Revert "Use usleep instead of sched_yield by default"

Return a somewhat sane default value for L2 cache size if cpuid retur… (#1611 )
* Return a somewhat sane default value for L2 cache size if cpuid returned something unexpected Fixes #1610, the KVM hypervisor on Google Chromebooks returning zero for CPUID 0x80000006, causing DYNAMIC_ARCH builds of OpenBLAS to hang
2018-06-11 17:05:27 +02:00 · 2018-06-11 13:26:19 +02:00 · 2018-06-09 19:57:33 +02:00 · 2018-06-09 17:54:36 +02:00 · 2018-06-09 12:42:34 +02:00 · 2018-06-07 15:29:13 +01:00
192 changed files with 13733 additions and 527 deletions
@@ -7,6 +7,7 @@ language: c
 jobs:
  include:
    - &test-ubuntu
+      os: linux
      stage: test
      compiler: gcc
      addons:
@@ -57,7 +58,8 @@ jobs:
        - TARGET_BOX=LINUX32
        - BTYPE="BINARY=32"

-    - stage: test
+    - os: linux
+      stage: test
      compiler: gcc
      addons:
        apt:
@@ -77,6 +79,7 @@ jobs:
    # which is slower than container-based infrastructure used for jobs
    # that don't require sudo.
    - &test-alpine
+      os: linux
      stage: test
      dist: trusty
      sudo: true
@@ -120,6 +123,7 @@ jobs:
        - BTYPE="BINARY=64 NO_AFFINITY=1 USE_OPENMP=0 NO_LAPACK=0 TARGET=core2"

    - &test-cmake
+      os: linux
      stage: test
      compiler: clang
      addons:
@@ -147,6 +151,23 @@ jobs:
      env:
        - CMAKE=1

+    - &test-macos
+      os: osx
+      stage: test
+      osx_image: xcode8
+      before_script:
+        - COMMON_FLAGS="DYNAMIC_ARCH=1 TARGET=NEHALEM NUM_THREADS=32"
+        - brew update
+        - brew install gcc # for gfortran
+      script:
+        - travis_wait 45 make QUIET_MAKE=1 $COMMON_FLAGS $BTYPE
+      env:
+        - BTYPE="BINARY=64 INTERFACE64=1"
+
+    - <<: *test-macos
+      env:
+        - BTYPE="BINARY=32"
+
 # whitelist
 branches:
  only:
@@ -6,7 +6,7 @@ cmake_minimum_required(VERSION 2.8.5)
 project(OpenBLAS C ASM)
 set(OpenBLAS_MAJOR_VERSION 0)
 set(OpenBLAS_MINOR_VERSION 3)
-set(OpenBLAS_PATCH_VERSION 0.dev)
+set(OpenBLAS_PATCH_VERSION 1.dev)
 set(OpenBLAS_VERSION "${OpenBLAS_MAJOR_VERSION}.${OpenBLAS_MINOR_VERSION}.${OpenBLAS_PATCH_VERSION}")

 # Adhere to GNU filesystem layout conventions
@@ -294,9 +294,10 @@ endif

 lapack-test :
 	(cd $(NETLIB_LAPACK_DIR)/TESTING && rm -f x* *.out)
-	$(MAKE) -j 1 -C $(NETLIB_LAPACK_DIR)/TESTING xeigtstc  xeigtstd  xeigtsts  xeigtstz  xlintstc  xlintstd  xlintstds  xlintstrfd  xlintstrfz  xlintsts  xlintstz  xlintstzc xlintstrfs xlintstrfc
+	$(MAKE) -j 1 -C $(NETLIB_LAPACK_DIR)/TESTING/EIG xeigtstc  xeigtstd  xeigtsts  xeigtstz 
+	$(MAKE) -j 1 -C $(NETLIB_LAPACK_DIR)/TESTING/LIN xlintstc  xlintstd  xlintstds  xlintstrfd  xlintstrfz  xlintsts  xlintstz  xlintstzc xlintstrfs xlintstrfc
 ifneq ($(CROSS), 1)
-	( cd $(NETLIB_LAPACK_DIR)/INSTALL; ./testlsame; ./testslamch; ./testdlamch; \
+	( cd $(NETLIB_LAPACK_DIR)/INSTALL; make all; ./testlsame; ./testslamch; ./testdlamch; \
        ./testsecond; ./testdsecnd; ./testieee; ./testversion )
 	(cd $(NETLIB_LAPACK_DIR); ./lapack_testing.py -r )
 endif
@@ -308,9 +309,9 @@ lapack-runtest:


 blas-test:
-	(cd $(NETLIB_LAPACK_DIR)/BLAS && rm -f x* *.out)
+	(cd $(NETLIB_LAPACK_DIR)/BLAS/TESTING && rm -f x* *.out)
 	$(MAKE) -j 1 -C $(NETLIB_LAPACK_DIR) blas_testing
-	(cd $(NETLIB_LAPACK_DIR)/BLAS && cat *.out)
+	(cd $(NETLIB_LAPACK_DIR)/BLAS/TESTING && cat *.out)


 dummy :
@@ -96,8 +96,9 @@ endif

 #Generating openblas.pc
 	@echo Generating openblas.pc in "$(DESTDIR)$(OPENBLAS_PKGCONFIG_DIR)"
-	@echo 'libdir='$(OPENBLAS_LIBRARY_DIR) >> "$(DESTDIR)$(OPENBLAS_PKGCONFIG_DIR)/openblas.pc"
+	@echo 'libdir='$(OPENBLAS_LIBRARY_DIR) > "$(DESTDIR)$(OPENBLAS_PKGCONFIG_DIR)/openblas.pc"
 	@echo 'includedir='$(OPENBLAS_INCLUDE_DIR) >> "$(DESTDIR)$(OPENBLAS_PKGCONFIG_DIR)/openblas.pc"
+	@echo 'openblas_config= USE_64BITINT='$(USE_64BITINT) 'DYNAMIC_ARCH='$(DYNAMIC_ARCH) 'NO_CBLAS='$(NO_CBLAS) 'NO_LAPACK='$(NO_LAPACK) 'NO_LAPACKE='$(NO_LAPACKE) 'NO_AFFINITY='$(NO_AFFINITY) 'USE_OPENMP='$(USE_OPENMP) $(CORE) 'MAX_THREADS='$(NUM_THREADS)>> "$(DESTDIR)$(OPENBLAS_PKGCONFIG_DIR)/openblas.pc"
 	@echo 'version='$(VERSION) >> "$(DESTDIR)$(OPENBLAS_PKGCONFIG_DIR)/openblas.pc"
 	@echo 'extralib='$(EXTRALIB) >> "$(DESTDIR)$(OPENBLAS_PKGCONFIG_DIR)/openblas.pc"
 	@cat openblas.pc.in >> "$(DESTDIR)$(OPENBLAS_PKGCONFIG_DIR)/openblas.pc"
@@ -17,6 +17,10 @@ ifdef CPUIDEMU
 EXFLAGS = -DCPUIDEMU -DVENDOR=99
 endif

+ifeq ($(TARGET), 1004K)
+TARGET_FLAGS = -mips32r2
+endif
+
 ifeq ($(TARGET), P5600)
 TARGET_FLAGS = -mips32r5
 endif
@@ -3,7 +3,7 @@
 #

 # This library's version
-VERSION = 0.3.0.dev
+VERSION = 0.3.1.dev

 # If you set the suffix, the library name will be libopenblas_$(LIBNAMESUFFIX).a
 # and libopenblas_$(LIBNAMESUFFIX).so. Meanwhile, the soname in shared library
@@ -60,6 +60,13 @@ VERSION = 0.3.0.dev
 # automatically detected by the the script.
 # NUM_THREADS = 24

+# If you have enabled USE_OPENMP and your application would call
+# OpenBLAS's calculation API from multi threads, please comment it in.
+# This flag defines how many instances of OpenBLAS's calculation API can 
+# actually run in parallel. If more threads call OpenBLAS's calculation API,
+# they need to wait for the preceding API calls to finish or risk data corruption.
+# NUM_PARALLEL = 2
+
 # if you don't need to install the static library, please comment it in.
 # NO_STATIC = 1

@@ -62,6 +62,9 @@ ifeq ($(BINARY), 32)
 ifeq ($(TARGET), HASWELL)
 GETARCH_FLAGS := -DFORCE_NEHALEM
 endif
+ifeq ($(TARGET), SKYLAKEX)
+GETARCH_FLAGS := -DFORCE_NEHALEM
+endif
 ifeq ($(TARGET), SANDYBRIDGE)
 GETARCH_FLAGS := -DFORCE_NEHALEM
 endif
@@ -95,6 +98,9 @@ ifeq ($(BINARY), 32)
 ifeq ($(TARGET_CORE), HASWELL)
 GETARCH_FLAGS := -DFORCE_NEHALEM
 endif
+ifeq ($(TARGET_CORE), SKYLAKEX)
+GETARCH_FLAGS := -DFORCE_NEHALEM
+endif
 ifeq ($(TARGET_CORE), SANDYBRIDGE)
 GETARCH_FLAGS := -DFORCE_NEHALEM
 endif
@@ -141,6 +147,10 @@ ifeq ($(NO_AVX2), 1)
 GETARCH_FLAGS	+= -DNO_AVX2
 endif

+ifeq ($(NO_AVX512), 1)
+GETARCH_FLAGS	+= -DNO_AVX512
+endif
+
 ifeq ($(DEBUG), 1)
 GETARCH_FLAGS	+= -g
 endif
@@ -184,6 +194,10 @@ endif

 endif

+ifndef NUM_PARALLEL
+NUM_PARALLEL = 1
+endif
+
 ifndef NUM_THREADS
 NUM_THREADS = $(NUM_CORES)
 endif
@@ -433,7 +447,7 @@ CCOMMON_OPT    += -fopenmp
 endif

 ifeq ($(C_COMPILER), INTEL)
-CCOMMON_OPT    += -openmp
+CCOMMON_OPT    += -fopenmp
 endif

 ifeq ($(C_COMPILER), PGI)
@@ -465,6 +479,11 @@ endif
 ifneq ($(NO_AVX2), 1)
 DYNAMIC_CORE += HASWELL ZEN
 endif
+ifneq ($(NO_AVX512), 1)
+ifneq ($(NO_AVX2), 1)
+DYNAMIC_CORE += SKYLAKEX
+endif
+endif
 endif

 # If DYNAMIC_CORE is not set, DYNAMIC_ARCH cannot do anything, so force it to empty
@@ -564,9 +583,14 @@ CCOMMON_OPT += -march=mips64
 FCOMMON_OPT += -march=mips64
 endif

+ifeq ($(CORE), 1004K)
+CCOMMON_OPT += -mips32r2   $(MSA_FLAGS)
+FCOMMON_OPT += -mips32r2   $(MSA_FLAGS)
+endif
+
 ifeq ($(CORE), P5600)
-CCOMMON_OPT += -mips32r5 -mnan=2008 -mtune=p5600 $(MSA_FLAGS)
-FCOMMON_OPT += -mips32r5 -mnan=2008 -mtune=p5600 $(MSA_FLAGS)
+CCOMMON_OPT += -mips32r5 -mnan=2008 -mtune=p5600  $(MSA_FLAGS)
+FCOMMON_OPT += -mips32r5 -mnan=2008 -mtune=p5600  $(MSA_FLAGS)
 endif

 ifeq ($(CORE), I6400)
@@ -713,7 +737,7 @@ FCOMMON_OPT += -i8
 endif
 endif
 ifeq ($(USE_OPENMP), 1)
-FCOMMON_OPT += -openmp
+FCOMMON_OPT += -fopenmp
 endif
 endif

@@ -915,6 +939,10 @@ ifeq ($(NO_AVX2), 1)
 CCOMMON_OPT	+= -DNO_AVX2
 endif

+ifeq ($(NO_AVX512), 1)
+CCOMMON_OPT	+= -DNO_AVX512
+endif
+
 ifdef SMP
 CCOMMON_OPT	+= -DSMP_SERVER

@@ -961,6 +989,8 @@ endif

 CCOMMON_OPT	+= -DMAX_CPU_NUMBER=$(NUM_THREADS)

+CCOMMON_OPT	+= -DMAX_PARALLEL_NUMBER=$(NUM_PARALLEL)
+
 ifdef USE_SIMPLE_THREADED_LEVEL3
 CCOMMON_OPT	+= -DUSE_SIMPLE_THREADED_LEVEL3
 endif
@@ -1219,6 +1249,7 @@ export MSA_FLAGS
 export KERNELDIR
 export FUNCTION_PROFILE
 export TARGET_CORE
+export NO_AVX512

 export SGEMM_UNROLL_M
 export SGEMM_UNROLL_N
@@ -20,6 +20,7 @@ DUNNINGTON
 NEHALEM
 SANDYBRIDGE
 HASWELL
+SKYLAKEX
 ATOM

 b)AMD CPU:
@@ -56,6 +57,7 @@ CELL

 3.MIPS CPU:
 P5600
+1004K

 4.MIPS64 CPU:
 SICORTEX
@@ -201,6 +201,21 @@ $architecture = zarch  if ($data =~ /ARCH_ZARCH/);
 $binformat    = bin32;
 $binformat    = bin64  if ($data =~ /BINARY_64/);

+$no_avx512= 0;
+if (($architecture eq "x86") || ($architecture eq "x86_64")) {
+    $code = '"vbroadcastss -4 * 4(%rsi), %zmm2"';
+    print $tmpf "int main(void){ __asm__ volatile($code); }\n";
+    $args = " -o $tmpf.o -x c $tmpf";
+    my @cmd = ("$compiler_name $args");
+    system(@cmd) == 0;
+    if ($? != 0) {
+	$no_avx512 = 1;
+    } else {
+	$no_avx512 = 0;
+    }
+    unlink("tmpf.o");
+}
+
 $data = `$compiler_name -S ctest1.c && grep globl ctest1.s | head -n 1 && rm -f ctest1.s`;

 $data =~ /globl\s([_\.]*)(.*)/;
@@ -288,6 +303,7 @@ print MAKEFILE "CROSS=1\n" if $cross != 0;
 print MAKEFILE "CEXTRALIB=$linker_L $linker_l $linker_a\n";
 print MAKEFILE "HAVE_MSA=1\n" if $have_msa eq 1;
 print MAKEFILE "MSA_FLAGS=$msa_flags\n" if $have_msa eq 1;
+print MAKEFILE "NO_AVX512=1\n" if $no_avx512 eq 1;

 $os           =~ tr/[a-z]/[A-Z]/;
 $architecture =~ tr/[a-z]/[A-Z]/;
@@ -56,6 +56,9 @@ if (DYNAMIC_ARCH)
    if (NOT NO_AVX2)
      set(DYNAMIC_CORE ${DYNAMIC_CORE} HASWELL ZEN)
    endif ()
+    if (NOT NO_AVX512)
+      set(DYNAMIC_CORE ${DYNAMIC_CORE} SKYLAKEX)
+    endif ()
  endif ()

  if (NOT DYNAMIC_CORE)
@@ -1,6 +1,7 @@
 libdir=@CMAKE_INSTALL_FULL_LIBDIR@
 includedir=@CMAKE_INSTALL_FULL_INCLUDEDIR@

+openblas_config=USE_64BITINT=@USE_64BITINT@ NO_CBLAS=@NO_CBLAS@ NO_LAPACK=@NO_LAPACK@ NO_LAPACKE=@NO_LAPACKE@ DYNAMIC_ARCH=@DYNAMIC_ARCH@ NO_AFFINITY=@NO_AFFINITY@ USE_OPENMP=@USE_OPENMP@ @CORE@ MAX_THREADS=@NUM_THREADS@ 
 Name: OpenBLAS
 Description: OpenBLAS is an optimized BLAS library based on GotoBLAS2 1.13 BSD version
 Version: @OPENBLAS_VERSION@
@@ -33,7 +33,7 @@ endif ()
 if (DEFINED BINARY AND DEFINED TARGET AND BINARY EQUAL 32)
  message(STATUS "Compiling a ${BINARY}-bit binary.")
  set(NO_AVX 1)
-  if (${TARGET} STREQUAL "HASWELL" OR ${TARGET} STREQUAL "SANDYBRIDGE")
+  if (${TARGET} STREQUAL "HASWELL" OR ${TARGET} STREQUAL "SANDYBRIDGE" OR ${TARGET} STREQUAL "SKYLAKEX")
    set(TARGET "NEHALEM")
  endif ()
  if (${TARGET} STREQUAL "BULLDOZER" OR ${TARGET} STREQUAL "PILEDRIVER" OR ${TARGET} STREQUAL "ZEN")
@@ -96,6 +96,10 @@ if (NOT CMAKE_CROSSCOMPILING)

 endif()

