From d93991e290618a685b67506c78090350e6aee43f Mon Sep 17 00:00:00 2001 From: David Mansell Date: Fri, 6 Jul 2018 14:52:52 +0100 Subject: COMPMID-1380: Pre-work for SVE support. This patch makes the needed infrastructure changes to allow SVE kernels to be added later on. Change-Id: Ide5bccac2f47278e93fff3d648231aee2d5f8c2e Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/139070 Reviewed-by: Anthony Barbier Tested-by: Jenkins --- .../NEON/kernels/arm_gemm/std_transforms_fixed.hpp | 69 ++++++++++++++++++++++ 1 file changed, 69 insertions(+) create mode 100644 src/core/NEON/kernels/arm_gemm/std_transforms_fixed.hpp (limited to 'src/core/NEON/kernels/arm_gemm/std_transforms_fixed.hpp') diff --git a/src/core/NEON/kernels/arm_gemm/std_transforms_fixed.hpp b/src/core/NEON/kernels/arm_gemm/std_transforms_fixed.hpp new file mode 100644 index 0000000000..44124a7b41 --- /dev/null +++ b/src/core/NEON/kernels/arm_gemm/std_transforms_fixed.hpp @@ -0,0 +1,69 @@ +/* + * Copyright (c) 2017-2018 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 + +namespace arm_gemm { + +/* + * Define "standard" transforms for the blocked GEMMs with fixed vector + * length. + * + * This assumes that A is interleaved 'height' ways, B is interleaved + * 'width' ways and transposed, and that the merge needs to work in 'height' + * x 'width' blocks. + * + * The optional 'block' parameter is for kernels using dot-product type + * instructions like UDOT and SDOT. + */ +template +class StdTransformsFixed +{ +public: + template + void PrepareA(TOperand *out, const TIn *in, const int stride, const int y0, + const int ymax, const int k0, const int kmax, bool transposed) { + if (transposed) { + Transform(out, in, stride, y0, ymax, k0, kmax); + } else { + Transform(out, in, stride, y0, ymax, k0, kmax); + } + } + + template + void PrepareB(TOperand *out, const TIn *in, const int stride, const int x0, + const int xmax, const int k0, const int kmax, bool transposed) { + if (transposed) { + Transform(out, in, stride, x0, xmax, k0, kmax); + } else { + Transform(out, in, stride, x0, xmax, k0, kmax); + } + } + + template + void Merge(TOut *out, const TResult *in, int stride, int y0, int ymax, int x0, int xmax, const TOut alpha, const TOut beta) { + MergeResults(out, in, stride, y0, ymax, x0, xmax, alpha, beta); + } +}; + +} // namespace arm_gemm -- cgit v1.2.1