[CPU][s390x] Optimize inference perf and add oneDNN INT8 GEMM for s390x (#50219)

Signed-off-by: Rehan Khan <Rehan.Khan7@ibm.com>
Co-authored-by: Li, Jiang <jiang1.li@intel.com>
This commit is contained in:
Rehan Khan
2026-07-31 17:49:31 +08:00
committed by GitHub
co-authored by Li, Jiang <jiang1.li@intel.com>
parent 482cfc20f0
commit 88bc8fb56f
6 changed files with 182 additions and 27 deletions
+11
View File
@@ -183,4 +183,15 @@
#endif // __powerpc__
// IBM Z (s390x) VXE
#ifdef __s390x__
// FP32Vec16::exp() in cpu_types_vxe.hpp delegates to FP32Vec8::exp(), which
// implements a vectorised 5-term minimax polynomial using VXE intrinsics.
#define DEFINE_FAST_EXP \
auto fast_exp = [&](const vec_op::FP32Vec16& vec) \
__attribute__((always_inline)) { return vec.exp(); }; \
auto fast_exp_f16 = fast_exp;
#endif // __s390x__
#endif