+if (NOT DEFINED NUM_PARALLEL)
+  set(NUM_PARALLEL 1)
+endif()
+
 if (NOT DEFINED NUM_THREADS)
  if (DEFINED NUM_CORES AND NOT NUM_CORES EQUAL 0)
    # HT?
@@ -224,6 +228,8 @@ endif ()

 set(CCOMMON_OPT "${CCOMMON_OPT} -DMAX_CPU_NUMBER=${NUM_THREADS}")

+set(CCOMMON_OPT "${CCOMMON_OPT} -DMAX_PARALLEL_NUMBER=${NUM_PARALLEL}")
+
 if (USE_SIMPLE_THREADED_LEVEL3)
  set(CCOMMON_OPT "${CCOMMON_OPT} -DUSE_SIMPLE_THREADED_LEVEL3")
 endif ()
@@ -66,3 +66,12 @@ else()
  set(BINARY32 1)
 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 ${PROJECT_BINARY_DIR}/avx512.tmp RESULT_VARIABLE NO_AVX512)
+if (NO_AVX512 EQUAL 1)
+set (CCOMMON_OPT "${CCOMMON_OPT} -DNO_AVX512")
+endif()
+  file(REMOVE "avx512.tmp" "avx512.o")
+endif()
+
@@ -179,7 +179,7 @@ extern "C" {

 #define ALLOCA_ALIGN 63UL

-#define NUM_BUFFERS (MAX_CPU_NUMBER * 2)
+#define NUM_BUFFERS (MAX_CPU_NUMBER * 2 * MAX_PARALLEL_NUMBER)

 #ifdef NEEDBUNDERSCORE
 #define BLASFUNC(FUNC) FUNC##_
@@ -642,6 +642,7 @@ void gotoblas_profile_init(void);
 void gotoblas_profile_quit(void);

 #ifdef USE_OPENMP
+
 #ifndef C_MSVC
 int omp_in_parallel(void);
 int omp_get_num_procs(void);
@@ -649,6 +650,21 @@ int omp_get_num_procs(void);
 __declspec(dllimport) int __cdecl omp_in_parallel(void);
 __declspec(dllimport) int __cdecl omp_get_num_procs(void);
 #endif
+
+#if (__STDC_VERSION__ >= 201112L)
+#if defined(C_GCC) && ( __GNUC__ < 7) 
+// workaround for GCC bug 65467
+#ifndef _Atomic
+#define _Atomic volatile
+#endif
+#endif
+#include <stdatomic.h>
+#else
+#ifndef _Atomic
+#define _Atomic volatile
+#endif
+#endif
+
 #else
 #ifdef __ELF__
 int omp_in_parallel  (void) __attribute__ ((weak));
@@ -178,7 +178,13 @@ static __inline int blas_quickdivide(unsigned int x, unsigned int y){
  result = x/y;
  return result;
 #else
-
+#if (MAX_CPU_NUMBER > 64)
+  if ( y > 64) {
+	  result = x/y;
+	  return result;
+  }
+#endif
+	
  y = blas_quick_divide_table[y];

  __asm__ __volatile__  ("mull %0" :"=d" (result) :"a"(x), "0" (y));
@@ -196,6 +196,13 @@ static __inline int blas_quickdivide(unsigned int x, unsigned int y){

  if (y <= 1) return x;

+#if (MAX_CPU_NUMBER > 64)  
+  if (y > 64) { 
+	  result = x / y;
+	  return result;
+  }
+#endif
+	
  y = blas_quick_divide_table[y];

  __asm__ __volatile__  ("mull %0" :"=d" (result) :"a"(x), "0" (y));
@@ -115,6 +115,7 @@
 #define CORE_STEAMROLLER 25
 #define CORE_EXCAVATOR   26
 #define CORE_ZEN         27
+#define CORE_SKYLAKEX    28

 #define HAVE_SSE      (1 <<  0)
 #define HAVE_SSE2     (1 <<  1)
@@ -137,6 +138,7 @@
 #define HAVE_AVX      (1 <<  18)
 #define HAVE_FMA4     (1 <<  19)
 #define HAVE_FMA3     (1 <<  20)
+#define HAVE_AVX512VL (1 <<  21)

 #define CACHE_INFO_L1_I     1
 #define CACHE_INFO_L1_D     2
@@ -211,5 +213,6 @@ typedef struct {
 #define CPUTYPE_STEAMROLLER 		49
 #define CPUTYPE_EXCAVATOR 		50
 #define CPUTYPE_ZEN 			51
+#define CPUTYPE_SKYLAKEX		52

 #endif
@@ -121,7 +121,7 @@ int detect(void)
 			return CPU_VULCAN;
 		else if (strstr(cpu_part, "0x0a1") && strstr(cpu_implementer, "0x43"))
 			return CPU_THUNDERX;
-		else if (strstr(cpu_part, "0xFFF") && strstr(cpu_implementer, "0x43")) /* TODO */
+		else if (strstr(cpu_part, "0x0af") && strstr(cpu_implementer, "0x43"))
 			return CPU_THUNDERX2T99;
 	}

@@ -72,10 +72,12 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

 #define CPU_UNKNOWN     0
 #define CPU_P5600       1
+#define CPU_1004K	2

 static char *cpuname[] = {
  "UNKOWN",
-  "P5600"
+  "P5600",
+  "1004K"
 };

 int detect(void){
@@ -90,7 +92,7 @@ int detect(void){
    if (!strncmp("cpu", buffer, 3)){
 	p = strchr(buffer, ':') + 2;
 #if 0
-	fprintf(stderr, "%s\n", p);
+	fprintf(stderr, "%s \n", p);
 #endif
 	break;
      }
@@ -99,43 +101,13 @@ int detect(void){
  fclose(infile);

  if(p != NULL){
-  if (strstr(p, "Loongson-3A")){
-    return CPU_LOONGSON3A;
-  }else if(strstr(p, "Loongson-3B")){
-    return CPU_LOONGSON3B;
-  }else if (strstr(p, "Loongson-3")){
-    infile = fopen("/proc/cpuinfo", "r");
-    p = (char *)NULL;
-    while (fgets(buffer, sizeof(buffer), infile)){
-      if (!strncmp("system type", buffer, 11)){
-	p = strchr(buffer, ':') + 2;
-	break;
-      }
-    }
-    fclose(infile);
-    if (strstr(p, "loongson3a"))
-      return CPU_LOONGSON3A;
-  }else{
+  if (strstr(p, "5600")) {
+    return CPU_P5600;
+  } else if (strstr(p, "1004K")) {
+    return CPU_1004K;
+  } else  
    return CPU_UNKNOWN;
  }
-  }
-  //Check model name for Loongson3
-  infile = fopen("/proc/cpuinfo", "r");
-  p = (char *)NULL;
-  while (fgets(buffer, sizeof(buffer), infile)){
-    if (!strncmp("model name", buffer, 10)){
-      p = strchr(buffer, ':') + 2;
-      break;
-    }
-  }
-  fclose(infile);
-  if(p != NULL){
-  if (strstr(p, "Loongson-3A")){
-    return CPU_LOONGSON3A;
-  }else if(strstr(p, "Loongson-3B")){
-    return CPU_LOONGSON3B;
-  }
-  }
 #endif
    return CPU_UNKNOWN;
 }
@@ -149,7 +121,7 @@ void get_architecture(void){
 }

 void get_subarchitecture(void){
-  if(detect()==CPU_P5600){
+  if(detect()==CPU_P5600|| detect()==CPU_1004K){
    printf("P5600");
  }else{
    printf("UNKNOWN");
@@ -170,6 +142,14 @@ void get_cpuconfig(void){
    printf("#define DTB_DEFAULT_ENTRIES 64\n");
    printf("#define DTB_SIZE 4096\n");
    printf("#define L2_ASSOCIATIVE 8\n");
+  } else if (detect()==CPU_1004K) {
+    printf("#define MIPS1004K\n");
+    printf("#define L1_DATA_SIZE 32768\n");
+    printf("#define L1_DATA_LINESIZE 32\n");
+    printf("#define L2_SIZE 26144\n");
+    printf("#define DTB_DEFAULT_ENTRIES 8\n");
+    printf("#define DTB_SIZE 4096\n");
+    printf("#define L2_ASSOCIATIVE 4\n");
  }else{
    printf("#define UNKNOWN\n");
  }
@@ -178,6 +158,8 @@ void get_cpuconfig(void){
 void get_libname(void){
  if(detect()==CPU_P5600) {
    printf("p5600\n");
+  } else if (detect()==CPU_1004K) {
+    printf("1004K\n");
  }else{
    printf("mips\n");
  }
@@ -50,6 +50,8 @@
 #ifdef NO_AVX
 #define CPUTYPE_HASWELL CPUTYPE_NEHALEM
 #define CORE_HASWELL CORE_NEHALEM
+#define CPUTYPE_SKYLAKEX CPUTYPE_NEHALEM
+#define CORE_SKYLAKEX CORE_NEHALEM
 #define CPUTYPE_SANDYBRIDGE CPUTYPE_NEHALEM
 #define CORE_SANDYBRIDGE CORE_NEHALEM
 #define CPUTYPE_BULLDOZER CPUTYPE_BARCELONA
@@ -1299,6 +1301,19 @@ int get_cpuname(void){
          else
 	    return CPUTYPE_NEHALEM;
 	case 5:
+	  // Skylake X
+#ifndef NO_AVX512
+	  return CPUTYPE_SKYLAKEX;
+#else
+	  if(support_avx())
+#ifndef NO_AVX2
+	  return CPUTYPE_HASWELL;
+#else
+	  return CPUTYPE_SANDYBRIDGE;
+#endif
+	  else
+	  return CPUTYPE_NEHALEM;
+#endif			
        case 14:
 	  // Skylake
          if(support_avx())
@@ -1556,6 +1571,7 @@ static char *cpuname[] = {
  "STEAMROLLER",
  "EXCAVATOR",
  "ZEN",
+  "SKYLAKEX"	
 };

 static char *lowercpuname[] = {
@@ -1610,6 +1626,7 @@ static char *lowercpuname[] = {
  "steamroller",
  "excavator",
  "zen",
+  "skylakex"
 };

 static char *corename[] = {
@@ -1641,6 +1658,7 @@ static char *corename[] = {
  "STEAMROLLER",
  "EXCAVATOR",
  "ZEN",
+  "SKYLAKEX"	
 };

 static char *corename_lower[] = {
@@ -1672,6 +1690,7 @@ static char *corename_lower[] = {
  "steamroller",
  "excavator",
  "zen",
+  "skylakex"	
 };


@@ -1860,6 +1879,19 @@ int get_coretype(void){
          else
 	    return CORE_NEHALEM;
 	case 5:
+	 // Skylake X
+#ifndef NO_AVX512
+	    return CORE_SKYLAKEX;
+#else
+	  if(support_avx())
+#ifndef NO_AVX2
+	    return CORE_HASWELL;
+#else
+	    return CORE_SANDYBRIDGE;
+#endif
+	  else
+	    return CORE_NEHALEM;
+#endif			
 	case 14:
 	  // Skylake
          if(support_avx())
@@ -362,7 +362,7 @@ cgemm_ct.$(SUFFIX) : gemm.c level3.c  ../../param.h
 	$(CC) $(CFLAGS) $(BLOCKS) -c -UDOUBLE -DCOMPLEX -DCT $< -o $(@F)

 cgemm_cr.$(SUFFIX) : gemm.c level3.c  ../../param.h
-	$(CC) $(CFLAGS) $(BLOCKS) -c -UDOUBLE -DCOMPLEX -DCR $< -o $(@F)
+	$(CC) $(CFLAGS) $(BLOCKS) -c -UDOUBLE -DCOMPLEX -DCR=CR $< -o $(@F)

 cgemm_cc.$(SUFFIX) : gemm.c level3.c  ../../param.h
 	$(CC) $(CFLAGS) $(BLOCKS) -c -UDOUBLE -DCOMPLEX -DCC $< -o $(@F)
@@ -410,7 +410,7 @@ zgemm_ct.$(SUFFIX) : gemm.c level3.c ../../param.h
 	$(CC) $(CFLAGS) $(BLOCKS) -c -DDOUBLE -DCOMPLEX -DCT $< -o $(@F)

 zgemm_cr.$(SUFFIX) : gemm.c level3.c ../../param.h
-	$(CC) $(CFLAGS) $(BLOCKS) -c -DDOUBLE -DCOMPLEX -DCR $< -o $(@F)
+	$(CC) $(CFLAGS) $(BLOCKS) -c -DDOUBLE -DCOMPLEX -DCR=CR $< -o $(@F)

 zgemm_cc.$(SUFFIX) : gemm.c level3.c ../../param.h
 	$(CC) $(CFLAGS) $(BLOCKS) -c -DDOUBLE -DCOMPLEX -DCC $< -o $(@F)
@@ -458,7 +458,7 @@ xgemm_ct.$(SUFFIX) : gemm.c level3.c ../../param.h
 	$(CC) $(CFLAGS) $(BLOCKS) -c -DXDOUBLE -DCOMPLEX -DCT $< -o $(@F)

 xgemm_cr.$(SUFFIX) : gemm.c level3.c ../../param.h
-	$(CC) $(CFLAGS) $(BLOCKS) -c -DXDOUBLE -DCOMPLEX -DCR $< -o $(@F)
+	$(CC) $(CFLAGS) $(BLOCKS) -c -DXDOUBLE -DCOMPLEX -DCR=CR $< -o $(@F)

 xgemm_cc.$(SUFFIX) : gemm.c level3.c ../../param.h
 	$(CC) $(CFLAGS) $(BLOCKS) -c -DXDOUBLE -DCOMPLEX -DCC $< -o $(@F)
@@ -558,7 +558,7 @@ cgemm_thread_ct.$(SUFFIX) : gemm.c level3_thread.c ../../param.h
 	$(CC) $(CFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -UDOUBLE -DCOMPLEX -DCT $< -o $(@F)

 cgemm_thread_cr.$(SUFFIX) : gemm.c level3_thread.c ../../param.h
-	$(CC) $(CFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -UDOUBLE -DCOMPLEX -DCR $< -o $(@F)
+	$(CC) $(CFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -UDOUBLE -DCOMPLEX -DCR=CR $< -o $(@F)

 cgemm_thread_cc.$(SUFFIX) : gemm.c level3_thread.c ../../param.h
 	$(CC) $(CFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -UDOUBLE -DCOMPLEX -DCC $< -o $(@F)
@@ -606,7 +606,7 @@ zgemm_thread_ct.$(SUFFIX) : gemm.c level3_thread.c ../../param.h
 	$(CC) $(CFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DDOUBLE -DCOMPLEX -DCT $< -o $(@F)

 zgemm_thread_cr.$(SUFFIX) : gemm.c level3_thread.c ../../param.h
-	$(CC) $(CFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DDOUBLE -DCOMPLEX -DCR $< -o $(@F)
+	$(CC) $(CFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DDOUBLE -DCOMPLEX -DCR=CR $< -o $(@F)

 zgemm_thread_cc.$(SUFFIX) : gemm.c level3_thread.c ../../param.h
 	$(CC) $(CFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DDOUBLE -DCOMPLEX -DCC $< -o $(@F)
@@ -654,7 +654,7 @@ xgemm_thread_ct.$(SUFFIX) : gemm.c level3_thread.c ../../param.h
 	$(CC) $(CFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DXDOUBLE -DCOMPLEX -DCT $< -o $(@F)

 xgemm_thread_cr.$(SUFFIX) : gemm.c level3_thread.c ../../param.h
-	$(CC) $(CFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DXDOUBLE -DCOMPLEX -DCR $< -o $(@F)
+	$(CC) $(CFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DXDOUBLE -DCOMPLEX -DCR=CR $< -o $(@F)

 xgemm_thread_cc.$(SUFFIX) : gemm.c level3_thread.c ../../param.h
 	$(CC) $(CFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DXDOUBLE -DCOMPLEX -DCC $< -o $(@F)
@@ -1821,7 +1821,7 @@ cgemm3m_ct.$(SUFFIX) : gemm3m.c gemm3m_level3.c
 	$(CC) $(CFLAGS) $(BLOCKS) -c -UDOUBLE -DCOMPLEX -DCT $< -o $(@F)

