diff options
author | Pablo Marquez Tello <pablo.tello@arm.com> | 2023-10-16 14:14:14 +0100 |
---|---|---|
committer | Pablo Marquez Tello <pablo.tello@arm.com> | 2023-10-30 14:30:21 +0000 |
commit | e5362e7e5dbccf81c5296a7e77154e11e1a14d2f (patch) | |
tree | f1a00380b62d62da8ecd5b839652275b9b14d9ac /src/cpu/kernels/directconv2d/nchw/fp16.cpp | |
parent | dfcd41af4eea67f7c348c8688e05cb12e3eee5f5 (diff) | |
download | ComputeLibrary-e5362e7e5dbccf81c5296a7e77154e11e1a14d2f.tar.gz |
DirectConv and Im2Col changes to enable fp16 in armv8a multi_isa builds
* FP16 kernels must be instantiated in fp16.cpp.
* Partially resolves MLCE-1102
Change-Id: Iab9c29dbfd89358f2f663862ff5010c88aeccf8c
Signed-off-by: Pablo Marquez Tello <pablo.tello@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/10496
Reviewed-by: Anitha Raj <Anitha.Raj@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Benchmark: Arm Jenkins <bsgcomp@arm.com>
Diffstat (limited to 'src/cpu/kernels/directconv2d/nchw/fp16.cpp')
-rw-r--r-- | src/cpu/kernels/directconv2d/nchw/fp16.cpp | 84 |
1 files changed, 84 insertions, 0 deletions
diff --git a/src/cpu/kernels/directconv2d/nchw/fp16.cpp b/src/cpu/kernels/directconv2d/nchw/fp16.cpp new file mode 100644 index 0000000000..a9cab42f56 --- /dev/null +++ b/src/cpu/kernels/directconv2d/nchw/fp16.cpp @@ -0,0 +1,84 @@ +/* + * Copyright (c) 2022-2023 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. + */ + +#include "src/cpu/kernels/directconv2d/impl.h" +#include "src/cpu/kernels/directconv2d/nchw/impl.h" + +namespace arm_compute +{ +namespace cpu +{ +namespace kernels +{ +#if defined(__ARM_FEATURE_FP16_VECTOR_ARITHMETIC) && defined(ENABLE_FP16_KERNELS) +void neon_fp16_nchw_directconv2d( + const Window &window, const ITensor *src, const ITensor *weights, ITensor *dst, const PadStrideInfo &conv_info) +{ + convolve_nchw<float16_t>(window, src, weights, dst, conv_info); +} +#endif //defined(__ARM_FEATURE_FP16_VECTOR_ARITHMETIC) && defined(ENABLE_FP16_KERNELS) + +void run_im2col_fp16_nchw_pad(const ITensor *src, + ITensor *dst, + const Window &window, + DataLayout data_layout, + const PadStrideInfo &conv_info, + std::pair<unsigned int, unsigned int> convolved_dims, + const Size2D &kernel_dims, + const Size2D &dilation, + uint32_t input_pad_right, + bool has_bias) +{ +#if defined(__ARM_FEATURE_FP16_VECTOR_ARITHMETIC) && defined(ENABLE_FP16_KERNELS) + arm_compute::cpu::kernels::run_im2col<float16_t, true, true>( + src, dst, window, data_layout, conv_info, convolved_dims, kernel_dims, dilation, input_pad_right, has_bias); +#else // defined(__ARM_FEATURE_FP16_VECTOR_ARITHMETIC) && defined(ENABLE_FP16_KERNELS) + ARM_COMPUTE_UNUSED(src, dst, window, data_layout, conv_info, convolved_dims, kernel_dims, dilation, input_pad_right, + has_bias); +#endif // defined(__ARM_FEATURE_FP16_VECTOR_ARITHMETIC) && defined(ENABLE_FP16_KERNELS) +} + +void run_im2col_fp16_nchw_nopad(const ITensor *src, + ITensor *dst, + const Window &window, + DataLayout data_layout, + const PadStrideInfo &conv_info, + std::pair<unsigned int, unsigned int> convolved_dims, + const Size2D &kernel_dims, + const Size2D &dilation, + uint32_t input_pad_right, + bool has_bias) +{ +#if defined(__ARM_FEATURE_FP16_VECTOR_ARITHMETIC) && defined(ENABLE_FP16_KERNELS) + arm_compute::cpu::kernels::run_im2col<float16_t, false, true>( + src, dst, window, data_layout, conv_info, convolved_dims, kernel_dims, dilation, input_pad_right, has_bias); +#else // defined(__ARM_FEATURE_FP16_VECTOR_ARITHMETIC) && defined(ENABLE_FP16_KERNELS) + ARM_COMPUTE_UNUSED(src, dst, window, data_layout, conv_info, convolved_dims, kernel_dims, dilation, input_pad_right, + has_bias); +#endif // defined(__ARM_FEATURE_FP16_VECTOR_ARITHMETIC) && defined(ENABLE_FP16_KERNELS) +} + +} // namespace kernels +} // namespace cpu +} // namespace arm_compute |