From ae0fc8612dba6faebf58c3ebbfae8d6e639d432d Mon Sep 17 00:00:00 2001
From: Georgios Pinitas <georgios.pinitas@arm.com>
Date: Mon, 30 Sep 2019 12:39:40 +0100
Subject: COMPMID-2452: Dot product optimizations on merge/transforms

-Adds optimized gemm transforms for AArch64.
-Optimized gemm merger to only support alpha==1 and (beta==0 || beta=1) cases

Change-Id: I55793b12a0381f4fd53f521d0e57416809904d96
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Reviewed-on: https://review.mlplatform.org/c/2003
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
---
 src/core/NEON/kernels/arm_gemm/gemm_uint8.cpp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'src/core/NEON/kernels/arm_gemm/gemm_uint8.cpp')

diff --git a/src/core/NEON/kernels/arm_gemm/gemm_uint8.cpp b/src/core/NEON/kernels/arm_gemm/gemm_uint8.cpp
index 84a45f700e..f5a30720d2 100644
--- a/src/core/NEON/kernels/arm_gemm/gemm_uint8.cpp
+++ b/src/core/NEON/kernels/arm_gemm/gemm_uint8.cpp
@@ -98,7 +98,7 @@ static const GemmImplementation<uint8_t, uint32_t> gemm_u8_methods[] = {
 {
     GemmMethod::GEMM_INTERLEAVED,
     "gemm_u8_12x8",
-    [](const GemmArgs<uint32_t> &args) { return args._ci->has_dotprod(); },
+    [](const GemmArgs<uint32_t> &args) { return args._ci->has_dotprod() && args._alpha==1 && (args._beta==0 || args._beta==1); },
     nullptr,
     [](const GemmArgs<uint32_t> &args) { return new GemmInterleaved<gemm_u8_12x8, uint8_t, uint32_t>(args); }
 },
-- 
cgit v1.2.1