 cgemm3m_cr.$(SUFFIX) : gemm3m.c gemm3m_level3.c
-	$(CC) $(CFLAGS) $(BLOCKS) -c -UDOUBLE -DCOMPLEX -DCR $< -o $(@F)
+	$(CC) $(CFLAGS) $(BLOCKS) -c -UDOUBLE -DCOMPLEX -DCR=CR $< -o $(@F)

 cgemm3m_cc.$(SUFFIX) : gemm3m.c gemm3m_level3.c
 	$(CC) $(CFLAGS) $(BLOCKS) -c -UDOUBLE -DCOMPLEX -DCC $< -o $(@F)
@@ -1869,7 +1869,7 @@ zgemm3m_ct.$(SUFFIX) : gemm3m.c gemm3m_level3.c
 	$(CC) $(CFLAGS) $(BLOCKS) -c -DDOUBLE -DCOMPLEX -DCT $< -o $(@F)

 zgemm3m_cr.$(SUFFIX) : gemm3m.c gemm3m_level3.c
-	$(CC) $(CFLAGS) $(BLOCKS) -c -DDOUBLE -DCOMPLEX -DCR $< -o $(@F)
+	$(CC) $(CFLAGS) $(BLOCKS) -c -DDOUBLE -DCOMPLEX -DCR=CR $< -o $(@F)

 zgemm3m_cc.$(SUFFIX) : gemm3m.c gemm3m_level3.c
 	$(CC) $(CFLAGS) $(BLOCKS) -c -DDOUBLE -DCOMPLEX -DCC $< -o $(@F)
@@ -1917,7 +1917,7 @@ xgemm3m_ct.$(SUFFIX) : gemm3m.c gemm3m_level3.c ../../param.h
 	$(CC) $(CFLAGS) $(BLOCKS) -c -DXDOUBLE -DCOMPLEX -DCT $< -o $(@F)

 xgemm3m_cr.$(SUFFIX) : gemm3m.c gemm3m_level3.c ../../param.h
-	$(CC) $(CFLAGS) $(BLOCKS) -c -DXDOUBLE -DCOMPLEX -DCR $< -o $(@F)
+	$(CC) $(CFLAGS) $(BLOCKS) -c -DXDOUBLE -DCOMPLEX -DCR=CR $< -o $(@F)

 xgemm3m_cc.$(SUFFIX) : gemm3m.c gemm3m_level3.c ../../param.h
 	$(CC) $(CFLAGS) $(BLOCKS) -c -DXDOUBLE -DCOMPLEX -DCC $< -o $(@F)
@@ -1974,7 +1974,7 @@ cgemm3m_thread_ct.$(SUFFIX) : gemm3m.c level3_gemm3m_thread.c ../../param.h
 	$(CC) $(CFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -UDOUBLE -DCOMPLEX -DCT $< -o $(@F)

 cgemm3m_thread_cr.$(SUFFIX) : gemm3m.c level3_gemm3m_thread.c ../../param.h
-	$(CC) $(CFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -UDOUBLE -DCOMPLEX -DCR $< -o $(@F)
+	$(CC) $(CFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -UDOUBLE -DCOMPLEX -DCR=CR $< -o $(@F)

 cgemm3m_thread_cc.$(SUFFIX) : gemm3m.c level3_gemm3m_thread.c ../../param.h
 	$(CC) $(CFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -UDOUBLE -DCOMPLEX -DCC $< -o $(@F)
@@ -2022,7 +2022,7 @@ zgemm3m_thread_ct.$(SUFFIX) : gemm3m.c level3_gemm3m_thread.c ../../param.h
 	$(CC) $(CFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DDOUBLE -DCOMPLEX -DCT $< -o $(@F)

 zgemm3m_thread_cr.$(SUFFIX) : gemm3m.c level3_gemm3m_thread.c ../../param.h
-	$(CC) $(CFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DDOUBLE -DCOMPLEX -DCR $< -o $(@F)
+	$(CC) $(CFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DDOUBLE -DCOMPLEX -DCR=CR $< -o $(@F)

 zgemm3m_thread_cc.$(SUFFIX) : gemm3m.c level3_gemm3m_thread.c ../../param.h
 	$(CC) $(CFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DDOUBLE -DCOMPLEX -DCC $< -o $(@F)
@@ -2070,7 +2070,7 @@ xgemm3m_thread_ct.$(SUFFIX) : gemm3m.c level3_gemm3m_thread.c ../../param.h
 	$(CC) $(CFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DXDOUBLE -DCOMPLEX -DCT $< -o $(@F)

 xgemm3m_thread_cr.$(SUFFIX) : gemm3m.c level3_gemm3m_thread.c ../../param.h
-	$(CC) $(CFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DXDOUBLE -DCOMPLEX -DCR $< -o $(@F)
+	$(CC) $(CFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DXDOUBLE -DCOMPLEX -DCR=CR $< -o $(@F)

 xgemm3m_thread_cc.$(SUFFIX) : gemm3m.c level3_gemm3m_thread.c ../../param.h
 	$(CC) $(CFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DXDOUBLE -DCOMPLEX -DCC $< -o $(@F)
@@ -2731,7 +2731,7 @@ cgemm_ct.$(PSUFFIX) : gemm.c level3.c  ../../param.h
 	$(CC) $(PFLAGS) $(BLOCKS) -c -UDOUBLE -DCOMPLEX -DCT $< -o $(@F)

 cgemm_cr.$(PSUFFIX) : gemm.c level3.c  ../../param.h
-	$(CC) $(PFLAGS) $(BLOCKS) -c -UDOUBLE -DCOMPLEX -DCR $< -o $(@F)
+	$(CC) $(PFLAGS) $(BLOCKS) -c -UDOUBLE -DCOMPLEX -DCR=CR $< -o $(@F)

 cgemm_cc.$(PSUFFIX) : gemm.c level3.c  ../../param.h
 	$(CC) $(PFLAGS) $(BLOCKS) -c -UDOUBLE -DCOMPLEX -DCC $< -o $(@F)
@@ -2779,7 +2779,7 @@ zgemm_ct.$(PSUFFIX) : gemm.c level3.c ../../param.h
 	$(CC) $(PFLAGS) $(BLOCKS) -c -DDOUBLE -DCOMPLEX -DCT $< -o $(@F)

 zgemm_cr.$(PSUFFIX) : gemm.c level3.c ../../param.h
-	$(CC) $(PFLAGS) $(BLOCKS) -c -DDOUBLE -DCOMPLEX -DCR $< -o $(@F)
+	$(CC) $(PFLAGS) $(BLOCKS) -c -DDOUBLE -DCOMPLEX -DCR=CR $< -o $(@F)

 zgemm_cc.$(PSUFFIX) : gemm.c level3.c ../../param.h
 	$(CC) $(PFLAGS) $(BLOCKS) -c -DDOUBLE -DCOMPLEX -DCC $< -o $(@F)
@@ -2827,7 +2827,7 @@ xgemm_ct.$(PSUFFIX) : gemm.c level3.c ../../param.h
 	$(CC) $(PFLAGS) $(BLOCKS) -c -DXDOUBLE -DCOMPLEX -DCT $< -o $(@F)

 xgemm_cr.$(PSUFFIX) : gemm.c level3.c ../../param.h
-	$(CC) $(PFLAGS) $(BLOCKS) -c -DXDOUBLE -DCOMPLEX -DCR $< -o $(@F)
+	$(CC) $(PFLAGS) $(BLOCKS) -c -DXDOUBLE -DCOMPLEX -DCR=CR $< -o $(@F)

 xgemm_cc.$(PSUFFIX) : gemm.c level3.c ../../param.h
 	$(CC) $(PFLAGS) $(BLOCKS) -c -DXDOUBLE -DCOMPLEX -DCC $< -o $(@F)
@@ -2927,7 +2927,7 @@ cgemm_thread_ct.$(PSUFFIX) : gemm.c level3_thread.c ../../param.h
 	$(CC) $(PFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -UDOUBLE -DCOMPLEX -DCT $< -o $(@F)

 cgemm_thread_cr.$(PSUFFIX) : gemm.c level3_thread.c ../../param.h
-	$(CC) $(PFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -UDOUBLE -DCOMPLEX -DCR $< -o $(@F)
+	$(CC) $(PFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -UDOUBLE -DCOMPLEX -DCR=CR $< -o $(@F)

 cgemm_thread_cc.$(PSUFFIX) : gemm.c level3_thread.c ../../param.h
 	$(CC) $(PFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -UDOUBLE -DCOMPLEX -DCC $< -o $(@F)
@@ -2975,7 +2975,7 @@ zgemm_thread_ct.$(PSUFFIX) : gemm.c level3_thread.c ../../param.h
 	$(CC) $(PFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DDOUBLE -DCOMPLEX -DCT $< -o $(@F)

 zgemm_thread_cr.$(PSUFFIX) : gemm.c level3_thread.c ../../param.h
-	$(CC) $(PFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DDOUBLE -DCOMPLEX -DCR $< -o $(@F)
+	$(CC) $(PFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DDOUBLE -DCOMPLEX -DCR=CR $< -o $(@F)

 zgemm_thread_cc.$(PSUFFIX) : gemm.c level3_thread.c ../../param.h
 	$(CC) $(PFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DDOUBLE -DCOMPLEX -DCC $< -o $(@F)
@@ -3023,7 +3023,7 @@ xgemm_thread_ct.$(PSUFFIX) : gemm.c level3_thread.c ../../param.h
 	$(CC) $(PFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DXDOUBLE -DCOMPLEX -DCT $< -o $(@F)

 xgemm_thread_cr.$(PSUFFIX) : gemm.c level3_thread.c ../../param.h
-	$(CC) $(PFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DXDOUBLE -DCOMPLEX -DCR $< -o $(@F)
+	$(CC) $(PFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DXDOUBLE -DCOMPLEX -DCR=CR $< -o $(@F)

 xgemm_thread_cc.$(PSUFFIX) : gemm.c level3_thread.c ../../param.h
 	$(CC) $(PFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DXDOUBLE -DCOMPLEX -DCC $< -o $(@F)
@@ -4190,7 +4190,7 @@ cgemm3m_ct.$(PSUFFIX) : gemm3m.c gemm3m_level3.c
 	$(CC) $(PFLAGS) $(BLOCKS) -c -UDOUBLE -DCOMPLEX -DCT $< -o $(@F)

 cgemm3m_cr.$(PSUFFIX) : gemm3m.c gemm3m_level3.c
-	$(CC) $(PFLAGS) $(BLOCKS) -c -UDOUBLE -DCOMPLEX -DCR $< -o $(@F)
+	$(CC) $(PFLAGS) $(BLOCKS) -c -UDOUBLE -DCOMPLEX -DCR=CR $< -o $(@F)

 cgemm3m_cc.$(PSUFFIX) : gemm3m.c gemm3m_level3.c
 	$(CC) $(PFLAGS) $(BLOCKS) -c -UDOUBLE -DCOMPLEX -DCC $< -o $(@F)
@@ -4238,7 +4238,7 @@ zgemm3m_ct.$(PSUFFIX) : gemm3m.c gemm3m_level3.c
 	$(CC) $(PFLAGS) $(BLOCKS) -c -DDOUBLE -DCOMPLEX -DCT $< -o $(@F)

 zgemm3m_cr.$(PSUFFIX) : gemm3m.c gemm3m_level3.c
-	$(CC) $(PFLAGS) $(BLOCKS) -c -DDOUBLE -DCOMPLEX -DCR $< -o $(@F)
+	$(CC) $(PFLAGS) $(BLOCKS) -c -DDOUBLE -DCOMPLEX -DCR=CR $< -o $(@F)

 zgemm3m_cc.$(PSUFFIX) : gemm3m.c gemm3m_level3.c
 	$(CC) $(PFLAGS) $(BLOCKS) -c -DDOUBLE -DCOMPLEX -DCC $< -o $(@F)
@@ -4286,7 +4286,7 @@ xgemm3m_ct.$(PSUFFIX) : gemm3m.c gemm3m_level3.c ../../param.h
 	$(CC) $(PFLAGS) $(BLOCKS) -c -DXDOUBLE -DCOMPLEX -DCT $< -o $(@F)

 xgemm3m_cr.$(PSUFFIX) : gemm3m.c gemm3m_level3.c ../../param.h
-	$(CC) $(PFLAGS) $(BLOCKS) -c -DXDOUBLE -DCOMPLEX -DCR $< -o $(@F)
+	$(CC) $(PFLAGS) $(BLOCKS) -c -DXDOUBLE -DCOMPLEX -DCR=CR $< -o $(@F)

 xgemm3m_cc.$(PSUFFIX) : gemm3m.c gemm3m_level3.c ../../param.h
 	$(CC) $(PFLAGS) $(BLOCKS) -c -DXDOUBLE -DCOMPLEX -DCC $< -o $(@F)
@@ -4343,7 +4343,7 @@ cgemm3m_thread_ct.$(PSUFFIX) : gemm3m.c level3_gemm3m_thread.c ../../param.h
 	$(CC) $(PFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -UDOUBLE -DCOMPLEX -DCT $< -o $(@F)

 cgemm3m_thread_cr.$(PSUFFIX) : gemm3m.c level3_gemm3m_thread.c ../../param.h
-	$(CC) $(PFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -UDOUBLE -DCOMPLEX -DCR $< -o $(@F)
+	$(CC) $(PFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -UDOUBLE -DCOMPLEX -DCR=CR $< -o $(@F)

 cgemm3m_thread_cc.$(PSUFFIX) : gemm3m.c level3_gemm3m_thread.c ../../param.h
 	$(CC) $(PFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -UDOUBLE -DCOMPLEX -DCC $< -o $(@F)
@@ -4391,7 +4391,7 @@ zgemm3m_thread_ct.$(PSUFFIX) : gemm3m.c level3_gemm3m_thread.c ../../param.h
 	$(CC) $(PFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DDOUBLE -DCOMPLEX -DCT $< -o $(@F)

 zgemm3m_thread_cr.$(PSUFFIX) : gemm3m.c level3_gemm3m_thread.c ../../param.h
-	$(CC) $(PFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DDOUBLE -DCOMPLEX -DCR $< -o $(@F)
+	$(CC) $(PFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DDOUBLE -DCOMPLEX -DCR=CR $< -o $(@F)

 zgemm3m_thread_cc.$(PSUFFIX) : gemm3m.c level3_gemm3m_thread.c ../../param.h
 	$(CC) $(PFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DDOUBLE -DCOMPLEX -DCC $< -o $(@F)
@@ -4439,7 +4439,7 @@ xgemm3m_thread_ct.$(PSUFFIX) : gemm3m.c level3_gemm3m_thread.c ../../param.h
 	$(CC) $(PFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DXDOUBLE -DCOMPLEX -DCT $< -o $(@F)

 xgemm3m_thread_cr.$(PSUFFIX) : gemm3m.c level3_gemm3m_thread.c ../../param.h
-	$(CC) $(PFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DXDOUBLE -DCOMPLEX -DCR $< -o $(@F)
+	$(CC) $(PFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DXDOUBLE -DCOMPLEX -DCR=CR $< -o $(@F)

 xgemm3m_thread_cc.$(PSUFFIX) : gemm3m.c level3_gemm3m_thread.c ../../param.h
 	$(CC) $(PFLAGS) $(BLOCKS) -c -DTHREADED_LEVEL3 -DXDOUBLE -DCOMPLEX -DCC $< -o $(@F)
@@ -91,7 +91,12 @@
 #endif

 typedef struct {
-  volatile BLASLONG working[MAX_CPU_NUMBER][CACHE_LINE_SIZE * DIVIDE_RATE];
+#if __STDC_VERSION__ >= 201112L
+  _Atomic
+#else
+  volatile
+#endif  
+   BLASLONG working[MAX_CPU_NUMBER][CACHE_LINE_SIZE * DIVIDE_RATE];
 } job_t;


@@ -67,7 +67,12 @@
 #endif

 typedef struct {
-  volatile BLASLONG working[MAX_CPU_NUMBER][CACHE_LINE_SIZE * DIVIDE_RATE];
+#if __STDC_VERSION__ >= 201112L
+_Atomic
+#else 
+  volatile
+#endif
+   BLASLONG working[MAX_CPU_NUMBER][CACHE_LINE_SIZE * DIVIDE_RATE];
 } job_t;


