44, 36, %x3	\n\t"
        "xvmulsp		45, 37, %x3	\n\t"
-       "lxvw4x		36, %7, %2	\n\t"
-       "lxvw4x		37, %8, %2	\n\t"
+       "lxvd2x		36, %7, %2	\n\t"
+       "lxvd2x		37, %8, %2	\n\t"
        "xvmulsp		46, 38, %x3	\n\t"
        "xvmulsp		47, 39, %x3	\n\t"
-       "lxvw4x		38, %9, %2	\n\t"
-       "lxvw4x		39, %10, %2	\n\t"
+       "lxvd2x		38, %9, %2	\n\t"
+       "lxvd2x		39, %10, %2	\n\t"
 
        "addi		%2, %2, -128	\n\t"
 
-       "stxvw4x		40, 0, %2	\n\t"
-       "stxvw4x		41, %4, %2	\n\t"
-       "stxvw4x		42, %5, %2	\n\t"
-       "stxvw4x		43, %6, %2	\n\t"
-       "stxvw4x		44, %7, %2	\n\t"
-       "stxvw4x		45, %8, %2	\n\t"
-       "stxvw4x		46, %9, %2	\n\t"
-       "stxvw4x		47, %10, %2	\n\t"
+       "stxvd2x		40, 0, %2	\n\t"
+       "stxvd2x		41, %4, %2	\n\t"
+       "stxvd2x		42, %5, %2	\n\t"
+       "stxvd2x		43, %6, %2	\n\t"
+       "stxvd2x		44, %7, %2	\n\t"
+       "stxvd2x		45, %8, %2	\n\t"
+       "stxvd2x		46, %9, %2	\n\t"
+       "stxvd2x		47, %10, %2	\n\t"
 
        "addi		%2, %2, 256	\n\t"
 
@@ -108,14 +108,14 @@ static void sscal_kernel_16 (long n, float *x, float alpha)
        "xvmulsp		46, 38, %x3	\n\t"
        "xvmulsp		47, 39, %x3	\n\t"
 
-       "stxvw4x		40, 0, %2	\n\t"
-       "stxvw4x		41, %4, %2	\n\t"
-       "stxvw4x		42, %5, %2	\n\t"
-       "stxvw4x		43, %6, %2	\n\t"
-       "stxvw4x		44, %7, %2	\n\t"
-       "stxvw4x		45, %8, %2	\n\t"
-       "stxvw4x		46, %9, %2	\n\t"
-       "stxvw4x		47, %10, %2	\n"
+       "stxvd2x		40, 0, %2	\n\t"
+       "stxvd2x		41, %4, %2	\n\t"
+       "stxvd2x		42, %5, %2	\n\t"
+       "stxvd2x		43, %6, %2	\n\t"
+       "stxvd2x		44, %7, %2	\n\t"
+       "stxvd2x		45, %8, %2	\n\t"
+       "stxvd2x		46, %9, %2	\n\t"
+       "stxvd2x		47, %10, %2	\n"
 
      "#n=%1 alpha=%3 x=%0=%2 o16=%4 o32=%5 o48=%6 o64=%7 o80=%8 o96=%9 o112=%10"
      :
@@ -150,14 +150,14 @@ static void sscal_kernel_16_zero (long n, float *x)
        ".p2align	5		\n"
      "1:				\n\t"
 
-       "stxvw4x		%x3, 0, %2	\n\t"
-       "stxvw4x		%x3, %4, %2	\n\t"
-       "stxvw4x		%x3, %5, %2	\n\t"
-       "stxvw4x		%x3, %6, %2	\n\t"
-       "stxvw4x		%x3, %7, %2	\n\t"
-       "stxvw4x		%x3, %8, %2	\n\t"
-       "stxvw4x		%x3, %9, %2	\n\t"
-       "stxvw4x		%x3, %10, %2	\n\t"
+       "stxvd2x		%x3, 0, %2	\n\t"
+       "stxvd2x		%x3, %4, %2	\n\t"
+       "stxvd2x		%x3, %5, %2	\n\t"
+       "stxvd2x		%x3, %6, %2	\n\t"
+       "stxvd2x		%x3, %7, %2	\n\t"
+       "stxvd2x		%x3, %8, %2	\n\t"
+       "stxvd2x		%x3, %9, %2	\n\t"
+       "stxvd2x		%x3, %10, %2	\n\t"
 
        "addi		%2, %2, 128	\n\t"
 

kernel/power/sswap_microk_power8.c

@@ -42,43 +42,43 @@ static void sswap_kernel_32 (long n, float *x, float *y)
        ".p2align	5		\n"
      "1:				\n\t"
 
