// Copyright 2015 The Gemmlowp Authors. All Rights Reserved. // // Licensed under the Apache License, Version 2.0 (the "License"); // you may not use this file except in compliance with the License. // You may obtain a copy of the License at // // http://www.apache.org/licenses/LICENSE-2.0 // // Unless required by applicable law or agreed to in writing, software // distributed under the License is distributed on an "AS IS" BASIS, // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. // See the License for the specific language governing permissions and // limitations under the License. // unpack.h: unpacking the result blocks computed by compute.h, // storing them into the destination matrix. #ifndef GEMMLOWP_INTERNAL_UNPACK_H_ #define GEMMLOWP_INTERNAL_UNPACK_H_ #include "allocator.h" #include "block_params.h" #include "output.h" #include "pack.h" #include namespace gemmlowp { class PackedResult { public: PackedResult(Allocator* _allocator, const BlockParams& _block_params) : allocator_(_allocator), block_params_(_block_params) { matrix_handle_ = allocator_->Reserve(block_params_.l2_rows * block_params_.l2_cols); } ~PackedResult() {} MatrixMap Map() { return MatrixMap( allocator_->GetPointer(matrix_handle_), block_params_.l2_rows, block_params_.l2_cols, block_params_.l2_rows); } MatrixMap Map() const { return MatrixMap( allocator_->GetPointer(matrix_handle_), block_params_.l2_rows, block_params_.l2_cols, block_params_.l2_rows); } private: Allocator* allocator_; Allocator::Handle matrix_handle_; const BlockParams& block_params_; }; struct MatrixBlockBounds { int start_row; int start_col; int rows; int cols; MatrixBlockBounds(int start_row_, int start_col_, int rows_, int cols_) : start_row(start_row_), start_col(start_col_), rows(rows_), cols(cols_) {} }; template void PrefetchResultBlock(const SrcMapType& src, const VectorMap& lhs_sums_of_each_slice, int src_row, int src_col) { const std::int32_t* src_data = src.data(src_row, src_col); const int src_stride = src.stride(); const std::int32_t* lhs_sums_data = lhs_sums_of_each_slice.data(src_row); for (int r = 0; r < Rows; r += 4) { Prefetch(lhs_sums_data + r); } for (int c = 0; c < Cols; c++) { for (int r = 0; r < Rows; r += 4) { Prefetch(src_data + r + c * src_stride); } } } template void UnpackResultBlock(const SrcMapType& src, const OutputPipelineExecutorType& executor, DstType* dst, const VectorMap& lhs_sums_of_each_slice, const VectorMap& rhs_sums_of_each_slice, const LhsOffset& lhs_offset, const RhsOffset& rhs_offset, int depth, int src_row, int src_col, int src_global_row, int src_global_col, int dst_row, int dst_col) { using KernelLhsScalar = typename KernelFormat::Lhs::Scalar; using KernelRhsScalar = typename KernelFormat::Rhs::Scalar; static constexpr int KernelLhsZeroPointInput = ZeroPointInputValue::kValue; static constexpr int KernelRhsZeroPointInput = ZeroPointInputValue::kValue; auto acc = Load(src, src_row, src_col); const auto& lhs_sums_of_each_slice_block = LoadForBroadcasting(lhs_sums_of_each_slice, src_row); const auto& rhs_sums_of_each_slice_block = LoadForBroadcasting(rhs_sums_of_each_slice, src_col); auto lhs_offset_block = LoadForBroadcasting(lhs_offset, src_row); auto rhs_offset_block = LoadForBroadcasting(rhs_offset, src_col); AddConstant(&lhs_offset_block); AddConstant(&rhs_offset_block); BroadcastMulAdd(lhs_sums_of_each_slice_block, rhs_offset_block, &acc); for (int i = 0; i < decltype(rhs_offset_block)::kRegisterCount; i++) { rhs_offset_block.buf.reg[i] = Mul(rhs_offset_block.buf.reg[i], depth); } BroadcastMulAdd(BroadcastAdd(rhs_sums_of_each_slice_block, rhs_offset_block), lhs_offset_block, &acc); executor.Execute(acc, dst, src_global_row, src_global_col, dst_row, dst_col); } template