@@ -91,7 +91,12 @@
 #endif

 typedef struct {
-  volatile BLASLONG working[MAX_CPU_NUMBER][CACHE_LINE_SIZE * DIVIDE_RATE];
+#if __STDC_VERSION__ >= 201112L
+_Atomic
+#else  
+  volatile
+#endif
+   BLASLONG working[MAX_CPU_NUMBER][CACHE_LINE_SIZE * DIVIDE_RATE];
 } job_t;


@@ -36,6 +36,7 @@
 /* or implied, of The University of Texas at Austin.                 */
 /*********************************************************************/

+#include <stdbool.h>
 #include <stdio.h>
 #include <stdlib.h>
 //#include <sys/mman.h>
@@ -49,11 +50,16 @@

 int blas_server_avail = 0;

-static void * blas_thread_buffer[MAX_CPU_NUMBER];
+static void * blas_thread_buffer[MAX_PARALLEL_NUMBER][MAX_CPU_NUMBER];
+#if __STDC_VERSION__ >= 201112L
+static atomic_bool blas_buffer_inuse[MAX_PARALLEL_NUMBER];
+#else
+static _Bool blas_buffer_inuse[MAX_PARALLEL_NUMBER];
+#endif

 void goto_set_num_threads(int num_threads) {

-  int i=0;
+  int i=0, j=0;

  if (num_threads < 1) num_threads = blas_num_threads;

@@ -68,15 +74,17 @@ void goto_set_num_threads(int num_threads) {
  omp_set_num_threads(blas_cpu_number);

  //adjust buffer for each thread
-  for(i=0; i<blas_cpu_number; i++){
-    if(blas_thread_buffer[i]==NULL){
-      blas_thread_buffer[i]=blas_memory_alloc(2);
+  for(i=0; i<MAX_PARALLEL_NUMBER; i++) {
+    for(j=0; j<blas_cpu_number; j++){
+      if(blas_thread_buffer[i][j]==NULL){
+        blas_thread_buffer[i][j]=blas_memory_alloc(2);
+      }
    }
-  }
-  for(; i<MAX_CPU_NUMBER; i++){
-    if(blas_thread_buffer[i]!=NULL){
-      blas_memory_free(blas_thread_buffer[i]);
-      blas_thread_buffer[i]=NULL;
+    for(; j<MAX_CPU_NUMBER; j++){
+      if(blas_thread_buffer[i][j]!=NULL){
+        blas_memory_free(blas_thread_buffer[i][j]);
+        blas_thread_buffer[i][j]=NULL;
+      }
    }
  }
 #if defined(ARCH_MIPS64)
@@ -92,30 +100,34 @@ void openblas_set_num_threads(int num_threads) {

 int blas_thread_init(void){

-  int i=0;
+  int i=0, j=0;

  blas_get_cpu_number();

  blas_server_avail = 1;

-  for(i=0; i<blas_num_threads; i++){
-    blas_thread_buffer[i]=blas_memory_alloc(2);
-  }
-  for(; i<MAX_CPU_NUMBER; i++){
-      blas_thread_buffer[i]=NULL;
+  for(i=0; i<MAX_PARALLEL_NUMBER; i++) {
+    for(j=0; j<blas_num_threads; j++){
+      blas_thread_buffer[i][j]=blas_memory_alloc(2);
+    }
+    for(; j<MAX_CPU_NUMBER; j++){
+      blas_thread_buffer[i][j]=NULL;
+    }
  }

  return 0;
 }

 int BLASFUNC(blas_thread_shutdown)(void){
-  int i=0;
+  int i=0, j=0;
  blas_server_avail = 0;

-  for(i=0; i<MAX_CPU_NUMBER; i++){
-    if(blas_thread_buffer[i]!=NULL){
-      blas_memory_free(blas_thread_buffer[i]);
-      blas_thread_buffer[i]=NULL;
+  for(i=0; i<MAX_PARALLEL_NUMBER; i++) {
+    for(j=0; j<MAX_CPU_NUMBER; j++){
+      if(blas_thread_buffer[i][j]!=NULL){
+        blas_memory_free(blas_thread_buffer[i][j]);
+        blas_thread_buffer[i][j]=NULL;
+      }
    }
  }

@@ -206,7 +218,7 @@ static void legacy_exec(void *func, int mode, blas_arg_t *args, void *sb){
      }
 }

-static void exec_threads(blas_queue_t *queue){
+static void exec_threads(blas_queue_t *queue, int buf_index){

  void *buffer, *sa, *sb;
  int pos=0, release_flag=0;
@@ -223,7 +235,7 @@ static void exec_threads(blas_queue_t *queue){
  if ((sa == NULL) && (sb == NULL) && ((queue -> mode & BLAS_PTHREAD) == 0)) {

    pos = omp_get_thread_num();
-    buffer = blas_thread_buffer[pos];
+    buffer = blas_thread_buffer[buf_index][pos];

    //fallback
    if(buffer==NULL) {
@@ -291,7 +303,7 @@ static void exec_threads(blas_queue_t *queue){

 int exec_blas(BLASLONG num, blas_queue_t *queue){

-  BLASLONG i;
+  BLASLONG i, buf_index;

  if ((num <= 0) || (queue == NULL)) return 0;

@@ -302,6 +314,23 @@ int exec_blas(BLASLONG num, blas_queue_t *queue){
  }
 #endif

+  while(true) {
+    for(i=0; i < MAX_PARALLEL_NUMBER; i++) {
+#if __STDC_VERSION__ >= 201112L
+      _Bool inuse = false;
+      if(atomic_compare_exchange_weak(&blas_buffer_inuse[i], &inuse, true)) {
+#else
+      if(blas_buffer_inuse[i] == false) {
+        blas_buffer_inuse[i] = true;
+#endif
+        buf_index = i;
+        break;
+      }
+    }
+    if(i != MAX_PARALLEL_NUMBER)
+      break;
+  }
+
 #pragma omp parallel for schedule(static)
  for (i = 0; i < num; i ++) {

@@ -309,9 +338,15 @@ int exec_blas(BLASLONG num, blas_queue_t *queue){
    queue[i].position = i;
 #endif

-    exec_threads(&queue[i]);
+    exec_threads(&queue[i], buf_index);
  }

+#if __STDC_VERSION__ >= 201112L
+  atomic_store(&blas_buffer_inuse[buf_index], false);
+#else
+  blas_buffer_inuse[buf_index] = false;
+#endif
+
  return 0;
 }

@@ -74,15 +74,22 @@ extern gotoblas_t  gotoblas_STEAMROLLER;
 extern gotoblas_t  gotoblas_EXCAVATOR;
 #ifdef NO_AVX2
 #define gotoblas_HASWELL gotoblas_SANDYBRIDGE
+#define gotoblas_SKYLAKEX gotoblas_SANDYBRIDGE
 #define gotoblas_ZEN gotoblas_SANDYBRIDGE
 #else
 extern gotoblas_t  gotoblas_HASWELL;
 extern gotoblas_t  gotoblas_ZEN;
+#ifndef NO_AVX512
+extern gotoblas_t  gotoblas_SKYLAKEX;
+#else
+#define gotoblas_SKYLAKEX gotoblas_HASWELL
+#endif
 #endif
 #else
 //Use NEHALEM kernels for sandy bridge
 #define gotoblas_SANDYBRIDGE gotoblas_NEHALEM
 #define gotoblas_HASWELL gotoblas_NEHALEM
+#define gotoblas_SKYLAKEX gotoblas_NEHALEM
 #define gotoblas_BULLDOZER gotoblas_BARCELONA
 #define gotoblas_PILEDRIVER gotoblas_BARCELONA
 #define gotoblas_STEAMROLLER gotoblas_BARCELONA
@@ -284,8 +291,21 @@ static gotoblas_t *get_coretype(void){
 	    return &gotoblas_NEHALEM; //OS doesn't support AVX. Use old kernels.
 	  }
 	}
+	if (model == 5) {	
+	// Intel Skylake X
+#ifndef NO_AVX512
+	  return &gotoblas_SKYLAKEX;
+#else		
+	  if(support_avx())
+	    return &gotoblas_HASWELL;
+	  else {
+	    openblas_warning(FALLBACK_VERBOSE, NEHALEM_FALLBACK);
+	    return &gotoblas_NEHALEM;
+	  }
+#endif		
+	}
 	//Intel Skylake
-	if (model == 14 || model == 5) {
+	if (model == 14) {
 	  if(support_avx())
 	    return &gotoblas_HASWELL;
 	  else{
@@ -445,7 +465,8 @@ static char *corename[] = {
    "Haswell",
    "Steamroller",
    "Excavator",
-    "Zen"
+    "Zen",
+    "SkylakeX"	
 };

 char *gotoblas_corename(void) {
@@ -473,7 +494,7 @@ char *gotoblas_corename(void) {
  if (gotoblas == &gotoblas_STEAMROLLER)  return corename[21];
  if (gotoblas == &gotoblas_EXCAVATOR)    return corename[22];
  if (gotoblas == &gotoblas_ZEN)          return corename[23];
-
+  if (gotoblas == &gotoblas_SKYLAKEX)     return corename[24];
  return corename[0];
 }

@@ -485,7 +506,7 @@ static gotoblas_t *force_coretype(char *coretype){
 	char message[128];
 	//char mname[20];

-	for ( i=1 ; i <= 23; i++)
+	for ( i=1 ; i <= 24; i++)
 	{
 		if (!strncasecmp(coretype,corename[i],20))
 		{
@@ -503,6 +524,7 @@ static gotoblas_t *force_coretype(char *coretype){

 	switch (found)
 	{
+		case 24: return (&gotoblas_SKYLAKEX);	
 		case 23: return (&gotoblas_ZEN);
 		case 22: return (&gotoblas_EXCAVATOR);
 		case 21: return (&gotoblas_STEAMROLLER);
@@ -180,7 +180,7 @@ int get_num_procs(void) {
 cpu_set_t *cpusetp;
 size_t size;
 int ret;
-// int i,n;
+int i,n;

  if (!nums) nums = sysconf(_SC_NPROCESSORS_CONF);
 #if !defined(OS_LINUX)
@@ -212,7 +212,8 @@ int ret;
  size = CPU_ALLOC_SIZE(nums);
  ret = sched_getaffinity(0,size,cpusetp);
  if (ret!=0) return nums;
-  nums = CPU_COUNT_S(size,cpusetp);
+  ret = CPU_COUNT_S(size,cpusetp);
+  if (ret > 0 && ret < nums) nums = ret;	
  CPU_FREE(cpusetp);
  return nums;
 #endif
@@ -54,6 +54,9 @@ static char* openblas_config_str=""
 #ifdef NO_AFFINITY
  "NO_AFFINITY "
 #endif
+#ifdef USE_OPENMP
+  "USE_OPENMP "
+#endif
 #ifndef DYNAMIC_ARCH
  CHAR_CORENAME
 #endif
@@ -61,18 +64,23 @@ static char* openblas_config_str=""

 #ifdef DYNAMIC_ARCH
 char *gotoblas_corename();
-static char tmp_config_str[256];
 #endif

+static char tmp_config_str[256];
+int openblas_get_parallel();

 char* CNAME() {
-#ifndef DYNAMIC_ARCH
-  return openblas_config_str;
-#else
+char tmpstr[20];
  strcpy(tmp_config_str, openblas_config_str);
+#ifdef DYNAMIC_ARCH
  strcat(tmp_config_str, gotoblas_corename());
-  return tmp_config_str;
 #endif
+if (openblas_get_parallel() == 0)
+  sprintf(tmpstr, " SINGLE_THREADED");
+else 
+  snprintf(tmpstr,19," MAX_THREADS=%d",MAX_CPU_NUMBER);
+  strcat(tmp_config_str, tmpstr);
+  return tmp_config_str;
 }


@@ -83,3 +91,4 @@ char* openblas_get_corename() {
  return gotoblas_corename();
 #endif
 }
+
@@ -167,7 +167,7 @@ int get_L2_size(void){
 #if defined(ATHLON) || defined(OPTERON) || defined(BARCELONA) || defined(BOBCAT) || defined(BULLDOZER) || \
    defined(CORE_PRESCOTT) || defined(CORE_CORE2) || defined(PENRYN) || defined(DUNNINGTON) || \
    defined(CORE_NEHALEM) || defined(CORE_SANDYBRIDGE) || defined(ATOM) || defined(GENERIC) || \
-    defined(PILEDRIVER) || defined(HASWELL) || defined(STEAMROLLER) || defined(EXCAVATOR) || defined(ZEN)
+    defined(PILEDRIVER) || defined(HASWELL) || defined(STEAMROLLER) || defined(EXCAVATOR) || defined(ZEN) || defined(SKYLAKEX)

  cpuid(0x80000006, &eax, &ebx, &ecx, &edx);

@@ -251,7 +251,7 @@ int get_L2_size(void){
 void blas_set_parameter(void){

  int factor;
-#if defined(BULLDOZER) || defined(PILEDRIVER) || defined(SANDYBRIDGE) || defined(NEHALEM) || defined(HASWELL) || defined(STEAMROLLER) || defined(EXCAVATOR) || defined(ZEN)
+#if defined(BULLDOZER) || defined(PILEDRIVER) || defined(SANDYBRIDGE) || defined(NEHALEM) || defined(HASWELL) || defined(STEAMROLLER) || defined(EXCAVATOR) || defined(ZEN) || defined(SKYLAKEX)
  int size = 16;
 #else
  int size = get_L2_size();
@@ -128,6 +128,8 @@ so : ../$(LIBSONAME)

 ifeq ($(OSNAME), Android)
 INTERNALNAME = $(LIBPREFIX).so
+FEXTRALIB += -lm
+EXTRALIB += -lm
 else
 INTERNALNAME = $(LIBPREFIX).so.$(MAJOR_VERSION)
 endif
@@ -97,7 +97,7 @@ if ($compiler eq "") {

 	if ($data =~ /Intel/) {
 	    $vendor = INTEL;
-	    $openmp = "-openmp";
+	    $openmp = "-fopenmp";
 	}

        if ($data =~ /Sun Fortran/) {
@@ -127,7 +127,7 @@ if ($compiler eq "") {

 	# for embeded underscore name, e.g. zho_ge, it may append 2 underscores.
 	$data = `$compiler -O2 -S ftest3.f > /dev/null 2>&1 && cat ftest3.s && rm -f ftest3.s`;
-	if ($data =~ /zho_ge__/) {
+	if ($data =~ / zho_ge__/) {
 	    $need2bu       = 1;
 	}
    }
@@ -155,7 +155,7 @@ if ($compiler eq "") {
 	if ($compiler =~ /ifort/) {
 	    $vendor = INTEL;
 	    $bu       = "_";
-	    $openmp = "-openmp";
+	    $openmp = "-fopenmp";
 	}

 	if ($compiler =~ /pathf/) {
@@ -326,6 +326,21 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #define CORENAME  "HASWELL"
 #endif

+#ifdef FORCE_SKYLAKEX
+#define FORCE
+#define FORCE_INTEL
+#define ARCHITECTURE    "X86"
+#define SUBARCHITECTURE "SKYLAKEX"
+#define ARCHCONFIG   "-DSKYLAKEX " \
+		     "-DL1_DATA_SIZE=32768 -DL1_DATA_LINESIZE=64 " \
+		     "-DL2_SIZE=262144 -DL2_LINESIZE=64 " \
+		     "-DDTB_DEFAULT_ENTRIES=64 -DDTB_SIZE=4096 " \
+		     "-DHAVE_CMOV -DHAVE_MMX -DHAVE_SSE -DHAVE_SSE2 -DHAVE_SSE3 -DHAVE_SSSE3 -DHAVE_SSE4_1 -DHAVE_SSE4_2 -DHAVE_AVX " \
+                     "-DFMA3 -DHAVE_AVX512VL -march=skylake-avx512"
+#define LIBNAME   "skylakex"
+#define CORENAME  "SKYLAKEX"
+#endif
+
 #ifdef FORCE_ATOM
 #define FORCE
 #define FORCE_INTEL
@@ -1181,9 +1196,7 @@ int main(int argc, char *argv[]){
 #elif NO_PARALLEL_MAKE==1
    printf("MAKE += -j 1\n");
 #else
-#ifndef OS_WINDOWS
    printf("MAKE += -j %d\n", get_num_cores());
-#endif
 #endif

    break;
@@ -44,6 +44,7 @@
 #endif

 #ifndef COMPLEX
+#define SMP_THRESHOLD_MIN 65536.0
 #ifdef XDOUBLE
 #define ERROR_NAME "QGEMM "
 #elif defined(DOUBLE)
@@ -52,6 +53,7 @@
 #define ERROR_NAME "SGEMM "
 #endif
 #else
+#define SMP_THRESHOLD_MIN 8192.0
 #ifndef GEMM3M
 #ifdef XDOUBLE
 #define ERROR_NAME "XGEMM "
@@ -121,8 +123,6 @@ void NAME(char *TRANSA, char *TRANSB,
  FLOAT *sa, *sb;

