From 2b9fa593a0a172bf36a02b5cdb840c6b9b361d7c Mon Sep 17 00:00:00 2001 From: Gunes Bayir Date: Wed, 17 Jan 2024 16:07:03 +0000 Subject: Use the stable CKW API in the GPU dynamic fusion backend - Refactor all kernels to work with the CKW stable API - Add support for sub-tile in the op_load/op_store CKW operator - Fix mismatch in resize - Add comments in all kernels written with CKW to help developers understand the structure of the code - Add texture image support in depthwise convolution written with CKW - Add support for different block sizes in depthwise convolution - Remove the use of the dynamic fusion helper functions. - Add support for floor in the op_unary() of CKW Resolves: COMPMID-6708, COMPMID-6743, COMPMID-6530 Signed-off-by: Gian Marco Iodice Signed-off-by: Gunes Bayir Signed-off-by: Viet-Hoa Do Signed-off-by: Jakub Sujak Change-Id: I8104ce4d04a3138a1aeb0b84940e1f1c89e76069 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/10914 Tested-by: Arm Jenkins Reviewed-by: Jakub Sujak Reviewed-by: Gunes Bayir Comments-Addressed: Arm Jenkins Benchmark: Arm Jenkins --- .../gpu/ckw_driver/components/utils/CkwHelper.cpp | 56 ++++++ .../gpu/ckw_driver/components/utils/CkwHelper.h | 65 +++++++ .../gpu/ckw_driver/components/utils/WriterHelper.h | 181 ------------------- .../components/utils/type_converter/Common.cpp | 162 +++++++++++++++++ .../components/utils/type_converter/Common.h | 194 +++++++-------------- .../utils/type_converter/ElementwiseBinary.cpp | 57 ++++++ .../utils/type_converter/ElementwiseBinary.h | 34 +--- 7 files changed, 410 insertions(+), 339 deletions(-) create mode 100644 src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/CkwHelper.cpp create mode 100644 src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/CkwHelper.h delete mode 100644 src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/WriterHelper.h create mode 100644 src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/type_converter/Common.cpp create mode 100644 src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/type_converter/ElementwiseBinary.cpp (limited to 'src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils') diff --git a/src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/CkwHelper.cpp b/src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/CkwHelper.cpp new file mode 100644 index 0000000000..1e6f0841ad --- /dev/null +++ b/src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/CkwHelper.cpp @@ -0,0 +1,56 @@ +/* + * Copyright (c) 2023-2024 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 "CkwHelper.h" + +namespace arm_compute +{ +namespace experimental +{ +namespace dynamic_fusion +{ +void get_coordinate_from_gws(GpuCkwScopedKernelWriter writer, + ckw::TileOperand &coord, + const ckw::TileOperand &gid, + ckw::TileOperand &step) +{ + writer->op_binary(coord, ckw::BinaryOp::Mul, gid, step); +} + +void get_coordinate_from_gws_overlapping_min(GpuCkwScopedKernelWriter writer, + ckw::TileOperand &coord, + const ckw::TileOperand &gid, + ckw::TileOperand &step, + ckw::TileOperand &shift_back, + ckw::TileOperand &const_0) +{ + // Applied formula: max((gid * step) - shift_back, 0) + // where the shift_back operand is: (step - leftover_step) % step + + writer->op_binary(coord, ckw::BinaryOp::Mul, gid, step); + writer->op_binary(coord, ckw::BinaryOp::Sub, coord, shift_back); + writer->op_binary(coord, ckw::BinaryOp::Max, coord, const_0); +} +} // namespace dynamic_fusion +} // namespace experimental +} // namespace arm_compute diff --git a/src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/CkwHelper.h