From 86690f9fd302cc6e8e0b647db8d1650dc133ccfd Mon Sep 17 00:00:00 2001 From: Arnav Sharma Date: Tue, 21 Dec 2021 16:49:11 +0530 Subject: [PATCH] Optimized AXPBYV Kernel using AVX2 Intrinsics Details: - Intrinsic implementation of axpbyv for AVX2 - Bench written for axpbyv - Added definitions in zen contexts AMD-Internal: [CPUPL-1963] Change-Id: I9bc21a6170f5c944eb6e9e9f0e994b9992f8b539 --- kernels/zen/bli_kernels_zen.h | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/kernels/zen/bli_kernels_zen.h b/kernels/zen/bli_kernels_zen.h index fea59796a..c092ab3ab 100644 --- a/kernels/zen/bli_kernels_zen.h +++ b/kernels/zen/bli_kernels_zen.h @@ -57,6 +57,16 @@ AXPBYV_KER_PROT( dcomplex, z, axpbyv_zen_int ) AXPBYV_KER_PROT( float, s, axpbyv_zen_int10 ) AXPBYV_KER_PROT( double, d, axpbyv_zen_int10 ) +// axpbyv (intrinsics) +AXPBYV_KER_PROT( float, s, axpbyv_zen_int ) +AXPBYV_KER_PROT( double, d, axpbyv_zen_int ) +AXPBYV_KER_PROT( scomplex, c, axpbyv_zen_int ) +AXPBYV_KER_PROT( dcomplex, z, axpbyv_zen_int ) + +// axpbyv (intrinsics, unrolled x10) +AXPBYV_KER_PROT( float, s, axpbyv_zen_int10 ) +AXPBYV_KER_PROT( double, d, axpbyv_zen_int10 ) + // axpyv (intrinsics) AXPYV_KER_PROT( float, s, axpyv_zen_int ) AXPYV_KER_PROT( double, d, axpyv_zen_int )