 #ifdef SMP
-  int nthreads_max;
-  int nthreads_avail;
  double MNK;
 #ifndef COMPLEX
 #ifdef XDOUBLE
@@ -245,8 +245,6 @@ void CNAME(enum CBLAS_ORDER order, enum CBLAS_TRANSPOSE TransA, enum CBLAS_TRANS
  XFLOAT *sa, *sb;

 #ifdef SMP
-  int nthreads_max;
-  int nthreads_avail;
  double MNK;
 #ifndef COMPLEX
 #ifdef XDOUBLE
@@ -411,25 +409,12 @@ void CNAME(enum CBLAS_ORDER order, enum CBLAS_TRANSPOSE TransA, enum CBLAS_TRANS
  mode |= (transa << BLAS_TRANSA_SHIFT);
  mode |= (transb << BLAS_TRANSB_SHIFT);

-  nthreads_max = num_cpu_avail(3);
-  nthreads_avail = nthreads_max;
-
-#ifndef COMPLEX
  MNK = (double) args.m * (double) args.n * (double) args.k;
-  if ( MNK <= (65536.0  * (double) GEMM_MULTITHREAD_THRESHOLD)  )
-	nthreads_max = 1;
-#else
-  MNK = (double) args.m * (double) args.n * (double) args.k;
-  if ( MNK <= (8192.0  * (double) GEMM_MULTITHREAD_THRESHOLD)  )
-	nthreads_max = 1;
-#endif
-  args.common = NULL;
-
-  if ( nthreads_max > nthreads_avail )
-  	args.nthreads = nthreads_avail;
+  if ( MNK <= (SMP_THRESHOLD_MIN  * (double) GEMM_MULTITHREAD_THRESHOLD)  )
+	args.nthreads = 1;
  else
-  	args.nthreads = nthreads_max;
-
+	args.nthreads = num_cpu_avail(3);
+  args.common = NULL;

 if (args.nthreads == 1) {
 #endif
@@ -366,12 +366,13 @@ void CNAME(enum CBLAS_ORDER order,
  mode |= (trans << BLAS_TRANSA_SHIFT);
  mode |= (side  << BLAS_RSIDE_SHIFT);

-  args.nthreads = num_cpu_avail(3);
  if ( args.m < 2*GEMM_MULTITHREAD_THRESHOLD )
 	args.nthreads = 1;
  else
 	if ( args.n < 2*GEMM_MULTITHREAD_THRESHOLD )
 		args.nthreads = 1;
+  else
+	args.nthreads = num_cpu_avail(3);
 		

  if (args.nthreads == 1) {
@@ -41,7 +41,11 @@
 #ifdef FUNCTION_PROFILE
 #include "functable.h"
 #endif
-
+#if  defined(Z13)
+#define MULTI_THREAD_MINIMAL  200000
+#else
+#define MULTI_THREAD_MINIMAL  10000
+#endif
 #ifndef CBLAS

 void NAME(blasint *N, FLOAT *ALPHA, FLOAT *x, blasint *INCX, FLOAT *y, blasint *INCY){
@@ -69,7 +73,7 @@ void CNAME(blasint n, FLOAT *ALPHA, FLOAT *x, blasint incx, FLOAT *y, blasint in
 #endif

 #ifndef CBLAS
-  PRINT_DEBUG_CNAME;
+  PRINT_DEBUG_NAME;
 #else
  PRINT_DEBUG_CNAME;
 #endif
@@ -93,6 +97,11 @@ void CNAME(blasint n, FLOAT *ALPHA, FLOAT *x, blasint incx, FLOAT *y, blasint in
  if (incx == 0 || incy == 0)
 	  nthreads = 1;

+  //Work around the low performance issue with small imput size &
+  //multithreads.
+  if (n <= MULTI_THREAD_MINIMAL) {
+	  nthreads = 1;
+  }
  if (nthreads == 1) {
 #endif

@@ -239,6 +239,9 @@ void CNAME(enum CBLAS_ORDER order, enum CBLAS_UPLO Uplo,
  } else
      nthreads = 1;

+/* FIXME TRMV multithreading appears to be broken, see issue 1332*/
+  nthreads = 1;
+
  if(nthreads > 1) {
    buffer_size = n > 16 ? 0 : n * 4 + 40;
  }
@@ -121,7 +121,7 @@ function (build_core TARGET_CORE KDIR TSUFFIX KERNEL_DEFINITIONS)
    # Makefile.L3
    set(USE_TRMM false)

-    if (ARM OR ARM64 OR "${TARGET_CORE}" STREQUAL "LONGSOON3B" OR "${TARGET_CORE}" STREQUAL "GENERIC" OR "${CORE}" STREQUAL "generic" OR "${TARGET_CORE}" STREQUAL "HASWELL" OR "${CORE}" STREQUAL "haswell" OR "${CORE}" STREQUAL "zen")
+    if (ARM OR ARM64 OR "${TARGET_CORE}" STREQUAL "LONGSOON3B" OR "${TARGET_CORE}" STREQUAL "GENERIC" OR "${CORE}" STREQUAL "generic" OR "${TARGET_CORE}" STREQUAL "HASWELL" OR "${CORE}" STREQUAL "haswell" OR "${CORE}" STREQUAL "zen" OR "${TARGET_CORE}" STREQUAL "SKYLAKEX" OR "${CORE}" STREQUAL "skylakex")
      set(USE_TRMM true)
    endif ()

@@ -29,9 +29,11 @@ USE_TRMM = 1
 endif

 ifeq ($(CORE), HASWELL)
-ifeq ($(ARCH), x86_64)
 USE_TRMM = 1
 endif
+
+ifeq ($(CORE), SKYLAKEX)
+USE_TRMM = 1
 endif

 ifeq ($(CORE), ZEN)
@@ -440,13 +440,13 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

 	cmp	N, #0
 	ble	axpy_kernel_L999
-
+/*
 	cmp	INC_X, #0
 	beq	axpy_kernel_L999

 	cmp	INC_Y, #0
 	beq	axpy_kernel_L999
-
+*/
 	cmp	INC_X, #1
 	bne	axpy_kernel_S_BEGIN

@@ -483,13 +483,13 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

 	cmp	N, #0
 	ble	rot_kernel_L999
-
+/*
 	cmp	INC_X, #0
 	beq	rot_kernel_L999

 	cmp	INC_Y, #0
 	beq	rot_kernel_L999
-
+*/
 	cmp	INC_X, #1
 	bne	rot_kernel_S_BEGIN

@@ -584,6 +584,12 @@ rot_kernel_S1:
 rot_kernel_S10:

 	KERNEL_S1
+	
+	cmp	INC_X, #0
+	beq	rot_kernel_L999
+
+	cmp	INC_Y, #0
+	beq	rot_kernel_L999

 	subs    I, I, #1
        bne     rot_kernel_S10
@@ -116,22 +116,34 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG posX, BLASLON
      if (m & 1) {

 	if (X > posY) {
-	  /* ao1 += 1;
-	  ao2 += 1; */
+	  ao1 += 1;
+	  ao2 += 1;
 	  b += 2;
 	} else
-#ifdef UNIT
 	  if (X < posY) {
-#endif 
-	     b[ 0] = *(ao1 + 0);
-#ifdef UNIT
+	    data01 = *(ao1 + 0);
+	    data02 = *(ao1 + 1);
+
+	    b[ 0] = data01;
+	    b[ 1] = data02;
+	    ao1 += lda;
+	    b += 2;
 	  } else {
+#ifdef UNIT
+	    data02 = *(ao1 + 1);

 	    b[ 0] = ONE;
+	    b[ 1] = data02;
+#else
+	    data01 = *(ao1 + 0);
+	    data02 = *(ao1 + 1);
+
+	    b[ 0] = data01;
+	    b[ 1] = data02;
+#endif
+	    ao1 += 2;
+	    b += 2;
 	  }
-#endif 
-	  b[ 1] = *(ao1 + 1);
-	  b += 2;
      }

      posY += 2;
@@ -178,7 +190,7 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG posX, BLASLON
 	} while (i > 0);
      }

-      // posY += 1;
+      posY += 1;
  }

  return 0;
@@ -518,7 +518,7 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG posX, BLASLON
      i = (m & 15);
      if (i > 0) {
 	if (X < posY) {
-	  /* a01 += i;
+	  a01 += i;
 	  a02 += i;
 	  a03 += i;
 	  a04 += i;
@@ -533,7 +533,7 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG posX, BLASLON
 	  a13 += i;
 	  a14 += i;
 	  a15 += i;
-	  a16 += i; */
+	  a16 += i;
 	  b += 16 * i;
 	} else
 	  if (X > posY) {
@@ -1130,14 +1130,14 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG posX, BLASLON
      i = (m & 7);
      if (i > 0) {
 	if (X < posY) {
-	  /* a01 += i;
+	  a01 += i;
 	  a02 += i;
 	  a03 += i;
 	  a04 += i;
 	  a05 += i;
 	  a06 += i;
 	  a07 += i;
-	  a08 += i; */
+	  a08 += i;
 	  b += 8 * i;
 	} else
 	  if (X > posY) {
@@ -1156,13 +1156,13 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG posX, BLASLON
 	      b += 8;
 	    }

-	    /* a02 += i * lda;
+	    a02 += i * lda;
 	    a03 += i * lda;
 	    a04 += i * lda;
 	    a05 += i * lda;
 	    a06 += i * lda;
 	    a07 += i * lda;
-	    a08 += i * lda; */
+	    a08 += i * lda;
 	  } else {
 #ifdef UNIT
 	    b[ 0] = ONE;
@@ -1371,10 +1371,10 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG posX, BLASLON
      i = (m & 3);
      if (i > 0) {
 	if (X < posY) {
-	  /* a01 += i;
+	  a01 += i;
 	  a02 += i;
 	  a03 += i;
-	  a04 += i; */
+	  a04 += i;
 	  b += 4 * i;
 	} else
 	  if (X > posY) {
@@ -1387,9 +1387,9 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG posX, BLASLON
 	      a01 += lda;
 	      b += 4;
 	    }
-	    /* a02 += lda;
+	    a02 += lda;
 	    a03 += lda;
-	    a04 += lda; */
+	    a04 += lda;
 	  } else {

 #ifdef UNIT
@@ -1487,19 +1487,23 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG posX, BLASLON
      if (X < posY) {
 	a01 ++;
 	a02 ++;
-      } else {
-#ifdef UNIT
+	b += 2;
+      } else
 	if (X > posY) {
-#endif
 	  b[  0] = *(a01 +  0);
-#ifdef UNIT
+	  b[  1] = *(a01 +  1);
+	  a01 += lda;
+	  b += 2;
 	} else {
+#ifdef UNIT
 	  b[  0] = ONE;
-	}
+	  b[  1] = *(a01 +  1);
+#else
+	  b[  0] = *(a01 +  0);
+	  b[  1] = *(a01 +  1);
 #endif
-	b[  1] = *(a01 +  1);
-      }
-      b += 2;
+	  b += 2;
+	}
    }
    posY += 2;
  }
@@ -1518,25 +1522,28 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG posX, BLASLON
    if (i > 0) {
      do {
 	if (X < posY) {
-	  a01 ++;
-	} else {
-#ifdef UNIT
+	  a01 += 1;
+	  b ++;
+	} else
 	  if (X > posY) {
-#endif
 	    b[  0] = *(a01 +  0);
-#ifdef UNIT
+	    a01 += lda;
+	    b ++;
 	  } else {
+#ifdef UNIT
 	    b[ 0] = ONE;
-	  }
+#else
+	    b[ 0] = *(a01 +  0);
 #endif
-	  a01 += lda;
-	}
-	b ++;
-	X ++;
-	i --;
+	    a01 += lda;
+	    b ++;
+	  }
+
+	  X += 1;
+	  i --;
      } while (i > 0);
    }
-    // posY += 1;
+    posY += 1;
  }

  return 0;
@@ -117,8 +117,8 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG posX, BLASLON
      if (m & 1) {

 	if (X < posY) {
-	  /* ao1 += 1;
-	  ao2 += 1; */
+	  ao1 += 1;
+	  ao2 += 1;
 	  b += 2;
 	} else
 	  if (X > posY) {
@@ -127,7 +127,7 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG posX, BLASLON

 	    b[ 0] = data01;
 	    b[ 1] = data02;
-	    // ao1 += lda;
+	    ao1 += lda;
 	    b += 2;
 	  } else {
 #ifdef UNIT
@@ -139,7 +139,7 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG posX, BLASLON
 	    b[ 0] = data01;
 	    b[ 1] = ZERO;
 #endif
-	    // ao1 += lda;
+	    ao1 += lda;
 	    b += 2;
 	  }
      }
@@ -161,18 +161,27 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG posX, BLASLON
    i = m;
    if (m > 0) {
      do {
+	if (X < posY) {
+	  b += 1;
+  	  ao1 += 1;
+	} else
+	  if (X > posY) {
+	    data01 = *(ao1 + 0);
+	    b[ 0] = data01;
+	    b += 1;
+	    ao1 += lda;
+	  } else {
 #ifdef UNIT
-        if (X > posY) {
+	    b[ 0] = ONE;
+#else
+	    data01 = *(ao1 + 0);
+	    b[ 0] = data01;
 #endif
-	  b[ 0] = *(ao1 + 0);
-#ifdef UNIT
-	} else {
-	  b[ 0] = ONE;
-	}
-#endif
-        b ++;
-        ao1 += lda;
-	X ++;
+	    b += 1;
+	    ao1 += lda;
+	  }
+
+	X += 1;
 	i --;
      } while (i > 0);
    }
@@ -201,18 +201,18 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG posX, BLASLON
 	if (X < posY) {

 	  if (m & 2) {
-	    /* ao1 += 2;
+	    ao1 += 2;
 	    ao2 += 2;
 	    ao3 += 2;
-	    ao4 += 2; */
+	    ao4 += 2;
 	    b += 8;
 	  }

 	  if (m & 1) {
-	    /* ao1 += 1;
+	    ao1 += 1;
 	    ao2 += 1;
 	    ao3 += 1;
-	    ao4 += 1; */
+	    ao4 += 1;
 	    b += 4;
 	  }

@@ -238,7 +238,7 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG posX, BLASLON
 	      b[ 7] = data08;

 	      ao1 += 2 * lda;
-	      // ao2 += 2 * lda;
+	      ao2 += 2 * lda;
 	      b   += 8;
 	    }

@@ -253,7 +253,7 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG posX, BLASLON
 	      b[ 2] = data03;
 	      b[ 3] = data04;

-	      // ao1 += lda;
+	      ao1 += lda;
 	      b += 4;
 	    }

@@ -401,7 +401,7 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG posX, BLASLON
      if (i) {

 	if (X < posY) {
-	  // ao1 += 2;
+	  ao1 += 2;
 	  b += 2;
 	} else
 	  if (X > posY) {
@@ -411,7 +411,7 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG posX, BLASLON
 	    b[ 0] = data01;
 	    b[ 1] = data02;