b/src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/CkwHelper.h new file mode 100644 index 0000000000..956e7c8ecb --- /dev/null +++ b/src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/CkwHelper.h @@ -0,0 +1,65 @@ +/* + * Copyright (c) 2023-2024 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. + */ +#ifndef ACL_SRC_DYNAMIC_FUSION_SKETCH_GPU_CKW_DRIVER_COMPONENTS_UTILS_CKWHELPER_H +#define ACL_SRC_DYNAMIC_FUSION_SKETCH_GPU_CKW_DRIVER_COMPONENTS_UTILS_CKWHELPER_H + +#include "src/dynamic_fusion/sketch/gpu/ckw_driver/GpuCkwScopedKernelWriter.h" + +namespace arm_compute +{ +namespace experimental +{ +namespace dynamic_fusion +{ +/** Get coordinate along one axis. + * + * @param[in,out] writer Writer + * @param[out] coord Resultant coordinate + * @param[in] gid Global work item id + * @param[in] step Step size / vector size + */ +void get_coordinate_from_gws(GpuCkwScopedKernelWriter writer, + ckw::TileOperand &coord, + const ckw::TileOperand &gid, + ckw::TileOperand &step); + +/** Get boundary aware coordinate along one axis. + * + * @param[in,out] writer Writer + * @param[out] coord Resultant coordinate + * @param[in] gid Global work item id + * @param[in] step Step size / vector size + * @param[in] shift_back It is (step - leftover_step) % step + * @param[in] const_0 Constant tile of value 0 + */ +void get_coordinate_from_gws_overlapping_min(GpuCkwScopedKernelWriter writer, + ckw::TileOperand &coord, + const ckw::TileOperand &gid, + ckw::TileOperand &step, + ckw::TileOperand &shift_back, + ckw::TileOperand &const_0); +} // namespace dynamic_fusion +} // namespace experimental +} // namespace arm_compute +#endif // ACL_SRC_DYNAMIC_FUSION_SKETCH_GPU_CKW_DRIVER_COMPONENTS_UTILS_CKWHELPER_H diff --git a/src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/WriterHelper.h b/src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/WriterHelper.h deleted file mode 100644 index 6ba2b2f651..0000000000 --- a/src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/WriterHelper.h +++ /dev/null @@ -1,181 +0,0 @@ -/* - * Copyright (c) 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. - */ -#ifndef ACL_SRC_DYNAMIC_FUSION_SKETCH_GPU_CKW_DRIVER_COMPONENTS_UTILS_WRITERHELPER_H -#define ACL_SRC_DYNAMIC_FUSION_SKETCH_GPU_CKW_DRIVER_COMPONENTS_UTILS_WRITERHELPER_H - -#include "arm_compute/core/utils/misc/Utility.h" -#include "ckw/TensorTileSampler.h" - -#include "src/dynamic_fusion/sketch/gpu/ckw_driver/GpuCkwComponentArgument.h" -#include "src/dynamic_fusion/sketch/gpu/ckw_driver/GpuCkwKernelWriter.h" -#include "src/dynamic_fusion/sketch/gpu/ckw_driver/GpuCkwScopedKernelWriter.h" - -#include -#include - -using namespace ckw; -namespace arm_compute -{ -namespace experimental -{ -namespace dynamic_fusion -{ -using SamplerCreator = std::function; - -/** Load src and dst tiles of dimension [m0, n0] only when not loaded and prepare the sampler - */ -inline void load_src_dst_tiles_and_prepare_sampler(GpuCkwScopedKernelWriter &writer, - GpuCkwComponentArgument *src, - GpuCkwComponentArgument *dst, - int32_t m0, - int32_t n0, - SamplerCreator create_sampler) -{ - if (!src->has_tile()) - { - const auto sampler = create_sampler(writer, m0, n0); - writer->op_load_once(src, sampler); - } - else - { - const auto &sampler = src->tile_sampler(); - writer->op_load_once(src, sampler); - } - - auto &src_tile = src->tile(); - const auto &sampler = src->tile_sampler(); - - // Prepare the output tile. - if (!dst->has_tile()) - { - auto &tile = writer->declare_tile("dst_tile", src_tile.tile_info()); - dst->init_virtual_tensor(tile, sampler); - } -} - -/** Get