commit f1f1f87132690a8061801ef1a4638d637c780df7
author Radu Salavat <radu.salavat@arm.com> Tue Feb 27 18:32:26 2024 +0000
committer Radu Salavat <radu.salavat@arm.com> Thu Apr 11 08:47:50 2024 +0000
tree 8ad4c3739217b3bc6281f4e0b9a7a63fe6c3f9bb
parent 1322065a3fbd15b00dbfb0969d6b438b5ba15530

Add in place summation to CPU GEMM kernels

Instead of dispatching the sum postop for GEMM kernels to a
separate kernel + add, that requires an extra destination sized
allocation, plus 3 extra load/stores per element,
just do it in the GEMM kernel.

Resolves: ONCPUML-1442

Signed-off-by: Radu Salavat <radu.salavat@arm.com>
Co-authored-by: Milos Puzovic <milos.puzovic@arm.com>
Change-Id: I7a1f2da3300875fa1ac88b705a34390969518077
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/11298
Reviewed-by: Gunes Bayir <gunes.bayir@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Benchmark: Arm Jenkins <bsgcomp@arm.com>

src/core/NEON/kernels/arm_gemm/gemm_int8.cpp

diff --git a/src/core/NEON/kernels/arm_gemm/gemm_int8.cpp b/src/core/NEON/kernels/arm_gemm/gemm_int8.cpp
index fd20e53..0dc0d55 100644
--- a/src/core/NEON/kernels/arm_gemm/gemm_int8.cpp
+++ b/src/core/NEON/kernels/arm_gemm/gemm_int8.cpp
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2017-2020, 2022-2023 Arm Limited.
+ * Copyright (c) 2017-2020, 2022-2024 Arm Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -128,14 +128,14 @@
 {
     GemmMethod::GEMM_HYBRID,
     "a64_smallK_hybrid_s8s32_dot_8x4",
-    [](const GemmArgs &args) { return args._ci->has_dotprod() && (args._Nsize % 4 == 0) && (args._Ksize<=32) && !args._indirect_input; },
+    [](const GemmArgs &args) { return args._ci->has_dotprod() && (args._Nsize % 4 == 0) && (args._Ksize<=32) && !args._indirect_input && !args._accumulate; },
     [](const GemmArgs &args) { return !(args._ci->has_svei8mm() || args._ci->has_i8mm()); },
     [](const GemmArgs &args) { return new GemmHybrid<cls_a64_smallK_hybrid_s8s32_dot_8x4, int8_t, int32_t>(args); }
 },
 {
     GemmMethod::GEMM_HYBRID,
     "a64_smallK_hybrid_s8s32_dot_6x4",
-    [](const GemmArgs &args) { return args._ci->has_dotprod() && (args._Nsize % 4 == 0) && (args._Ksize>32) && (args._Ksize<=64) && !args._indirect_input; },
+    [](const GemmArgs &args) { return args._ci->has_dotprod() && (args._Nsize % 4 == 0) && (args._Ksize>32) && (args._Ksize<=64) && !args._indirect_input && !args._accumulate; },
     [](const GemmArgs &args) { return !(args._ci->has_svei8mm() || args._ci->has_i8mm()); },
     [](const GemmArgs &args) { return new GemmHybrid<cls_a64_smallK_hybrid_s8s32_dot_6x4, int8_t, int32_t>(args); }
 },