-	    // ao1 += lda;
+	    ao1 += lda;
 	    b += 2;
 	  } else {
 #ifdef UNIT
@@ -443,21 +443,26 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG posX, BLASLON
 	do {

 	  if (X < posY) {
+	    b += 1;
 	    ao1 += 1;
-	  } else {
-#ifdef UNIT
+	  } else
 	    if (X > posY) {
-#endif
-	      b[ 0] = *(ao1 + 0);
-#ifdef UNIT
+	      data01 = *(ao1 + 0);
+	      b[ 0] = data01;
+	      ao1 += lda;
+	      b += 1;
 	    } else {
+#ifdef UNIT
 	      b[ 0] = ONE;
-	    }
+#else
+	      data01 = *(ao1 + 0);
+	      b[ 0] = data01;
 #endif
-	    ao1 += lda;
-	  }
-	  b ++;
-	  X ++;
+	      ao1 += lda;
+	      b += 1;
+	    }
+
+	  X += 1;
 	  i --;
 	} while (i > 0);
      }
@@ -206,7 +206,7 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG offset, FLOAT
      }

      a1 += 2 * lda;
-      // a2 += 2 * lda;
+      a2 += 2 * lda;
      b += 8;

      ii += 2;
@@ -139,18 +139,48 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG posX, BLASLON
      }

      if (m & 1) {
-#ifdef UNIT
+
+	if (X > posY) {
+	  ao1 += 2;
+	  ao2 += 2;
+	  b += 4;
+
+	} else
 	  if (X < posY) {
-#endif
-	    b[ 0] = *(ao1 + 0);
-	    b[ 1] = *(ao1 + 1);
-#ifdef UNIT
+	    data1 = *(ao1 + 0);
+	    data2 = *(ao1 + 1);
+	    data3 = *(ao1 + 2);
+	    data4 = *(ao1 + 3);
+
+	    b[ 0] = data1;
+	    b[ 1] = data2;
+	    b[ 2] = data3;
+	    b[ 3] = data4;
+
+	    ao1 += lda;
+	    b += 4;
 	  } else {
+#ifdef UNIT
+	    data3 = *(ao1 + 2);
+	    data4 = *(ao1 + 3);
+
 	    b[ 0] = ONE;
 	    b[ 1] = ZERO;
-	  }
+	    b[ 2] = data3;
+	    b[ 3] = data4;
+#else
+	    data1 = *(ao1 + 0);
+	    data2 = *(ao1 + 1);
+	    data3 = *(ao1 + 2);
+	    data4 = *(ao1 + 3);
+
+	    b[ 0] = data1;
+	    b[ 1] = data2;
+	    b[ 2] = data3;
+	    b[ 3] = data4;
 #endif
-	  b += 4;
+	    b += 4;
+	  }
      }

      posY += 2;
@@ -203,7 +233,7 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG posX, BLASLON
 	} while (i > 0);
      }

-      // posY += 1;
+      posY += 1;
  }

  return 0;
@@ -43,7 +43,7 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG offset, FLOAT

  BLASLONG i, ii, j, jj;

-  FLOAT data01 = 0.0, data02 = 0.0;
+  FLOAT data01, data02;
  FLOAT *a1;

  lda *= 2;
@@ -43,8 +43,8 @@ int CNAME(BLASLONG m, BLASLONG n, FLOAT *a, BLASLONG lda, BLASLONG offset, FLOAT

  BLASLONG i, ii, j, jj;

-  FLOAT data01 = 0.0, data02 = 0.0, data03, data04;
-  FLOAT data05, data06, data07 = 0.0, data08 = 0.0;
+  FLOAT data01, data02, data03, data04;
+  FLOAT data05, data06, data07, data08;
  FLOAT *a1, *a2;

  lda *= 2;
@@ -0,0 +1 @@
+include $(KERNELDIR)/KERNEL.P5600
@@ -38,8 +38,8 @@ ZASUMKERNEL  = ../mips/zasum_msa.c
 else
 SASUMKERNEL  = ../mips/asum.c
 DASUMKERNEL  = ../mips/asum.c
-CASUMKERNEL  = ../mips/asum.c
-ZASUMKERNEL  = ../mips/asum.c
+CASUMKERNEL  = ../mips/zasum.c
+ZASUMKERNEL  = ../mips/zasum.c
 endif

 ifdef HAVE_MSA
@@ -253,4 +253,4 @@ ZTRSMKERNEL_LN = ../generic/trsm_kernel_LN.c
 ZTRSMKERNEL_LT = ../generic/trsm_kernel_LT.c
 ZTRSMKERNEL_RN = ../generic/trsm_kernel_RN.c
 ZTRSMKERNEL_RT = ../generic/trsm_kernel_RT.c
-endif
+endif
@@ -484,10 +484,10 @@ int CNAME(BLASLONG m, BLASLONG n, BLASLONG dummy1, FLOAT alpha, FLOAT *A,
    FLOAT *pa0, *pa1, *pa2, *pa3, *pa4, *pa5, *pa6, *pa7;
    FLOAT temp, temp0, temp1, temp2, temp3, temp4, temp5, temp6, temp7;
    v2f64 v_alpha;
-    v2f64 x0, x1, x2, x3, y0 = 0.0, y1 = 0.0, y2 = 0.0, y3 = 0.0;
+    v2f64 x0, x1, x2, x3, y0 = {0,0}, y1 = {0,0}, y2 = {0,0}, y3 = {0,0};
    v2f64 t0, t1, t2, t3, t4, t5, t6, t7, t8, t9, t10, t11, t12, t13, t14, t15;
    v2f64 t16, t17, t18, t19, t20, t21, t22, t23, t24, t25, t26, t27, t28, t29;
-    v2f64 t30, t31, tp0 = 0.0, tp1 = 0.0, tp2 = 0.0, tp3 = 0.0, tp4 = 0.0, tp5 = 0.0, tp6 = 0.0, tp7 = 0.0;
+    v2f64 t30, t31, tp0 = {0,0}, tp1 = {0,0}, tp2 = {0,0}, tp3 = {0,0}, tp4 = {0,0}, tp5 = {0,0}, tp6 = {0,0}, tp7 = {0,0};

    v_alpha = COPY_DOUBLE_TO_VECTOR(alpha);

@@ -41,8 +41,11 @@ FLOAT CNAME(BLASLONG n, FLOAT *x, BLASLONG inc_x, FLOAT *y, BLASLONG inc_y)

 	while(i < n)
 	{
-
-		dot += y[iy] * x[ix] ;
+#if defined(DSDOT)
+		dot += (double)y[iy] * (double)x[ix] ;
+#else
+		dot += y[iy] * x[ix];
+#endif				
 		ix  += inc_x ;
 		iy  += inc_y ;
 		i++ ;
@@ -423,9 +423,9 @@ int CNAME(BLASLONG m, BLASLONG n, BLASLONG dummy1, FLOAT alpha, FLOAT *A,
    FLOAT *y_org = y;
    FLOAT *pa0, *pa1, *pa2, *pa3, *pa4, *pa5, *pa6, *pa7;
    FLOAT temp, temp0, temp1, temp2, temp3, temp4, temp5, temp6, temp7;
-    v4f32 v_alpha, x0, x1, y0 = 0.0, y1 = 0.0;
+    v4f32 v_alpha, x0, x1, y0 = {0,0,0,0}, y1 = {0,0,0,0};
    v4f32 t0, t1, t2, t3, t4, t5, t6, t7, t8, t9, t10, t11, t12, t13, t14, t15;
-    v4f32 tp0 = 0.0, tp1 = 0.0, tp2 = 0.0, tp3 = 0.0, tp4 = 0.0, tp5 = 0.0, tp6 = 0.0, tp7 = 0.0;
+    v4f32 tp0 = {0,0,0,0}, tp1 = {0,0,0,0}, tp2 = {0,0,0,0}, tp3 = {0,0,0,0}, tp4 = {0,0,0,0}, tp5 = {0,0,0,0}, tp6 = {0,0,0,0}, tp7 = {0,0,0,0};

    v_alpha = COPY_FLOAT_TO_VECTOR(alpha);

@@ -54,3 +54,6 @@ ZTRSMKERNEL_LN	=  ztrsm_kernel_power6_LN.S
 ZTRSMKERNEL_LT	=  ztrsm_kernel_power6_LT.S
 ZTRSMKERNEL_RN	=  ztrsm_kernel_power6_LT.S
 ZTRSMKERNEL_RT	=  ztrsm_kernel_power6_RT.S
+
+CROTKERNEL = ../arm/zrot.c
+ZROTKERNEL = ../arm/zrot.c
@@ -133,7 +133,7 @@ ZNRM2KERNEL  = ../arm/znrm2.c
 #
 SROTKERNEL   = srot.c
 DROTKERNEL   = drot.c
-#CROTKERNEL   = ../arm/zrot.c
+CROTKERNEL   = zrot.c
 ZROTKERNEL   = zrot.c
 #
 SSCALKERNEL  = sscal.c
@@ -647,7 +647,9 @@ static int get_l2_size_old(void){
      return 6144;
    }
  }
-  return 0;
+//  return 0;
+fprintf (stderr,"OpenBLAS WARNING - could not determine the L2 cache size on this system, assuming 256k\n");
+return 256;
 }
 #endif

@@ -660,6 +662,10 @@ static __inline__ int get_l2_size(void){
  l2 = BITMASK(ecx, 16, 0xffff);

 #ifndef ARCH_X86
+  if (l2 <= 0) {
+     fprintf (stderr,"OpenBLAS WARNING - could not determine the L2 cache size on this system, assuming 256k\n");
+     return 256;
+  }
  return l2;

 #else
@@ -871,6 +877,22 @@ static void init_parameter(void) {
 #endif
 #endif

+#ifdef SKYLAKEX
+
+#ifdef DEBUG
+  fprintf(stderr, "SkylakeX\n");
+#endif
+
+  TABLE_NAME.sgemm_p = SGEMM_DEFAULT_P;
+  TABLE_NAME.dgemm_p = DGEMM_DEFAULT_P;
+  TABLE_NAME.cgemm_p = CGEMM_DEFAULT_P;
+  TABLE_NAME.zgemm_p = ZGEMM_DEFAULT_P;
+#ifdef EXPRECISION
+  TABLE_NAME.qgemm_p = QGEMM_DEFAULT_P;
+  TABLE_NAME.xgemm_p = XGEMM_DEFAULT_P;
+#endif
+#endif
+

 #ifdef OPTERON

@@ -1,3 +1 @@
 include $(KERNELDIR)/KERNEL.PENRYN
-SSWAPKERNEL  = ../arm/swap.c
-DSWAPKERNEL  = ../arm/swap.c
@@ -138,6 +138,14 @@
 /* INCX != 1 or INCY != 1 */

 .L14:
+	cmpl	$0, %ebx
+	jne	.L141
+	cmpl	$0, %ecx
+	jne	.L141
+/* INCX == 0 and INCY == 0 */	
+	jmp	.L27
+
+.L141:	
 	movl	%edx, %eax
 	sarl	$2,   %eax
 	jle	.L28
@@ -62,7 +62,7 @@
 #define PREFETCHSIZE  (8 * 21 + 4)
 #endif

-#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL)
+#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL) || defined (SKYLAKEX)
 #define PREFETCH     prefetcht0
 #define PREFETCHSIZE  (8 * 21 + 4)
 #endif
@@ -62,7 +62,7 @@
 #define PREFETCHSIZE  (8 * 21 + 4)
 #endif

-#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL)
+#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL) || defined (SKYLAKEX)
 #define PREFETCH     prefetcht0
 #define PREFETCHSIZE  (8 * 21 + 4)
 #endif
@@ -62,7 +62,7 @@
 #define PREFETCHSIZE  (8 * 21 + 4)
 #endif

-#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL)
+#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL) || defined (SKYLAKEX)
 #define PREFETCH     prefetcht0
 #define PREFETCHSIZE  (8 * 21 + 4)
 #endif
@@ -62,7 +62,7 @@
 #define PREFETCHSIZE  (8 * 21 + 4)
 #endif

-#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL)
+#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL) || defined (SKYLAKEX)
 #define PREFETCH     prefetcht0
 #define PREFETCHSIZE  (8 * 21 + 4)
 #endif
@@ -62,7 +62,7 @@
 #define PREFETCHSIZE  (8 * 21 + 4)
 #endif

-#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL)
+#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL) || defined (SKYLAKEX)
 #define PREFETCH     prefetcht0
 #define PREFETCHSIZE  (8 * 21 + 4)
 #endif
@@ -62,7 +62,7 @@
 #define PREFETCHSIZE  (8 * 21 + 4)
 #endif

-#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL)
+#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL) || defined (SKYLAKEX)
 #define PREFETCH     prefetcht0
 #define PREFETCHSIZE  (8 * 21 + 4)
 #endif
@@ -61,7 +61,7 @@
 #define PREFETCHSIZE 84
 #endif

-#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL)
+#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL) || defined (SKYLAKEX)
 #define PREFETCH	prefetcht1
 #define PREFETCHSIZE 84
 #endif
@@ -63,7 +63,7 @@
 #define PREFETCHSIZE 84
 #endif

-#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL)
+#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL) || defined (SKYLAKEX)
 #define PREFETCH	prefetcht1
 #define PREFETCHSIZE 84
 #endif
@@ -61,7 +61,7 @@
 #define PREFETCHSIZE 84
 #endif

-#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL)
+#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL) || defined (SKYLAKEX)
 #define PREFETCH	prefetcht1
 #define PREFETCHSIZE 84
 #endif
@@ -63,7 +63,7 @@
 #define PREFETCHSIZE 84
 #endif

-#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL)
+#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL) || defined (SKYLAKEX)
 #define PREFETCH	prefetcht1
 #define PREFETCHSIZE 84
 #endif
@@ -61,7 +61,7 @@
 #define PREFETCHSIZE 84
 #endif

-#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL)
+#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL) || defined (SKYLAKEX)
 #define PREFETCH	prefetcht1
 #define PREFETCHSIZE 84
 #endif
@@ -0,0 +1,19 @@
+include $(KERNELDIR)/KERNEL.HASWELL
+
+SGEMMKERNEL    =  sgemm_kernel_16x4_skylakex.S
+
+
+DTRMMKERNEL    =  ../generic/trmmkernel_16x2.c
+DGEMMKERNEL    =  dgemm_kernel_16x2_skylakex.S
+DGEMMINCOPY    =  ../generic/gemm_ncopy_16.c
+DGEMMITCOPY    =  ../generic/gemm_tcopy_16.c
+DGEMMONCOPY    =  ../generic/gemm_ncopy_2.c
+DGEMMOTCOPY    =  ../generic/gemm_tcopy_2.c
+DGEMMINCOPYOBJ =  dgemm_incopy$(TSUFFIX).$(SUFFIX)
+DGEMMITCOPYOBJ =  dgemm_itcopy$(TSUFFIX).$(SUFFIX)
+DGEMMONCOPYOBJ =  dgemm_oncopy$(TSUFFIX).$(SUFFIX)
+DGEMMOTCOPYOBJ =  dgemm_otcopy$(TSUFFIX).$(SUFFIX)
+
+
+SGEMM_BETA = ../generic/gemm_beta.c
+DGEMM_BETA = ../generic/gemm_beta.c
@@ -33,7 +33,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #include "caxpy_microk_steamroller-2.c"
 #elif defined(BULLDOZER)
 #include "caxpy_microk_bulldozer-2.c"
-#elif defined(HASWELL) || defined(ZEN)
+#elif defined(HASWELL) || defined(ZEN) || defined(SKYLAKEX)
 #include "caxpy_microk_haswell-2.c"
 #elif defined(SANDYBRIDGE)
 #include "caxpy_microk_sandy-2.c"
@@ -34,7 +34,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #include "cdot_microk_bulldozer-2.c"
 #elif defined(STEAMROLLER) || defined(PILEDRIVER)  || defined(EXCAVATOR)
 #include "cdot_microk_steamroller-2.c"
-#elif defined(HASWELL) || defined(ZEN)
+#elif defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #include "cdot_microk_haswell-2.c"
 #elif defined(SANDYBRIDGE)
 #include "cdot_microk_sandy-2.c"
@@ -29,7 +29,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #include <stdio.h>
 #include "common.h"

-#if defined(HASWELL) || defined(ZEN)
+#if defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #include "cgemv_n_microk_haswell-4.c"
 #elif defined(BULLDOZER) || defined(PILEDRIVER) || defined(STEAMROLLER) || defined(EXCAVATOR)
 #include "cgemv_n_microk_bulldozer-4.c"
@@ -28,7 +28,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

 #include "common.h"

-#if defined(HASWELL) || defined(ZEN)
+#if defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #include "cgemv_t_microk_haswell-4.c"
 #elif defined(BULLDOZER) || defined(PILEDRIVER) || defined(STEAMROLLER)  || defined(EXCAVATOR)
 #include "cgemv_t_microk_bulldozer-4.c"
@@ -28,7 +28,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #include "common.h"


-#if defined(HASWELL) || defined(ZEN)
+#if defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #include "cscal_microk_haswell-2.c"
 #elif defined(BULLDOZER)  || defined(PILEDRIVER)
 #include "cscal_microk_bulldozer-2.c"
@@ -37,7 +37,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #include "daxpy_microk_steamroller-2.c"
 #elif defined(PILEDRIVER)
 #include "daxpy_microk_piledriver-2.c"
-#elif defined(HASWELL) || defined(ZEN)
+#elif defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #include "daxpy_microk_haswell-2.c"
 #elif defined(SANDYBRIDGE)
 #include "daxpy_microk_sandy-2.c"
@@ -37,7 +37,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #include "ddot_microk_piledriver-2.c"
 #elif defined(NEHALEM) 
 #include "ddot_microk_nehalem-2.c"
-#elif defined(HASWELL) || defined(ZEN)
+#elif defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #include "ddot_microk_haswell-2.c"
 #elif defined(SANDYBRIDGE)
 #include "ddot_microk_sandy-2.c"
@@ -31,7 +31,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