boundary aware coordinate along one axis. Load and store of size step_v at the coordinate will not be out of bound - * - * @param[in,out] writer Writer - * @param[out] coord Resultant coordinate - * @param[in] gid Global work item id - * @param[in] step_v Step size / vector size - * @param[in] leftover_step_v Leftover step size at the boundary - * @param[in] prefix Prefix to all the tiles declared within this function - * @param[in] const_0 Constant tile of value 0 - */ -inline void get_coord(GpuCkwScopedKernelWriter writer, - TileOperand &coord, - const TileOperand &gid, - int32_t step_v, - int32_t leftover_step_v, - const std::string &prefix, - const TileOperand &const_0) -{ - auto &step = writer->declare_tile(prefix + "step", step_v); - auto &leftover_step = writer->declare_tile(prefix + "leftover_step", leftover_step_v); - - // step - leftover_step - auto &step_minus_leftover = writer->declare_tile(prefix + "step_minus_leftover", ckw::DataType::Int32); - writer->op_binary_expression(step_minus_leftover, step, ckw::BinaryOp::Sub, leftover_step); - - // (step - leftover_step) % step - auto &coord_correction = writer->declare_tile(prefix + "coord_correction", ckw::DataType::Int32); - writer->op_binary_expression(coord_correction, step_minus_leftover, ckw::BinaryOp::Mod, step); - - // (gid * step) - auto &raw_coord = writer->declare_tile(prefix + "raw_coord", ckw::DataType::Int32); - writer->op_binary_expression(raw_coord, gid, ckw::BinaryOp::Mul, step); - - // (gid * step) - (step - leftover_step) % step - auto &corrected_coord = writer->declare_tile(prefix + "corrected_coord", ckw::DataType::Int32); - writer->op_binary_expression(corrected_coord, raw_coord, ckw::BinaryOp::Sub, coord_correction); - - // max((gid * step) - (step - leftover_step) % step, 0) - writer->op_binary_elementwise_function(coord, ckw::BinaryFunction::Max, corrected_coord, const_0); -} - -/** Declare coordinate tiles "{prefix}_dim0_coord" and "{prefix}_dim1_coord", and create a boundary-aware sampler from tile of size [n0, m0], against the overall dimensions [dim0, dim1] - * The load and store of tile [n0, m0] will never be out of bound of [dim0, dim1] - */ - -/** Declare coordinate tiles "{prefix}_dim0_coord" and "{prefix}_dim1_coord", and create a boundary-aware sampler from tile of size [n0, m0], against the overall dimensions [dim0, dim1] - * The load and store of tile [n0, m0] will never be out of bound of [dim0, dim1] - * - * @param[in,out] writer Writer - * @param[in] gid_0 Global work item id 0 - * @param[in] gid_1 Global work item id 1 - * @param[in] dim0_v Dimension 0 - * @param[in] dim1_v Dimension 1 - * @param[in] n0_v Tile size dimension 0 - * @param[in] m0_v Tile size dimension 1 - * @param[in] prefix Prefix to all the tiles declared within this function - * @param[in] const_0 Constant tile of value 0 - * - * @return TensorTileSampler - */ -inline TensorTileSampler create_boundary_aware_2d_sampler(GpuCkwScopedKernelWriter writer, - TileOperand &gid_0, - TileOperand &gid_1, - int32_t dim0_v, - int32_t dim1_v, - int32_t n0_v, - int32_t m0_v, - const std::string prefix, - TileOperand &const_0) -{ - // Clamp tile size [n0, m0] against dimension [dim0, dim1] - // This is needed to: - // * Guard against tile sizes are bigger than the tensor dimensions - // * Handle broadcasting tiles (e.g. src tensor is of size 1 in one of the dimensions) - n0_v = utility::clamp(n0_v, 1, dim0_v); - m0_v = utility::clamp(m0_v, 1, dim1_v); - const int32_t partial_n0_v = dim0_v % n0_v; - const