Update GEMM assembly kernels
- Introduce Fp32 kernels with internal calculations in Bfloat16 when
fast_mode is enabled
- Improve kernel selection heuristics
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Change-Id: I68a9e7e862b6fd2721b46e0d7cc791091c4ab279
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5965
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
diff --git a/src/core/NEON/kernels/arm_gemm/gemv_batched.hpp b/src/core/NEON/kernels/arm_gemm/gemv_batched.hpp
index 1221600..4fc9b34 100644
--- a/src/core/NEON/kernels/arm_gemm/gemv_batched.hpp
+++ b/src/core/NEON/kernels/arm_gemm/gemv_batched.hpp
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2017-2020 Arm Limited.
+ * Copyright (c) 2017-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -95,6 +95,18 @@
void set_pretransposed_B_data(void *buffer) override {
_subgemm->set_pretransposed_B_data(buffer);
}
+
+ GemmConfig get_config() override {
+ GemmConfig c = _subgemm->get_config();
+
+ std::string new_filter = "gemv_batched[";
+ new_filter.append(c.filter);
+ new_filter.append("]");
+
+ c.filter = new_filter;
+
+ return c;
+ }
};
} // namespace arm_gemm