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>
diff --git a/src/cpu/operators/CpuGemmLowpMatrixMultiplyCore.cpp b/src/cpu/operators/CpuGemmLowpMatrixMultiplyCore.cpp
index b25505a..94e86c6 100644
--- a/src/cpu/operators/CpuGemmLowpMatrixMultiplyCore.cpp
+++ b/src/cpu/operators/CpuGemmLowpMatrixMultiplyCore.cpp
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2021-2023 Arm Limited.
+ * Copyright (c) 2021-2024 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -65,6 +65,7 @@
asm_info.activation_info = info.activation_info();
asm_info.output_stage = info.gemmlowp_output_stage();
asm_info.fast_mode = info.fast_math();
+ asm_info.accumulate = info.accumulate();
return asm_info;
}
@@ -343,6 +344,13 @@
ARM_COMPUTE_RETURN_ERROR_ON_MSG(gemm_info.is_a_reshaped(), "Matrix A already reshaped is not supported");
ARM_COMPUTE_RETURN_ERROR_ON_MSG(gemm_info.is_b_reshaped(), "Matrix B already reshaped is not supported");
+ // When using accumulation(in place summation), for now, the only supported DataType for output is S32.
+ if (gemm_info.accumulate())
+ {
+ ARM_COMPUTE_RETURN_ERROR_ON_MSG(gemm_info.gemmlowp_output_stage().type != GEMMLowpOutputStageType::NONE,
+ "Accumulation is not supported for output QASYMM8/QASYMM8_SIGNED");
+ }
+
GEMMInfo info = gemm_info;
const ITensorInfo *matrix_a_info = a;
const ITensorInfo *matrix_b_info = b;