int32_t partial_m0_v = dim1_v % m0_v; - - // Declare #prefix_dim0_coord and #prefix_dim1_coord - auto &dim0_coord = writer->declare_tile(prefix + "dim0_coord", ckw::DataType::Int32); - get_coord(writer, dim0_coord, gid_0, n0_v, partial_n0_v, prefix + "dim0_", const_0); - auto &dim1_coord = writer->declare_tile(prefix + "dim1_coord", ckw::DataType::Int32); - get_coord(writer, dim1_coord, gid_1, m0_v, partial_m0_v, prefix + "dim1_", const_0); - - // Set sampler - // Only set fields related to boundary aware loading/storing. Other info (e.g. format) is not responsibility of this function - TensorTileSampler sampler; - - sampler.x(dim0_coord); - sampler.y(dim1_coord); - - sampler.width(n0_v); - sampler.height(m0_v); - - sampler.address_mode_x(TensorSamplerAddressModeX::None); - sampler.address_mode_y(TensorSamplerAddressModeY::None); - sampler.address_mode_z(TensorSamplerAddressModeZ::None); - - return sampler; -} -} // namespace dynamic_fusion -} // namespace experimental -} // namespace arm_compute -#endif // ACL_SRC_DYNAMIC_FUSION_SKETCH_GPU_CKW_DRIVER_COMPONENTS_UTILS_WRITERHELPER_H diff --git a/src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/type_converter/Common.cpp b/src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/type_converter/Common.cpp new file mode 100644 index 0000000000..ad31b06362 --- /dev/null +++ b/src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/type_converter/Common.cpp @@ -0,0 +1,162 @@ +/* + * Copyright (c) 2023-2024 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 "Common.h" + +namespace arm_compute +{ +namespace experimental +{ +namespace dynamic_fusion +{ +ckw::DataType to_ckw(DataType dt) +{ + switch (dt) + { + case DataType::F32: + return ckw::DataType::Fp32; + case DataType::F16: + return ckw::DataType::Fp16; + case DataType::S32: + return ckw::DataType::Int32; + case DataType::S16: + return ckw::DataType::Int16; + case DataType::S8: + case DataType::QASYMM8_SIGNED: + return ckw::DataType::Int8; + case DataType::U32: + return ckw::DataType::Uint32; + case DataType::U16: + return ckw::DataType::Uint16; + case DataType::U8: + case DataType::QASYMM8: + return ckw::DataType::Uint8; + default: + return ckw::DataType::Unknown; + } +} + +ckw::TensorShape to_ckw(const TensorShape &shape) +{ + ARM_COMPUTE_ERROR_ON(shape.num_max_dimensions < std::tuple_size{}); + ARM_COMPUTE_ERROR_ON(std::tuple_size{} != 5); + /// NOTE: Overflow danger. Use size_t? + return ckw::TensorShape{static_cast(shape[0]), static_cast(shape[1]), + static_cast(shape[2]), static_cast(shape[3]), + static_cast(shape[4])}; +} + +ckw::TensorDataLayout to_ckw(DataLayout dl) +{ + switch (dl) + { + case DataLayout::NHWC: + return ckw::TensorDataLayout::Nhwc; + case DataLayout::NDHWC: + return ckw::TensorDataLayout::Ndhwc; + default: + return ckw::TensorDataLayout::Unknown; + } +} + +ckw::TensorInfo to_ckw(const ITensorInfo &tensor_info) +{ + return ckw::TensorInfo{to_ckw(tensor_info.data_type()), to_ckw(tensor_info.tensor_shape()), + to_ckw(tensor_info.data_layout()), tensor_info.id()}; +} + +ckw::TensorStorageType to_ckw(const TensorStorageType &storage) +{ + switch (storage) + { + case TensorStorageType::ClBufferUint8Ptr: + return ckw::TensorStorageType::BufferUint8Ptr; + case TensorStorageType::ClImage2dReadOnly: + return ckw::TensorStorageType::Texture2dReadOnly; + case TensorStorageType::ClImage2dWriteOnly: + return ckw::TensorStorageType::Texture2dWriteOnly; + case TensorStorageType::Unknown: + return ckw::TensorStorageType::Unknown; + default: + ARM_COMPUTE_ERROR("Unknown tensor storage