#else
#define SGEMM_DEFAULT_UNROLL_M 16
-#ifdef DYNAMIC_ARCH
-#define DGEMM_DEFAULT_UNROLL_M 4
-#else
#define DGEMM_DEFAULT_UNROLL_M 16
-#endif
#define QGEMM_DEFAULT_UNROLL_M 2
#define CGEMM_DEFAULT_UNROLL_M 8
#define ZGEMM_DEFAULT_UNROLL_M 4
#define XGEMM_DEFAULT_UNROLL_M 1
#define SGEMM_DEFAULT_UNROLL_N 4
-#ifdef DYNAMIC_ARCH
-#define DGEMM_DEFAULT_UNROLL_N 8
-#else
#define DGEMM_DEFAULT_UNROLL_N 2
-#endif
#define QGEMM_DEFAULT_UNROLL_N 2
#define CGEMM_DEFAULT_UNROLL_N 2
#define ZGEMM_DEFAULT_UNROLL_N 2
#else
#define SGEMM_DEFAULT_P 448
-#ifndef DYNAMIC_ARCH
#define DGEMM_DEFAULT_P 192
-#else
-#define DGEMM_DEFAULT_P 384
-#endif
#define CGEMM_DEFAULT_P 384
#define ZGEMM_DEFAULT_P 256
#define SGEMM_DEFAULT_Q 448
-#ifndef DYNAMIC_ARCH
#define DGEMM_DEFAULT_Q 384
-#else
-#define DGEMM_DEFAULT_Q 168
-#endif
#define CGEMM_DEFAULT_Q 192
#define ZGEMM_DEFAULT_Q 128
#define SGEMM_DEFAULT_R sgemm_r
-#ifndef DYNAMIC_ARCH
#define DGEMM_DEFAULT_R 8640
-#else
-#define DGEMM_DEFAULT_R 13824
-#endif
#define CGEMM_DEFAULT_R cgemm_r
#define ZGEMM_DEFAULT_R zgemm_r
#define SYMV_P 16
-#if defined(CORTEXA57) || \
+#if defined(CORTEXA57) || defined(CORTEXX1) || \
defined(CORTEXA72) || defined(CORTEXA73) || \
- defined(FALKOR) || defined(TSV110) || defined(EMAG8180) || defined(VORTEX)
+ defined(FALKOR) || defined(TSV110) || defined(EMAG8180) || defined(VORTEX) || defined(FT2000)
#define SGEMM_DEFAULT_UNROLL_M 16
#define SGEMM_DEFAULT_UNROLL_N 4
/*FIXME: this should be using the cache size, but there is currently no easy way to
query that on ARM. So if getarch counted more than 8 cores we simply assume the host
is a big desktop or server with abundant cache rather than a phone or embedded device */
-#if NUM_CORES > 8 || defined(TSV110) || defined(EMAG8180) || defined(VORTEX)
+#if NUM_CORES > 8 || defined(TSV110) || defined(EMAG8180) || defined(VORTEX)|| defined(CORTEXX1)
#define SGEMM_DEFAULT_P 512
#define DGEMM_DEFAULT_P 256
#define CGEMM_DEFAULT_P 256
#elif defined(NEOVERSEN2)
+#undef SBGEMM_DEFAULT_UNROLL_M
+#undef SBGEMM_DEFAULT_UNROLL_N
+#define SBGEMM_DEFAULT_UNROLL_M 8
+#define SBGEMM_DEFAULT_UNROLL_N 4
+
#define SGEMM_DEFAULT_UNROLL_M 16
#define SGEMM_DEFAULT_UNROLL_N 4
#define CGEMM_DEFAULT_R 4096
#define ZGEMM_DEFAULT_R 4096
-#elif defined(ARMV8SVE) || defined(A64FX)
+#elif defined(ARMV8SVE) || defined(A64FX) || defined(ARMV9) || defined(CORTEXA510)|| defined(CORTEXA710) || defined(CORTEXX2)
/* When all BLAS3 routines are implemeted with SVE, SGEMM_DEFAULT_UNROLL_M should be "sve_vl".
Until then, just keep it different than DGEMM_DEFAULT_UNROLL_N to keep copy routines in both directions seperated. */