Skip to content

Commit f4ee3ae

Browse files
authored
Allow VortexM4 on the SME fast path only with non-gcc compilers
1 parent e01b109 commit f4ee3ae

1 file changed

Lines changed: 6 additions & 2 deletions

File tree

interface/trsm.c

Lines changed: 6 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -356,10 +356,14 @@ void CNAME(enum CBLAS_ORDER order,
356356
}
357357

358358
#if !defined(COMPLEX) && !defined(DOUBLE) && !defined(BFLOAT16) && !defined(HFLOAT16)
359-
#if defined(ARCH_ARM64) && (defined(USE_STRMM_KERNEL_DIRECT)||defined(DYNAMIC_ARCH))
359+
#if defined(ARCH_ARM64) && (defined(USE_STRMM_KERNEL_DIRECT)||defined(DYNAMIC_AR
360+
CH))
360361
#if defined(DYNAMIC_ARCH)
361-
if (support_sme1())
362+
if (strcmp(gotoblas_corename(), "armv9sme") == 0
363+
#if defined(__clang__)
364+
|| strcmp(gotoblas_corename(), "vortexm4") == 0
362365
#endif
366+
)
363367
if (args.m == 0 || args.n == 0) return;
364368
if (order == CblasRowMajor && Diag == CblasNonUnit && Side == CblasLeft && m == lda && n == ldb) {
365369
if (Trans == CblasNoTrans) {

0 commit comments

Comments
 (0)