floraachy/OpenBLAS
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

# The first commit's message is:

Browse Source 
Optimizations for APM's xgene-1 (aarch64).

1) general system updates to support armv8 better.  Make all did not work, one needed to supply TARGET=ARMV8.
2) sgem 4x4 kernel in assembler using SIMD, and configuration changes to use it.
3) strmm 4x4 kernel in C.  Since the sgem kernel does 4x4, the trmm kernel must also do 4xN.

Added Dave Nuechterlein to the contributors list.
This commit is contained in:
Benedikt Huber
2014-10-09 06:52:10 -07:00
committed by Zhang Xianyi
parent 2987bc7b40
commit 58c90d5937
8 changed files with 2442 additions and 16 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
getarch.c
View File

@@ -746,12 +746,11 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#define SUBARCHITECTURE "ARMV8"
#define SUBDIRNAME "arm64"
#define ARCHCONFIG "-DARMV8 " \
"-DL1_DATA_SIZE=65536 -DL1_DATA_LINESIZE=32 " \
"-DL2_SIZE=512488 -DL2_LINESIZE=32 " \
"-DDTB_DEFAULT_ENTRIES=64 -DDTB_SIZE=4096 -DL2_ASSOCIATIVE=4 " \
"-DHAVE_VFP -DHAVE_VFPV3 -DHAVE_VFPV4"
"-DL1_DATA_SIZE=32768 -DL1_DATA_LINESIZE=64 " \
"-DL2_SIZE=262144 -DL2_LINESIZE=64 " \
"-DDTB_DEFAULT_ENTRIES=64 -DDTB_SIZE=4096 -DL2_ASSOCIATIVE=32 "
#define LIBNAME "armv8"
#define CORENAME "ARMV8"
#define CORENAME "XGENE1"
#else
#endif
@@ -801,6 +800,11 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
#define OPENBLAS_SUPPORTED
#endif
#ifdef __aarch64__
#include "cpuid_arm64.c"
#define OPENBLAS_SUPPORTED
#endif
#ifndef OPENBLAS_SUPPORTED
#error "This arch/CPU is not supported by OpenBLAS."
@@ -856,7 +860,7 @@ int main(int argc, char *argv[]){
#ifdef FORCE
printf("CORE=%s\n", CORENAME);
#else
#if defined(__i386__) || defined(__x86_64__) || defined(POWER) || defined(__mips__) || defined(__arm__)
#if defined(__i386__) || defined(__x86_64__) || defined(POWER) || defined(__mips__) || defined(__arm__) || defined(__aarch64__)
printf("CORE=%s\n", get_corename());
#endif
#endif
@@ -956,7 +960,7 @@ int main(int argc, char *argv[]){
#ifdef FORCE
printf("#define CHAR_CORENAME \"%s\"\n", CORENAME);
#else
#if defined(__i386__) || defined(__x86_64__) || defined(POWER) || defined(__mips__) || defined(__arm__)
#if defined(__i386__) || defined(__x86_64__) || defined(POWER) || defined(__mips__) || defined(__arm__) || defined(__aarch64__)
printf("#define CHAR_CORENAME \"%s\"\n", get_corename());
#endif
#endif