 #if defined(NEHALEM)
 #include "dgemv_n_microk_nehalem-4.c"
-#elif defined(HASWELL) || defined(ZEN) || defined(STEAMROLLER) || defined(EXCAVATOR)
+#elif defined(HASWELL) || defined(ZEN) || defined(STEAMROLLER) || defined(EXCAVATOR) || defined (SKYLAKEX)
 #include "dgemv_n_microk_haswell-4.c"
 #endif

@@ -28,7 +28,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

 #include "common.h"

-#if defined(HASWELL) || defined(ZEN) || defined(STEAMROLLER)  || defined(EXCAVATOR)
+#if defined(HASWELL) || defined(ZEN) || defined(STEAMROLLER)  || defined(EXCAVATOR) || defined (SKYLAKEX)
 #include "dgemv_t_microk_haswell-4.c"
 #endif

@@ -31,7 +31,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #include "dscal_microk_bulldozer-2.c"
 #elif defined(SANDYBRIDGE)
 #include "dscal_microk_sandy-2.c"
-#elif defined(HASWELL) || defined(ZEN)
+#elif defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #include "dscal_microk_haswell-2.c"
 #endif

@@ -30,7 +30,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

 #if defined(BULLDOZER) || defined(PILEDRIVER) || defined(STEAMROLLER) || defined(EXCAVATOR)
 #include "dsymv_L_microk_bulldozer-2.c"
-#elif defined(HASWELL) || defined(ZEN)
+#elif defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #include "dsymv_L_microk_haswell-2.c"
 #elif defined(SANDYBRIDGE)
 #include "dsymv_L_microk_sandy-2.c"
@@ -31,7 +31,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

 #if defined(BULLDOZER) || defined(PILEDRIVER) || defined(STEAMROLLER)  || defined(EXCAVATOR)
 #include "dsymv_U_microk_bulldozer-2.c"
-#elif defined(HASWELL) || defined(ZEN)
+#elif defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #include "dsymv_U_microk_haswell-2.c"
 #elif defined(SANDYBRIDGE)
 #include "dsymv_U_microk_sandy-2.c"
@@ -31,7 +31,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

 #if defined(NEHALEM)
 #include "saxpy_microk_nehalem-2.c"
-#elif defined(HASWELL) || defined(ZEN)
+#elif defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #include "saxpy_microk_haswell-2.c"
 #elif defined(SANDYBRIDGE)
 #include "saxpy_microk_sandy-2.c"
@@ -34,7 +34,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #include "sdot_microk_steamroller-2.c"
 #elif defined(NEHALEM)
 #include "sdot_microk_nehalem-2.c"
-#elif defined(HASWELL) || defined(ZEN)
+#elif defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #include "sdot_microk_haswell-2.c"
 #elif defined(SANDYBRIDGE)
 #include "sdot_microk_sandy-2.c"
@@ -35,7 +35,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #include "sgemv_n_microk_nehalem-4.c"
 #elif defined(SANDYBRIDGE)
 #include "sgemv_n_microk_sandy-4.c"
-#elif defined(HASWELL) || defined(ZEN)
+#elif defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #include "sgemv_n_microk_haswell-4.c"
 #endif

@@ -34,7 +34,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #include "sgemv_t_microk_bulldozer-4.c"
 #elif defined(SANDYBRIDGE)
 #include "sgemv_t_microk_sandy-4.c"
-#elif defined(HASWELL) || defined(ZEN)
+#elif defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #include "sgemv_t_microk_haswell-4.c"
 #endif

@@ -32,7 +32,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #include "ssymv_L_microk_bulldozer-2.c"
 #elif defined(NEHALEM)
 #include "ssymv_L_microk_nehalem-2.c"
-#elif defined(HASWELL) || defined(ZEN)
+#elif defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #include "ssymv_L_microk_haswell-2.c"
 #elif defined(SANDYBRIDGE)
 #include "ssymv_L_microk_sandy-2.c"
@@ -33,7 +33,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #include "ssymv_U_microk_bulldozer-2.c"
 #elif defined(NEHALEM)
 #include "ssymv_U_microk_nehalem-2.c"
-#elif defined(HASWELL) || defined(ZEN)
+#elif defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #include "ssymv_U_microk_haswell-2.c"
 #elif defined(SANDYBRIDGE)
 #include "ssymv_U_microk_sandy-2.c"
@@ -57,7 +57,7 @@
 #define PREFETCHSIZE	(16 * 12)
 #endif

-#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL) || defined(ZEN)
+#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #define PREFETCH	prefetcht0
 #define PREFETCHW	prefetcht0
 #define PREFETCHSIZE	(16 * 12)
@@ -57,7 +57,7 @@
 #define PREFETCHSIZE	(16 * 12)
 #endif

-#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL) || defined(ZEN)
+#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #define PREFETCH	prefetcht0
 #define PREFETCHW	prefetcht0
 #define PREFETCHSIZE	(16 * 12)
@@ -57,7 +57,7 @@
 #define PREFETCHSIZE	(16 * 12)
 #endif

-#if defined(NEHALEM) || defined(SANDYBRIDGE)  || defined(HASWELL) || defined(ZEN)
+#if defined(NEHALEM) || defined(SANDYBRIDGE)  || defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #define PREFETCH	prefetcht0
 #define PREFETCHW	prefetcht0
 #define PREFETCHSIZE	(16 * 12)
@@ -57,7 +57,7 @@
 #define PREFETCHSIZE	(16 * 12)
 #endif

-#if defined(NEHALEM) || defined(SANDYBRIDGE)  || defined(HASWELL) || defined(ZEN)
+#if defined(NEHALEM) || defined(SANDYBRIDGE)  || defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #define PREFETCH	prefetcht0
 #define PREFETCHW	prefetcht0
 #define PREFETCHSIZE	(16 * 24)
@@ -33,7 +33,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #include "zaxpy_microk_bulldozer-2.c"
 #elif defined(PILEDRIVER) || defined(STEAMROLLER) || defined(EXCAVATOR)
 #include "zaxpy_microk_steamroller-2.c"
-#elif defined(HASWELL) || defined(ZEN)
+#elif defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #include "zaxpy_microk_haswell-2.c"
 #elif defined(SANDYBRIDGE)
 #include "zaxpy_microk_sandy-2.c"
@@ -33,7 +33,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #include "zdot_microk_bulldozer-2.c"
 #elif defined(STEAMROLLER) || defined(PILEDRIVER) || defined(EXCAVATOR)
 #include "zdot_microk_steamroller-2.c"
-#elif defined(HASWELL) || defined(ZEN)
+#elif defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #include "zdot_microk_haswell-2.c"
 #elif defined(SANDYBRIDGE)
 #include "zdot_microk_sandy-2.c"
@@ -30,7 +30,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #include "common.h"


-#if defined(HASWELL) || defined(ZEN)
+#if defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #include "zgemv_n_microk_haswell-4.c"
 #elif defined(SANDYBRIDGE)
 #include "zgemv_n_microk_sandy-4.c"
@@ -31,7 +31,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

 #if defined(BULLDOZER) || defined(PILEDRIVER) || defined(STEAMROLLER)  || defined(EXCAVATOR)
 #include "zgemv_t_microk_bulldozer-4.c"
-#elif defined(HASWELL) || defined(ZEN)
+#elif defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #include "zgemv_t_microk_haswell-4.c"
 #endif

@@ -28,7 +28,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 #include "common.h"


-#if defined(HASWELL) || defined(ZEN)
+#if defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #include "zscal_microk_haswell-2.c"
 #elif defined(BULLDOZER)  || defined(PILEDRIVER)
 #include "zscal_microk_bulldozer-2.c"
@@ -57,7 +57,7 @@
 #define PREFETCHSIZE	(16 * 24)
 #endif