type"); + } +} + +TensorComponentType from_ckw(const ckw::TensorComponentType &component) +{ + switch (component) + { + case ckw::TensorComponentType::OffsetFirstElement: + return TensorComponentType::OffsetFirstElement; + case ckw::TensorComponentType::Stride0: + return TensorComponentType::Stride0; + case ckw::TensorComponentType::Stride1: + return TensorComponentType::Stride1; + case ckw::TensorComponentType::Stride2: + return TensorComponentType::Stride2; + case ckw::TensorComponentType::Stride3: + return TensorComponentType::Stride3; + case ckw::TensorComponentType::Stride4: + return TensorComponentType::Stride4; + case ckw::TensorComponentType::Dim0: + return TensorComponentType::Dim0; + case ckw::TensorComponentType::Dim1: + return TensorComponentType::Dim1; + case ckw::TensorComponentType::Dim2: + return TensorComponentType::Dim2; + case ckw::TensorComponentType::Dim3: + return TensorComponentType::Dim3; + case ckw::TensorComponentType::Dim4: + return TensorComponentType::Dim4; + case ckw::TensorComponentType::Dim1xDim2: + return TensorComponentType::Dim1xDim2; + case ckw::TensorComponentType::Dim2xDim3: + return TensorComponentType::Dim2xDim3; + case ckw::TensorComponentType::Dim1xDim2xDim3: + return TensorComponentType::Dim1xDim2xDim3; + case ckw::TensorComponentType::Unknown: + return TensorComponentType::Unknown; + default: + ARM_COMPUTE_ERROR("Unknown CKW tensor component"); + } +} + +TensorStorageType from_ckw(const ckw::TensorStorageType &storage) +{ + switch (storage) + { + case ckw::TensorStorageType::BufferUint8Ptr: + return TensorStorageType::ClBufferUint8Ptr; + case ckw::TensorStorageType::Texture2dReadOnly: + return TensorStorageType::ClImage2dReadOnly; + case ckw::TensorStorageType::Texture2dWriteOnly: + return TensorStorageType::ClImage2dWriteOnly; + case ckw::TensorStorageType::Unknown: + return TensorStorageType::Unknown; + default: + ARM_COMPUTE_ERROR("Unknown CKW tensor storage type"); + } +} +} // namespace dynamic_fusion +} // namespace experimental +} // namespace arm_compute diff --git a/src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/type_converter/Common.h b/src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/type_converter/Common.h index 5da317bf38..26740cdd04 100644 --- a/src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/type_converter/Common.h +++ b/src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/type_converter/Common.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2023 Arm Limited. + * Copyright (c) 2023-2024 Arm Limited. * * SPDX-License-Identifier: MIT * @@ -21,151 +21,83 @@ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. */ -#ifndef ACL_SRC_DYNAMIC_FUSION_SKETCH_GPU_CKW_DRIVER_COMPONENTS_UTILS_TYPE_CONVERTER_COMMON -#define ACL_SRC_DYNAMIC_FUSION_SKETCH_GPU_CKW_DRIVER_COMPONENTS_UTILS_TYPE_CONVERTER_COMMON +#ifndef ACL_SRC_DYNAMIC_FUSION_SKETCH_GPU_CKW_DRIVER_COMPONENTS_UTILS_TYPE_CONVERTER_COMMON_H +#define ACL_SRC_DYNAMIC_FUSION_SKETCH_GPU_CKW_DRIVER_COMPONENTS_UTILS_TYPE_CONVERTER_COMMON_H +#include "arm_compute/core/CoreTypes.h" #include "arm_compute/core/ITensorInfo.h" #include "arm_compute/core/TensorShape.h" -#include "arm_compute/core/Types.h" -#include "ckw/TensorInfo.h" #include "src/dynamic_fusion/sketch/gpu/GpuKernelArgument.h" +#include "compute_kernel_writer/include/ckw/TensorInfo.h" +#include "compute_kernel_writer/include/ckw/types/DataType.h" +#include "compute_kernel_writer/include/ckw/types/TensorComponentType.h" +#include "compute_kernel_writer/include/ckw/types/TensorStorageType.h" + namespace arm_compute { namespace