-       "lxvw4x		32, 0, %4	\n\t"
-       "lxvw4x		33, %5, %4	\n\t"
-       "lxvw4x		34, %6, %4	\n\t"
-       "lxvw4x		35, %7, %4	\n\t"
-       "lxvw4x		36, %8, %4	\n\t"
-       "lxvw4x		37, %9, %4	\n\t"
-       "lxvw4x		38, %10, %4	\n\t"
-       "lxvw4x		39, %11, %4	\n\t"
+       "lxvd2x		32, 0, %4	\n\t"
+       "lxvd2x		33, %5, %4	\n\t"
+       "lxvd2x		34, %6, %4	\n\t"
+       "lxvd2x		35, %7, %4	\n\t"
+       "lxvd2x		36, %8, %4	\n\t"
+       "lxvd2x		37, %9, %4	\n\t"
+       "lxvd2x		38, %10, %4	\n\t"
+       "lxvd2x		39, %11, %4	\n\t"
 
-       "lxvw4x		40, 0, %3	\n\t"
-       "lxvw4x		41, %5, %3	\n\t"
-       "lxvw4x		42, %6, %3	\n\t"
-       "lxvw4x		43, %7, %3	\n\t"
-       "lxvw4x		44, %8, %3	\n\t"
-       "lxvw4x		45, %9, %3	\n\t"
-       "lxvw4x		46, %10, %3	\n\t"
-       "lxvw4x		47, %11, %3	\n\t"
+       "lxvd2x		40, 0, %3	\n\t"
+       "lxvd2x		41, %5, %3	\n\t"
+       "lxvd2x		42, %6, %3	\n\t"
+       "lxvd2x		43, %7, %3	\n\t"
+       "lxvd2x		44, %8, %3	\n\t"
+       "lxvd2x		45, %9, %3	\n\t"
+       "lxvd2x		46, %10, %3	\n\t"
+       "lxvd2x		47, %11, %3	\n\t"
 
-       "stxvw4x		32, 0, %3	\n\t"
-       "stxvw4x		33, %5, %3	\n\t"
-       "stxvw4x		34, %6, %3	\n\t"
-       "stxvw4x		35, %7, %3	\n\t"
-       "stxvw4x		36, %8, %3	\n\t"
-       "stxvw4x		37, %9, %3	\n\t"
-       "stxvw4x		38, %10, %3	\n\t"
-       "stxvw4x		39, %11, %3	\n\t"
+       "stxvd2x		32, 0, %3	\n\t"
+       "stxvd2x		33, %5, %3	\n\t"
+       "stxvd2x		34, %6, %3	\n\t"
+       "stxvd2x		35, %7, %3	\n\t"
+       "stxvd2x		36, %8, %3	\n\t"
+       "stxvd2x		37, %9, %3	\n\t"
+       "stxvd2x		38, %10, %3	\n\t"
+       "stxvd2x		39, %11, %3	\n\t"
 
        "addi		%3, %3, 128	\n\t"
 
-       "stxvw4x		40, 0, %4	\n\t"
-       "stxvw4x		41, %5, %4	\n\t"
-       "stxvw4x		42, %6, %4	\n\t"
-       "stxvw4x		43, %7, %4	\n\t"
-       "stxvw4x		44, %8, %4	\n\t"
-       "stxvw4x		45, %9, %4	\n\t"
-       "stxvw4x		46, %10, %4	\n\t"
-       "stxvw4x		47, %11, %4	\n\t"
+       "stxvd2x		40, 0, %4	\n\t"
+       "stxvd2x		41, %5, %4	\n\t"
+       "stxvd2x		42, %6, %4	\n\t"
+       "stxvd2x		43, %7, %4	\n\t"
+       "stxvd2x		44, %8, %4	\n\t"
+       "stxvd2x		45, %9, %4	\n\t"
+       "stxvd2x		46, %10, %4	\n\t"
+       "stxvd2x		47, %11, %4	\n\t"
 
        "addi		%4, %4, 128	\n\t"
 

utest/CMakeLists.txt

@@ -21,6 +21,10 @@ if(${CMAKE_SYSTEM_NAME} MATCHES "Linux")
 target_link_libraries(${OpenBLAS_utest_bin} m)
 endif()
 
+if (${CMAKE_SYSTEM_NAME} STREQUAL "WindowsStore")
+set_target_properties( ${OpenBLAS_utest_bin} PROPERTIES COMPILE_DEFINITIONS "_CRT_SECURE_NO_WARNINGS")
+endif()
+
 #Set output for utest
 set_target_properties( ${OpenBLAS_utest_bin} PROPERTIES RUNTIME_OUTPUT_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR})
 foreach (OUTPUTCONFIG ${CMAKE_CONFIGURATION_TYPES})