-#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL) || defined(ZEN)
+#if defined(NEHALEM) || defined(SANDYBRIDGE) || defined(HASWELL) || defined(ZEN) || defined (SKYLAKEX)
 #define PREFETCH	prefetcht0
 #define PREFETCHW	prefetcht0
 #define PREFETCHSIZE	(16 * 24)
--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
Martin Kroeker	de8fff671d	Revert "Use usleep instead of sched_yield by default"	2018-06-11 17:05:27 +02:00
Martin Kroeker	6f71c0fce4	Return a somewhat sane default value for L2 cache size if cpuid retur… (#1611 ) * Return a somewhat sane default value for L2 cache size if cpuid returned something unexpected Fixes #1610, the KVM hypervisor on Google Chromebooks returning zero for CPUID 0x80000006, causing DYNAMIC_ARCH builds of OpenBLAS to hang	2018-06-11 13:26:19 +02:00
Martin Kroeker	3313e4b946	Merge pull request #1608 from martin-frbg/issue874 Enable parallel make on MS Windows by default	2018-06-09 19:57:33 +02:00
Martin Kroeker	e9cd11768c	Enable parallel make on MS Windows by default fixes #874	2018-06-09 17:54:36 +02:00
Martin Kroeker	0297b3211a	Merge pull request #1605 from oon3m0oo/develop Improve performance of GEMM for small matrices when SMP is defined.	2018-06-09 12:42:34 +02:00
Craig Donner	66316b9f4c	Improve performance of GEMM for small matrices when SMP is defined. Always checking num_cpu_avail() regardless of whether threading will actually be used adds noticeable overhead for small matrices. Most other uses of num_cpu_avail() do so only if threading will be used, so do the same here.	2018-06-07 15:29:13 +01:00
Martin Kroeker	6adc4b7b36	Merge pull request #1601 from martin-frbg/zaxpy Use a single thread for small input size in zaxpy	2018-06-07 14:09:58 +02:00
Martin Kroeker	2ade0ef085	Merge pull request #1600 from martin-frbg/noyield Use usleep instead of sched_yield by default	2018-06-07 12:42:00 +02:00
Martin Kroeker	e8880c1699	Use a single thread for small input size copies daxpy improvement from #27, see #1560	2018-06-07 10:26:55 +02:00
Martin Kroeker	ed7c4a043b	Use usleep instead of sched_yield by default sched_yield only burns cpu cycles, fixes #900, see also #923, #1560	2018-06-07 10:18:26 +02:00
Martin Kroeker	cf234a0561	Merge pull request #1589 from fenrus75/skylakex Initial support for SkylakeX / AVX512	2018-06-06 22:07:09 +02:00
Martin Kroeker	ae2a33128b	Merge pull request #1599 from martin-frbg/c_check_avx512 Improved AVX512 test case for c_check	2018-06-06 18:42:42 +02:00
Martin Kroeker	e4718b1fee	Better AVX512 test case	2018-06-06 16:51:30 +02:00
Martin Kroeker	9b87b64262	Improve AVX512 testcase clang 3.4 managed to accept the original test code, only to fail on the actual Skylake asm later	2018-06-06 16:49:00 +02:00
Martin Kroeker	0218b884c1	Merge pull request #1598 from martin-frbg/issue1593-2 Restore _Atomic define before stdatomic.h for old gcc	2018-06-06 12:48:26 +02:00
Martin Kroeker	83da278093	Update common.h	2018-06-06 09:27:49 +02:00
Martin Kroeker	358d4df2bd	Merge branch 'develop' into issue1593-2	2018-06-06 09:21:41 +02:00
Martin Kroeker	06d43760e4	Restore _Atomic define before stdatomic.h for old gcc see #1593	2018-06-06 09:18:10 +02:00
Martin Kroeker	a4af8861ff	Merge pull request #1597 from martin-frbg/cmake-avx512 Check build system support for AVX512 instructions	2018-06-06 07:22:20 +02:00
Martin Kroeker	7fb62aed7e	Check build system support for AVX512 instructions	2018-06-05 23:29:33 +02:00
Martin Kroeker	f6021c798d	Re-enable QUIET_MAKE	2018-06-05 19:09:38 +02:00
Martin Kroeker	e8002536ec	disable quiet_make for the moment	2018-06-05 18:23:01 +02:00
Martin Kroeker	ce6317f6c0	Merge pull request #1594 from martin-frbg/issue1593 Fix inverted condition in _Atomic declaration	2018-06-05 16:02:51 +02:00
Martin Kroeker	15a78d6b66	export NO_AVX512 setting	2018-06-05 15:58:34 +02:00
Martin Kroeker	354a976a59	Fix inverted condition in _Atomic declaration fixes #1593	2018-06-05 10:31:34 +02:00
Martin Kroeker	38ad05bd04	Extend loop range to find SkylakeX in force_coretype	2018-06-05 10:26:49 +02:00
Martin Kroeker	b7feded85a	Propagate NO_AVX512 via CCOMMON_OPT	2018-06-05 10:24:05 +02:00
Martin Kroeker	dc9fe05ab5	Update cpuid_x86.c	2018-06-04 17:10:19 +02:00
Martin Kroeker	8be027e4c6	Update dynamic.c	2018-06-04 14:36:39 +02:00
Martin Kroeker	ac7b6e3e9a	Fix misplaced endif	2018-06-04 08:23:40 +02:00
Martin Kroeker	fc66a0ec0b	Merge pull request #1590 from martin-frbg/avx512_check Disable AVX512 (Skylake X) support if the build system is too old	2018-06-04 08:18:38 +02:00
Arjan van de Ven	89372e0993	Use AVX512 also for DGEMM this required switching to the generic gemm_beta code (which is faster anyway on SKX) for both DGEMM and SGEMM Performance for the not-retuned version is in the 30% range	2018-06-03 22:17:27 +00:00
Martin Kroeker	ef626c6824	typo fix	2018-06-04 00:13:19 +02:00
Martin Kroeker	83fec56a3f	Disable AVX512 (Skylake X) support if the build system is too old	2018-06-04 00:01:11 +02:00
Martin Kroeker	5a51cf4576	Separate Skylake X from Skylake	2018-06-03 23:41:33 +02:00
Martin Kroeker	5a92b311e0	Separate Skylake X from Skylake	2018-06-03 23:29:07 +02:00
Martin Kroeker	a7d0f49cec	Add SKYLAKEX to DYNAMIC_CORE list only if AVX512 is available	2018-06-03 23:13:25 +02:00
Martin Kroeker	f1fb9a4745	Propagate NO_AVX512 if needed	2018-06-03 13:48:27 +02:00
Martin Kroeker	0023515733	Typo fix (misplaced parenthesis)	2018-06-03 13:22:59 +02:00
Arjan van de Ven	99c7bba8e4	Initial support for SkylakeX / AVX512 This patch adds the basic infrastructure for adding the SkylakeX (Intel Skylake server) target. The SkylakeX target will use the AVX512 (AVX512VL level) instruction set, which brings 2 basic things: 1) 512 bit wide SIMD (2x width of AVX2) 2) 32 SIMD registers (2x the number on AVX2) This initial patch only contains a trivial transofrmation of the Haswell SGEMM kernel to AVX512VL; more will follow later but this patch aims to get the infrastructure in place for this "later". Full performance tuning has not been done yet; with more registers and wider SIMD it's in theory possible to retune the kernels but even without that there's an interesting enough performance increase (30-40% range) with just this change.	2018-06-03 07:58:52 +00:00
Martin Kroeker	36c4523d85	Merge pull request #1587 from matthew-brett/fix-compile-error-early-glibc Revert "take out unused variables"	2018-06-02 10:02:38 +02:00
Matthew Brett	a8002e283a	Revert "take out unused variables" This reverts commit `e5752ff9b3`. The variables i and n are used in the `#if !__GLIBC_PREREQ(2, 7)` branch. Closes gh-1586.	2018-06-01 23:20:00 +01:00
Martin Kroeker	401adddb2b	Merge pull request #1585 from martin-frbg/lapack-253 Fixes from Lapack-Reference PR 253	2018-06-01 18:59:33 +02:00
Martin Kroeker	c5b13d4e10	Fixes from netlib PR 253	2018-06-01 15:14:45 +02:00
Martin Kroeker	677e42d7b0	Fixes from netlib PR 253 When minimal workspace is given in ?hesv_aa, ?sysv_aa, ?hesv_aa_2stage, ?sysv_aa_2stage, now no error is given Quick return for ?laqr1	2018-06-01 15:12:59 +02:00
Martin Kroeker	e2a8c35e5a	Fixes from netlib PR253 LAPACKE interfaces for Aasen's functions now call ?sytrf_aa and ?hetrf_aa instead of ?sytrf and ?hetrf	2018-06-01 15:08:14 +02:00
Martin Kroeker	1a49fb1c05	Merge pull request #1584 from martin-frbg/issue1503 Work around name clash with Windows10's winnt.h	2018-05-31 21:56:04 +02:00
Martin Kroeker	8562d5787a	Merge pull request #1583 from martin-frbg/issue1575 Handle INCX=0,INCY=0 case	2018-05-31 21:55:26 +02:00
Martin Kroeker	93f1eb09c3	Merge pull request #1582 from martin-frbg/develop-031 Update version number on the develop branch to 0.3.1.dev	2018-05-31 21:55:07 +02:00
Martin Kroeker	c90bbda3df	Merge pull request #1581 from martin-frbg/issue1574-2 Fix paths to LIN and EIG tests	2018-05-31 21:54:45 +02:00
Martin Kroeker	7df8c4f76f	typo fix	2018-05-31 17:23:08 +02:00
Martin Kroeker	2fc748bf72	Restore optimized swap kernel now that we have a proper fix	2018-05-31 13:41:12 +02:00
Martin Kroeker	a91f1587b9	Work around name clash with Windows10's winnt.h fixes #1503	2018-05-31 13:26:00 +02:00
Martin Kroeker	d1b7be14aa	Handle INCX=0,INCY=0 case Fixes #1575 (sswap/dswap failing the swap utest on x86) as suggested by atsampson.	2018-05-31 12:52:04 +02:00
Martin Kroeker	b491b10057	Update version to 0.3.1.dev	2018-05-31 12:44:36 +02:00
Martin Kroeker	5fae96fb70	Update version to 0.3.1.dev	2018-05-31 12:43:45 +02:00
Martin Kroeker	a7dbd4c57d	Fix paths to LIN and EIG tests should fix 1574	2018-05-31 11:19:33 +02:00
Martin Kroeker	2cae104b5e	Merge pull request #1579 from martin-frbg/issue1574 Adapt lapack-test and blas-test to changes in netlib directory layout	2018-05-29 22:02:06 +02:00
Martin Kroeker	908d40be71	Adapt lapack-test and blas-test to changes in netlib directory layout partial fix for #1574 - the problem with lapack_testing.py looks like an upstream bug	2018-05-29 14:27:46 +02:00
Zhang Xianyi	43e592ceb3	Add -lm for Android. Conflicts: exports/Makefile	2018-05-24 21:02:42 +08:00
Martin Kroeker	f0f27868d8	Merge pull request #1572 from martin-frbg/issue1571 Use the new zrot.c on POWER8 for crot as well	2018-05-23 22:55:37 +02:00
Martin Kroeker	961d25e9c7	Use the new zrot.c on POWER8 for crot as well fixes #1571 (the old zrot.S assembly does not handle incx=0 correctly)	2018-05-23 22:54:39 +02:00
Martin Kroeker	f5959f2543	Merge pull request #1567 from martin-frbg/mipstrmm Revert " Switch mips32 target to USE_TRMM to fix complex TRMM"	2018-05-17 20:50:23 +02:00
Martin Kroeker	82012b960b	Revert " Switch mips32 target to USE_TRMM to fix complex TRMM" ... as it was just a silly workaround for the issue seen in #1563, caused by #1419	2018-05-17 20:30:03 +02:00
Martin Kroeker	8dd3515fa2	Merge pull request #1565 from martin-frbg/mipstypo Remove extraneous brace from previous commit of mips dsdot fix	2018-05-17 20:22:58 +02:00
Martin Kroeker	95f7f0229c	Remove extraneous brace from previous commit	2018-05-17 18:43:59 +02:00
Martin Kroeker	5082fe4306	Merge pull request #1564 from martin-frbg/issue1563 Revert changes from PR#1419	2018-05-17 14:04:13 +02:00
Martin Kroeker	7a7619af6d	Revert changes from PR#1419 at least one of these changes apparently is an oversimplification, leading to TRMM breakage on some platforms as observed in #1563	2018-05-17 11:40:08 +02:00
Martin Kroeker	9a400b7014	Merge pull request #1562 from martin-frbg/issue1561 Use correct data type for initializers of v2f64, v4f32	2018-05-15 17:46:09 +02:00
Martin Kroeker	893b535540	Use correct data type for initializers of v2f64, v4f32 Fixes #1561	2018-05-15 14:42:12 +02:00
Martin Kroeker	6791294312	Merge pull request #1559 from martin-frbg/buildconf Add build-time configuration options to pkgconfig file	2018-05-14 18:49:53 +02:00
Martin Kroeker	ddb8b124de	Merge pull request #1558 from martin-frbg/instpc Overwrite any pre-existing openblas.pc rather than append to it	2018-05-14 17:38:12 +02:00
Martin Kroeker	191746c493	Merge pull request #1557 from martin-frbg/getconfig Add threading and OpenMP information to output	2018-05-14 17:37:55 +02:00
Martin Kroeker	eb9b021d38	Add build-time configuration options to pkgconfig file	2018-05-14 00:10:15 +02:00
Martin Kroeker	7d7564568c	Add build-time configuration options to pkgconfig file	2018-05-14 00:09:35 +02:00
Martin Kroeker	a07843bc93	Overwrite any pre-existing openblas.pc rather than append to it	2018-05-12 22:11:27 +02:00
Martin Kroeker	41ae8e8d67	Add threading and OpenMP information to output For #1416 and #1529, more information about the options OpenBLAS was built with is needed. Additionally we may want to add this data to the openblas.pc file (but not all projects use pkgconfig, and as far as I am aware the cmake module for accessing it does not make such "private" declarations available)	2018-05-12 12:11:38 +02:00
Zhang Xianyi	9c1aa0b0fe	Merge pull request #1556 from WestAlgo/develop move _Atomic define to common.h	2018-05-11 17:02:47 +08:00
zhiyong.dang	53457f222f	move _Atomic define to common.h	2018-05-11 00:13:16 -07:00
Zhang Xianyi	458e3af5b1	Merge pull request #1555 from WestAlgo/develop Change _STDC_VERSION__ to __STDC_VERSION__	2018-05-11 12:25:24 +08:00
Zhiyong Dang	3716267124	Change _STDC_VERSION__ to __STDC_VERSION__ Change-Id: Id3fa4e8d9eedd4ef7230df69b611e7f397301a42	2018-05-11 12:15:08 +08:00
Zhang Xianyi	50acc40613	Merge pull request #1536 from WestAlgo/develop Fix race condition in blas_server_omp.c	2018-05-11 10:09:14 +08:00
Martin Kroeker	c720f1f019	Merge pull request #1554 from martin-frbg/lapack-249 LAPACKE fixes from lapack PR249	2018-05-10 15:32:08 +02:00
Martin Kroeker	d7d950fcf2	LAPACKE fixes from lapack PR249 Copied from Reference-LAPACK/lapack#249, this fixes out-of-bounds memory accesses in the nancheck calls of the LAPACKE lacgv, lassq,larfg,larfb,larfx and mtr functions	2018-05-10 13:15:42 +02:00
Martin Kroeker	12398e53ce	Merge pull request #1553 from martin-frbg/ifort-openmpflag Change -openmp to -fopenmp for ifort entry as well	2018-05-09 14:39:52 +02:00
Martin Kroeker	193f835662	Change -openmp to -fopenmp for ifort entry as well	2018-05-09 12:34:09 +02:00
Martin Kroeker	7e3151ead7	Merge pull request #1551 from martin-frbg/f_check_fix Fixes for ifort 2018	2018-05-09 09:02:52 +02:00
Martin Kroeker	e3a069f108	Merge pull request #1550 from martin-frbg/ifort-openmpflag Update compiler flag for openmp use with ICC	2018-05-09 09:02:38 +02:00
Martin Kroeker	6fff8c626a	Merge pull request #1549 from martin-frbg/fix_ompcheck Drop C-style "L" suffx from OPENMP version number tests in the LAPACK source	2018-05-08 23:52:55 +02:00
Martin Kroeker	d2b9389f1b	Fixes for ifort 2018 1. the already deprecated -openmp option was removed in 2018, switch to -fopenmp 2. add leading blank in search for "zho_ge__" symbol to work around misleading tags in the 2018 assembly Expected to fix #1548	2018-05-08 21:55:37 +02:00
Martin Kroeker	65b8a5c5d8	Update compiler flag for openmp use with ICC The deprecated -openmp option was finally removed in favor of -qopenmp or -fopenmp, picking the latter to stay compatible with Intel compiler versions before 2015 (when -q options were introduced). Fixes #1546	2018-05-08 21:47:10 +02:00
Martin Kroeker	9795adc7ef	Drop C-style "L" suffix from OPENMP version number in check	2018-05-08 21:39:42 +02:00
Martin Kroeker	1a8e487c4a	Drop C-style "L" suffix from OPENMP version number in check	2018-05-08 21:38:25 +02:00
Martin Kroeker	5966fd52a2	Drop C-style "L" suffix from OPENMP version number in check	2018-05-08 21:36:56 +02:00
Martin Kroeker	dbafe6357b	Merge pull request #1543 from martin-frbg/mips32 Fix MIPS32 build and add MIPS 1004K cpu (MT7621 SOC)	2018-05-02 22:47:45 +02:00
Martin Kroeker	71051259e0	Restore compiler options for mips P5600 target	2018-05-02 20:37:06 +02:00
Martin Kroeker	73cc321190	Add MIPS 1004K target	2018-05-02 20:27:56 +02:00
Martin Kroeker	018f2dad27	Switch mips32 target to USE_TRMM to fix complex TRMM	2018-05-02 20:25:32 +02:00
Martin Kroeker	9d5098dbc9	Add MIPS 1004K target (Mediatek MT7621 SOC)	2018-05-02 20:20:44 +02:00
Martin Kroeker	d94d7baf7e	Add mips32r2 api target	2018-05-02 20:17:26 +02:00
Martin Kroeker	3af1b5c805	Make cpuid_mips compile again and add 1004K cpu	2018-05-02 20:12:25 +02:00
Martin Kroeker	88e224f4c0	Merge pull request #1542 from martin-frbg/quickdiv64 Avoid out-of-bounds accesses in blas_quickdivide on big X86 systems	2018-05-02 18:11:50 +02:00
Martin Kroeker	d0c0506588	Omit the divide table overflow check on small systems	2018-05-02 14:44:50 +02:00
Martin Kroeker	e93355e5e1	Omit the table overflow check when building for small systems	2018-05-02 14:43:08 +02:00
Martin Kroeker	c1eb06e102	Update common_x86_64.h	2018-04-29 14:40:12 +02:00
Martin Kroeker	8145ecd70b	Avoid out-of-bounds reads from blas_quick_divide_table on big systems	2018-04-29 14:38:55 +02:00
Martin Kroeker	26ce518d46	Avoid out of bounds reads from blas_quick_divide_table on big systems Should fix #1541	2018-04-29 14:34:33 +02:00
Martin Kroeker	1d27fa8507	Merge pull request #1539 from martin-frbg/ztrmv-1332 Disable multithreading in ztrmv	2018-04-27 23:10:21 +02:00
Martin Kroeker	802cf6b22d	Merge pull request #1486 from martin-frbg/atomic Use _Atomic instead of volatile for thread safety where C11 is supported	2018-04-27 23:09:57 +02:00
Martin Kroeker	894433a7c7	Update Makefile.rule	2018-04-27 12:08:06 +02:00
Zhiyong Dang	1b83341d19	Fix race condition in blas_server_omp.c Change-Id: Ic896276cd073d6b41930c7c5a29d66348cd1725d	2018-04-27 17:00:42 +08:00
Martin Kroeker	954f1832de	Merge pull request #1540 from martin-frbg/mips32-zasum Fix typo in MIPS P5600 complex ASUM code selection	2018-04-25 23:23:00 +02:00
Martin Kroeker	941ad280a8	Fix typo in MIPS P5600 complex ASUM code selection	2018-04-25 22:50:10 +02:00
Martin Kroeker	a8ed428bab	Disable multithreading in ztrmv BLAS-Tester shows that the same problem exists as with DTRMV (issue #1332)	2018-04-25 22:35:46 +02:00
Martin Kroeker	1da365312a	Merge pull request #1538 from martin-frbg/arm7utest Fix handling of zero INCX, INCY in ArmV7 AXPY and ROT	2018-04-25 08:38:58 +02:00
Martin Kroeker	2d0929fa7c	Move the test for zero incx,incy in ARMV7 ROT to pass the related utest (see #1469)	2018-04-24 22:43:00 +02:00
Martin Kroeker	125343cc88	Drop test for zero incx,incy in armv7 AXPY ...to pass the related utest (see #1469)	2018-04-24 22:39:50 +02:00
Martin Kroeker	8a3b6fa108	Use generic zrot.c on ppc64/POWER6 to work around utest failure from … (#1535 ) * Use generic C implementation of zrot on ppc64/POWER6 to work around utest failure from #1469	2018-04-23 19:05:49 +02:00
Martin Kroeker	78694f1b7e	Merge pull request #1534 from xianyi/revert-1333-haswell32 Revert "Fix 32bit HASWELL builds"	2018-04-22 23:34:17 +02:00
Martin Kroeker	9c5518319a	Revert "Fix 32bit HASWELL builds"	2018-04-22 20:20:04 +02:00
Martin Kroeker	86f49c529d	Merge pull request #1532 from martin-frbg/utest-cblas Do not try to build the fork utest when NO_CBLAS=1	2018-04-20 23:44:15 +02:00
Martin Kroeker	625c74a38f	fork utest depends on CBLAS	2018-04-20 15:43:59 +02:00
Martin Kroeker	5fcaca6438	fork utest depends on CBLAS	2018-04-20 15:42:13 +02:00
Martin Kroeker	4fcdd24459	Merge pull request #1530 from ashwinyes/develop_20180419_Tx2AutoDetect ARM64: Enable Auto Detection of ThunderX2T99	2018-04-19 14:10:57 +02:00
Ashwin Sekhar T K	68a3c4fca6	ARM64: Enable Auto Detection of ThunderX2T99	2018-04-19 09:05:25 +00:00
Martin Kroeker	0c4718c57a	Merge pull request #1523 from martin-frbg/utest_waith Include sys/types.h for proper typedefs related to wait()	2018-04-15 13:09:30 +02:00
Martin Kroeker	f29389c7ac	Merge pull request #1520 from martin-frbg/cpucounts Catch invalid cpu count returned by CPU_COUNT_S	2018-04-14 22:24:34 +02:00
Martin Kroeker	734d7c6a93	Include sys/types.h for proper typedefs related to wait() Should fix #1519	2018-04-14 18:59:46 +02:00
Martin Kroeker	7c861605b2	Catch invalid cpu count returned by CPU_COUNT_S mips32 was seen to return zero here, driving nthreads to zero with subsequent fpe in blas_quickdivide	2018-04-14 18:29:10 +02:00
Martin Kroeker	2ca0faf495	Merge pull request #1515 from martin-frbg/mipsdot Correct precision of mips dsdot	2018-04-11 08:21:25 +02:00
Martin Kroeker	0fe434598b	Fix precision of mips dsdot	2018-04-10 23:30:59 +02:00
Martin Kroeker	15c437e092	Merge pull request #1512 from ararslan/aa/travis-macos-2 Add macOS to the Travis testing matrix: Take 2!	2018-04-07 23:31:26 +02:00
Alex Arslan	b966bd79d5	Add a BINARY=32 build to macOS	2018-04-07 12:29:57 -07:00
Alex Arslan	2e988dbf35	Add macOS to the Travis testing matrix	2018-04-07 10:56:34 -07:00
Martin Kroeker	be6090d396	Merge pull request #1511 from xianyi/revert-1510-aa/travis-macos Revert "Add macOS to the Travis testing matrix"	2018-04-07 13:29:31 +02:00
Martin Kroeker	20c6c38e51	Merge branch 'develop' into atomic	2018-04-07 12:09:39 +02:00
Martin Kroeker	40160ff3c1	Use _Atomic instead of volatile for thread safety where C11 is supported	2018-03-10 00:15:44 +01:00
Martin Kroeker	6a99fcce94	Use _Atomic instead of volatile for thread safety where C11 is supported Suggested by dodomorandi in #660	2018-03-10 00:03:49 +01:00