experimental { namespace dynamic_fusion { -inline ckw::DataType to_ckw(DataType dt) -{ - switch (dt) - { - case DataType::F32: - return ckw::DataType::Fp32; - case DataType::F16: - return ckw::DataType::Fp16; - case DataType::S32: - return ckw::DataType::Int32; - case DataType::S16: - return ckw::DataType::Int16; - case DataType::S8: - case DataType::QASYMM8_SIGNED: - return ckw::DataType::Int8; - case DataType::U32: - return ckw::DataType::Uint32; - case DataType::U16: - return ckw::DataType::Uint16; - case DataType::U8: - case DataType::QASYMM8: - return ckw::DataType::Uint8; - default: - return ckw::DataType::Unknown; - } -} +/** Convert the Compute Library data type to Compute Kernel Writer data type + * + * @param[in] dt The Compute Library data type + * + * @return the Compute Kernel Writer data type (ckw::DataType) + */ +ckw::DataType to_ckw(DataType dt); -inline ckw::TensorShape to_ckw(const TensorShape &shape) -{ - ARM_COMPUTE_ERROR_ON(shape.num_max_dimensions < std::tuple_size{}); - ARM_COMPUTE_ERROR_ON(std::tuple_size{} != 5); - /// NOTE: Overflow danger. Use size_t? - return ckw::TensorShape{static_cast(shape[0]), static_cast(shape[1]), - static_cast(shape[2]), static_cast(shape[3]), - static_cast(shape[4])}; -} -inline ckw::TensorDataLayout to_ckw(DataLayout dl) -{ - switch (dl) - { - case DataLayout::NHWC: - return ckw::TensorDataLayout::Nhwc; - case DataLayout::NDHWC: - return ckw::TensorDataLayout::Ndhwc; - default: - return ckw::TensorDataLayout::Unknown; - } -} -inline ckw::TensorInfo to_ckw(const ITensorInfo &tensor_info) -{ - return ckw::TensorInfo{to_ckw(tensor_info.data_type()), to_ckw(tensor_info.tensor_shape()), - to_ckw(tensor_info.data_layout()), tensor_info.id()}; -} +/** Convert the Compute Library tensor shape to Compute Kernel Writer tensor shape + * + * @param[in] shape The Compute Library tensor shape + * + * @return the Compute Kernel Writer tensor shape (ckw::TensorShape) + */ +ckw::TensorShape to_ckw(const TensorShape &shape); -inline TensorComponentType from_ckw(const ckw::TensorComponentType &component) -{ - switch (component) - { - case ckw::TensorComponentType::OffsetFirstElement: - return TensorComponentType::OffsetFirstElement; - case ckw::TensorComponentType::Stride0: - return TensorComponentType::Stride0; - case ckw::TensorComponentType::Stride1: - return TensorComponentType::Stride1; - case ckw::TensorComponentType::Stride2: - return TensorComponentType::Stride2; - case ckw::TensorComponentType::Stride3: - return TensorComponentType::Stride3; - case ckw::TensorComponentType::Stride4: - return TensorComponentType::Stride4; - case ckw::TensorComponentType::Dim0: - return TensorComponentType::Dim0; - case ckw::TensorComponentType::Dim1: - return TensorComponentType::Dim1; - case ckw::TensorComponentType::Dim2: - return TensorComponentType::Dim2; - case ckw::TensorComponentType::Dim3: - return TensorComponentType::Dim3; - case ckw::TensorComponentType::Dim4: - return TensorComponentType::Dim4; - case ckw::TensorComponentType::Dim1xDim2: - return TensorComponentType::Dim1xDim2; - case ckw::TensorComponentType::Dim2xDim3: - return TensorComponentType::Dim2xDim3; - case ckw::TensorComponentType::Dim1xDim2xDim3: - return TensorComponentType::Dim1xDim2xDim3; - case ckw::TensorComponentType::Unknown: - return TensorComponentType::Unknown; - default: - ARM_COMPUTE_ERROR("Unknown CKW tensor component"); - return TensorComponentType::Unknown; - } -} +/** Convert the Compute Library data layout to Compute Kernel Writer data layout + * + * @param[in] dl The Compute Library data