Merge pull request #20 from xianyi/develop

rebase

getarch_2nd.c

@@ -4,7 +4,15 @@
 #else
 #include "config_kernel.h"
 #endif
-#include "common.h"
+#if (defined(__WIN32__) || defined(__WIN64__) || defined(__CYGWIN32__) || defined(__CYGWIN64__) || defined(_WIN32) || defined(_WIN64)) && defined(__64BIT__)
+typedef long long BLASLONG;
+typedef unsigned long long BLASULONG;
+#else
+typedef long BLASLONG;
+typedef unsigned long BLASULONG;
+#endif
+
+#include "param.h"
 
 int main(int argc, char **argv) {
 

kernel/power/KERNEL.POWER10

@@ -169,8 +169,13 @@ ZROTKERNEL   = zrot.c
 #
 SSCALKERNEL  = sscal.c
 DSCALKERNEL  = dscal.c
+ifeq ($(C_COMPILER), PGI)
+CSCALKERNEL  = ../arm/zscal.c
+ZSCALKERNEL  = ../arm/zscal.c
+else
 CSCALKERNEL  = zscal.c
 ZSCALKERNEL  = zscal.c
+endif
 #
 SSWAPKERNEL  = sswap.c
 DSWAPKERNEL  = dswap.c

kernel/power/KERNEL.POWER8

@@ -242,8 +242,13 @@ ZROTKERNEL   = zrot.c
 #
 SSCALKERNEL  = sscal.c
 DSCALKERNEL  = dscal.c
+ifeq ($(C_COMPILER), PGI)
+CSCALKERNEL  = ../arm/zscal.c
+ZSCALKERNEL  = ../arm/zscal.c
+else
 CSCALKERNEL  = zscal.c
 ZSCALKERNEL  = zscal.c
+endif
 #
 SSWAPKERNEL  = sswap.c
 DSWAPKERNEL  = dswap.c

kernel/power/KERNEL.POWER9

@@ -166,8 +166,13 @@ ZROTKERNEL   = zrot.c
 #
 SSCALKERNEL  = sscal.c
 DSCALKERNEL  = dscal.c
+ifeq ($(C_COMPILER), PGI)
+CSCALKERNEL  = ../arm/zscal.c
+ZSCALKERNEL  = ../arm/zscal.c
+else
 CSCALKERNEL  = zscal.c
 ZSCALKERNEL  = zscal.c
+endif
 #
 SSWAPKERNEL  = sswap.c
 DSWAPKERNEL  = dswap.c

kernel/x86_64/KERNEL

@@ -491,4 +491,3 @@ SSUMKERNEL = ../arm/sum.c
 DSUMKERNEL = ../arm/sum.c
 
 SOMATCOPY_RT = omatcopy_rt.c
-DOMATCOPY_RT = omatcopy_rt.c

kernel/x86_64/sgemm_direct_skylakex.c

@@ -1,10 +1,10 @@
 /* the direct sgemm code written by Arjan van der Ven */
-
+#include "common.h"
 
 #if defined(SKYLAKEX) || defined (COOPERLAKE)
 
 #include <immintrin.h>
-#include "common.h"
+
 
 /*
  * "Direct sgemm" code. This code operates directly on the inputs and outputs
@@ -472,7 +472,7 @@ void CNAME (BLASLONG M, BLASLONG N, BLASLONG K, float * __restrict A, BLASLONG s
 	}
 }
 #else
-#include "common.h"
+
 void CNAME (BLASLONG M, BLASLONG N, BLASLONG K, float * __restrict A, BLASLONG strideA, float * __restrict B, BLASLONG strideB , float * __restrict R, BLASLONG strideR)
 {}
 #endif

param.h

@@ -2466,13 +2466,13 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
 #define SGEMM_DEFAULT_P 512
 #define DGEMM_DEFAULT_P 384
-#define CGEMM_DEFAULT_P  512
+#define CGEMM_DEFAULT_P 512
 #define ZGEMM_DEFAULT_P 256
 
 #define SGEMM_DEFAULT_Q 512
 #define DGEMM_DEFAULT_Q 512
-#define CGEMM_DEFAULT_Q  1026
-#define ZGEMM_DEFAULT_Q 1026
+#define CGEMM_DEFAULT_Q 384
+#define ZGEMM_DEFAULT_Q 384 
 
 #define SGEMM_DEFAULT_R 4096
 #define DGEMM_DEFAULT_R 4096