aboutsummaryrefslogtreecommitdiff
path: root/src/core/NEON/kernels/arm_gemm/kernels/a64_sgemm_nativeA_pretransposeB_16x4.hpp
diff options
context:
space:
mode:
authorGeorgios Pinitas <georgios.pinitas@arm.com>2019-10-14 19:03:09 +0100
committerGeorgios Pinitas <georgios.pinitas@arm.com>2019-10-23 12:08:12 +0000
commit48b3ef89de5f21a0169d8416e3d54081f82c7bf8 (patch)
treef857d733ccf446c704823dc7ac796a96eb55095e /src/core/NEON/kernels/arm_gemm/kernels/a64_sgemm_nativeA_pretransposeB_16x4.hpp
parent1dce3101ef8d77c8cf0af7dfd4af6595a0136b91 (diff)
downloadComputeLibrary-48b3ef89de5f21a0169d8416e3d54081f82c7bf8.tar.gz
COMPMID-2577: Fuse bias addition and activation in gemm assembly kernels
Change-Id: I7f52112d2d05b1ea3d3f3d4b19b8eafab05d6c44 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/2141 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Marquez <pablo.tello@arm.com>
Diffstat (limited to 'src/core/NEON/kernels/arm_gemm/kernels/a64_sgemm_nativeA_pretransposeB_16x4.hpp')
-rw-r--r--src/core/NEON/kernels/arm_gemm/kernels/a64_sgemm_nativeA_pretransposeB_16x4.hpp78
1 files changed, 0 insertions, 78 deletions
diff --git a/src/core/NEON/kernels/arm_gemm/kernels/a64_sgemm_nativeA_pretransposeB_16x4.hpp b/src/core/NEON/kernels/arm_gemm/kernels/a64_sgemm_nativeA_pretransposeB_16x4.hpp
deleted file mode 100644
index 9b70827125..0000000000
--- a/src/core/NEON/kernels/arm_gemm/kernels/a64_sgemm_nativeA_pretransposeB_16x4.hpp
+++ /dev/null
@@ -1,78 +0,0 @@
-/*
- * Copyright (c) 2019 Arm Limited.
- *
- * SPDX-License-Identifier: MIT
- *
- * Permission is hereby granted, free of charge, to any person obtaining a copy
- * of this software and associated documentation files (the "Software"), to
- * deal in the Software without restriction, including without limitation the
- * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
- * sell copies of the Software, and to permit persons to whom the Software is
- * furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included in all
- * copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
- * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
- * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
- * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
- * SOFTWARE.
- */
-#pragma once
-
-#ifdef __aarch64__
-
-namespace arm_gemm {
-
-// Actual kernel implementations
-void a64_sgemm_nativeA_pretransposeB_16x4(const float *, int, const float *, float *, int, float, unsigned int, unsigned int, unsigned int);
-
-// Native A/Pretranspose B SGEMM "strategy" class.
-//
-// This describes the characteristics of a family of kernels, in terms of
-// the required interleave properties and the output block size.
-//
-// All kernels in the family must share these characteristics. The actual
-// kernel to be used can be chosen at runtime, based on the CPUInfo
-// structure.
-class sgemm_nativeA_pretransposeB_16x4 {
-public:
- typedef float operand_type;
- typedef float result_type;
-
- typedef void (*kern_type)(const float *, int, const float *, float *, int, float, unsigned int, unsigned int, unsigned int);
-
- /* Desired data layout for B buffer (used for pretranspose) */
- static const int B_interleave = 16;
- static const int B_block = 1;
- static const bool B_transpose = true;
-
- /* Kernel blocking parameters */
- static unsigned int out_width() {
- return 16;
- }
-
- static unsigned int out_height() {
- return 4;
- }
-
- static unsigned int k_unroll() {
- return 1;
- }
-
- StdTransformsFixed<operand_type, result_type, 4, 16> transforms = {};
-
- // Default to the generic kernel
- kern_type kernel=a64_sgemm_nativeA_pretransposeB_16x4;
-
- sgemm_nativeA_pretransposeB_16x4(const CPUInfo *ci) {
- UNUSED(ci);
- }
-};
-
-} // namespace arm_gemm
-
-#endif // __aarch64__