layout + * + * @return the Compute Kernel Writer data layout (ckw::TensorDataLayout) + */ +ckw::TensorDataLayout to_ckw(DataLayout dl); + +/** Convert the Compute Library tensor info to Compute Kernel Writer tensor info + * + * @param[in] tensor_info The Compute Library tensor info + * + * @return the Compute Kernel Writer tensor info (ckw::TensorInfo) + */ +ckw::TensorInfo to_ckw(const ITensorInfo &tensor_info); + +/** Convert the Compute Library tensor storage to Compute Kernel Writer tensor storage + * + * @param[in] storage The Compute Library tensor storage + * + * @return the Compute Kernel Writer tensor storate (ckw::TensorStorageType) + */ +ckw::TensorStorageType to_ckw(const TensorStorageType &storage); + +/** Convert the Compute Kernel Writer tensor component to Compute Library tensor component + * + * @param[in] component The Compute Kernel Writer tensor component + * + * @return the Compute Library tensor component + */ +TensorComponentType from_ckw(const ckw::TensorComponentType &component); + +/** Convert the Compute Kernel Writer tensor storage to Compute Library tensor storage + * + * @param[in] storage The Compute Kernel Writer tensor storage + * + * @return the Compute Library tensor storage + */ +TensorStorageType from_ckw(const ckw::TensorStorageType &storage); -inline ckw::TensorStorageType to_ckw(const TensorStorageType &storage) -{ - switch (storage) - { - case TensorStorageType::ClBufferUint8Ptr: - return ckw::TensorStorageType::BufferUint8Ptr; - case TensorStorageType::ClImage2dReadOnly: - return ckw::TensorStorageType::Texture2dReadOnly; - case TensorStorageType::ClImage2dWriteOnly: - return ckw::TensorStorageType::Texture2dWriteOnly; - case TensorStorageType::Unknown: - return ckw::TensorStorageType::Unknown; - default: - ARM_COMPUTE_ERROR("Unknown tensor storage type"); - return ckw::TensorStorageType::Unknown; - } -} -inline TensorStorageType from_ckw(const ckw::TensorStorageType &storage) -{ - switch (storage) - { - case ckw::TensorStorageType::BufferUint8Ptr: - return TensorStorageType::ClBufferUint8Ptr; - case ckw::TensorStorageType::Texture2dReadOnly: - return TensorStorageType::ClImage2dReadOnly; - case ckw::TensorStorageType::Texture2dWriteOnly: - return TensorStorageType::ClImage2dWriteOnly; - case ckw::TensorStorageType::Unknown: - return TensorStorageType::Unknown; - default: - ARM_COMPUTE_ERROR("Unknown CKW tensor storage type"); - return TensorStorageType::Unknown; - } -} } // namespace dynamic_fusion } // namespace experimental } // namespace arm_compute -#endif /* ACL_SRC_DYNAMIC_FUSION_SKETCH_GPU_CKW_DRIVER_COMPONENTS_UTILS_TYPE_CONVERTER_COMMON */ +#endif // ACL_SRC_DYNAMIC_FUSION_SKETCH_GPU_CKW_DRIVER_COMPONENTS_UTILS_TYPE_CONVERTER_COMMON_H diff --git a/src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/type_converter/ElementwiseBinary.cpp b/src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/type_converter/ElementwiseBinary.cpp new file mode 100644 index 0000000000..5630e390d5 --- /dev/null +++ b/src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/type_converter/ElementwiseBinary.cpp @@ -0,0 +1,57 @@ +/* + * Copyright (c) 2023-2024 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/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/type_converter/ElementwiseBinary.h" + +#include "src/dynamic_fusion/sketch/gpu/operators/internal/GpuElementwiseBinaryCommon.h" + +namespace arm_compute +{ +namespace experimental +{ +namespace dynamic_fusion +{ +ckw::BinaryOp to_ckw(const ElementwiseBinaryCommonAttributes &attributes) +{ + switch (attributes.operation()) + { + case ElementwiseBinaryCommonAttributes::ElementwiseOp::Add: + return ckw::BinaryOp::Add; + case ElementwiseBinaryCommonAttributes::ElementwiseOp::Sub: + return ckw::BinaryOp::Sub; + case ElementwiseBinaryCommonAttributes::ElementwiseOp::Div: + return ckw::BinaryOp::Div; + case ElementwiseBinaryCommonAttributes::ElementwiseOp::Mul: + return ckw::BinaryOp::Mul; + case ElementwiseBinaryCommonAttributes::ElementwiseOp::Min: + case ElementwiseBinaryCommonAttributes::ElementwiseOp::Max: + case ElementwiseBinaryCommonAttributes::ElementwiseOp::Power: + case ElementwiseBinaryCommonAttributes::ElementwiseOp::Prelu: + case ElementwiseBinaryCommonAttributes::ElementwiseOp::SquaredDiff: + default: + ARM_COMPUTE_ERROR("Cannot convert ElementwiseBinaryCommonAttributes to corresponding ckw::BinaryOp"); + } +} +} // namespace dynamic_fusion +} // namespace experimental +} // namespace arm_compute diff --git a/src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/type_converter/ElementwiseBinary.h b/src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/type_converter/ElementwiseBinary.h index 0cba258940..644a407702 100644 --- a/src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/type_converter/ElementwiseBinary.h +++ b/src/dynamic_fusion/sketch/gpu/ckw_driver/components/utils/type_converter/ElementwiseBinary.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2023 Arm Limited. + * Copyright (c) 2023-2024 Arm Limited. * * SPDX-License-Identifier: MIT * @@ -21,42 +21,22 @@ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. */ -#ifndef ACL_SRC_DYNAMIC_FUSION_SKETCH_GPU_CKW_DRIVER_COMPONENTS_UTILS_TYPE_CONVERTER_ELEMENTWISEBINARY -#define ACL_SRC_DYNAMIC_FUSION_SKETCH_GPU_CKW_DRIVER_COMPONENTS_UTILS_TYPE_CONVERTER_ELEMENTWISEBINARY - -#include "ckw/types/Operators.h" +#ifndef ACL_SRC_DYNAMIC_FUSION_SKETCH_GPU_CKW_DRIVER_COMPONENTS_UTILS_TYPE_CONVERTER_ELEMENTWISEBINARY_H +#define ACL_SRC_DYNAMIC_FUSION_SKETCH_GPU_CKW_DRIVER_COMPONENTS_UTILS_TYPE_CONVERTER_ELEMENTWISEBINARY_H #include "src/dynamic_fusion/sketch/gpu/operators/internal/GpuElementwiseBinaryCommon.h" +#include "compute_kernel_writer/include/ckw/types/Operators.h" + namespace arm_compute { namespace experimental { namespace dynamic_fusion { -inline ckw::BinaryOp to_ckw(const ElementwiseBinaryCommonAttributes &attributes) -{ - switch (attributes.operation()) - { - case ElementwiseBinaryCommonAttributes::ElementwiseOp::Add: - return ckw::BinaryOp::Add; - case ElementwiseBinaryCommonAttributes::ElementwiseOp::Sub: - return ckw::BinaryOp::Sub; - case ElementwiseBinaryCommonAttributes::ElementwiseOp::Div: - return ckw::BinaryOp::Div; - case ElementwiseBinaryCommonAttributes::ElementwiseOp::Mul: - return ckw::BinaryOp::Mul; - case ElementwiseBinaryCommonAttributes::ElementwiseOp::Min: - case ElementwiseBinaryCommonAttributes::ElementwiseOp::Max: - case ElementwiseBinaryCommonAttributes::ElementwiseOp::Power: - case ElementwiseBinaryCommonAttributes::ElementwiseOp::Prelu: - case ElementwiseBinaryCommonAttributes::ElementwiseOp::SquaredDiff: - default: - ARM_COMPUTE_ERROR("Cannot convert ElementwiseBinaryCommonAttributes to corresponding ckw::BinaryOp"); - } -} +ckw::BinaryOp to_ckw(const ElementwiseBinaryCommonAttributes &attributes); } // namespace dynamic_fusion } // namespace experimental } // namespace arm_compute -#endif /* ACL_SRC_DYNAMIC_FUSION_SKETCH_GPU_CKW_DRIVER_COMPONENTS_UTILS_TYPE_CONVERTER_ELEMENTWISEBINARY */ +#endif // ACL_SRC_DYNAMIC_FUSION_SKETCH_GPU_CKW_DRIVER_COMPONENTS_UTILS_TYPE_CONVERTER_ELEMENTWISEBINARY_H -- cgit v1.2.1