void UnpackResult(ResultBlockType* dst, const MatrixBlockBounds& dst_block, const PackedResultType& src, int depth, const std::int32_t* lhs_sums_of_each_slice_ptr, const std::int32_t* rhs_sums_of_each_slice_ptr, const LhsOffset& lhs_offset, const RhsOffset& rhs_offset, const OutputPipelineType& output_pipeline) { ScopedProfilingLabel label(ResultBlockType::kOrder == MapOrder::ColMajor ? "unpack to column-major" : "unpack to row-major"); assert(dst_block.start_row >= 0); assert(dst_block.start_row + dst_block.rows <= dst->rows()); assert(dst_block.start_col >= 0); assert(dst_block.start_col + dst_block.cols <= dst->cols()); const auto src_map = src.Map(); const VectorMap lhs_sums_of_each_slice( lhs_sums_of_each_slice_ptr, dst_block.rows); const VectorMap rhs_sums_of_each_slice( rhs_sums_of_each_slice_ptr, dst_block.cols); using Int32x1x1 = RegisterBlock; using Int32x4x1 = RegisterBlock; using Int32x8x1 = RegisterBlock; using Int32x1x4 = RegisterBlock; using Int32x4x4 = RegisterBlock; using Int32x8x4 = RegisterBlock; using DstScalarType = typename ResultBlockType::Scalar; using DstScalarx8x8 = RegisterBlock; OutputPipelineExecutor output_pipeline_executor_1x1(output_pipeline); OutputPipelineExecutor output_pipeline_executor_4x1(output_pipeline); OutputPipelineExecutor output_pipeline_executor_8x1(output_pipeline); OutputPipelineExecutor output_pipeline_executor_1x4(output_pipeline); OutputPipelineExecutor output_pipeline_executor_4x4(output_pipeline); OutputPipelineExecutor output_pipeline_executor_8x4(output_pipeline); int c8 = 0; if (ResultBlockType::kOrder == MapOrder::RowMajor) { for (; c8 <= dst_block.cols - 8; c8 += 8) { PrefetchResultBlock<8, 8>(src_map, lhs_sums_of_each_slice, 0, c8); int r = 0; for (; r <= dst_block.rows - 8; r += 8) { const int global_row = r + dst_block.start_row; PrefetchResultBlock<8, 8>(src_map, lhs_sums_of_each_slice, r + 8, c8); DstScalarType dst_colmajor_buf[64]; MatrixMap dst_colmajor_map( dst_colmajor_buf, 8, 8); for (int cx = 0; cx < 8; cx += 4) { const int c = c8 + cx; const int global_col = c + dst_block.start_col; UnpackResultBlock( src_map, output_pipeline_executor_8x4, &dst_colmajor_map, lhs_sums_of_each_slice, rhs_sums_of_each_slice, lhs_offset, rhs_offset, depth, r, c, global_row, global_col, 0, cx); } StoreFinalOutput(LoadContiguous(dst_colmajor_buf), dst, r + dst_block.start_row, c8 + dst_block.start_col); } for (; r <= dst_block.rows - 4; r += 4) { const int global_row = r + dst_block.start_row; for (int cx = 0; cx < 8; cx += 4) { const int c = c8 + cx; const int global_col = c + dst_block.start_col; UnpackResultBlock( src_map, output_pipeline_executor_4x4, dst, lhs_sums_of_each_slice, rhs_sums_of_each_slice, lhs_offset, rhs_offset, depth, r, c, global_row, global_col, global_row, global_col); } } for (; r < dst_block.rows; r++) { const int global_row = r + dst_block.start_row; for (int cx = 0; cx < 8; cx += 4) { const int c = c8 + cx; const int global_col = c + dst_block.start_col; UnpackResultBlock( src_map, output_pipeline_executor_1x4, dst, lhs_sums_of_each_slice, rhs_sums_of_each_slice, lhs_offset, rhs_offset, depth, r, c, global_row, global_col, global_row, global_col); } } } } int c = c8; for (; c <= dst_block.cols - 4; c += 4) { const int global_col = c + dst_block.start_col; PrefetchResultBlock<8, 4>(src_map, lhs_sums_of_each_slice, 0, c); int r = 0; for (; r <= dst_block.rows - 8; r += 8) { const int global_row = r + dst_block.start_row; PrefetchResultBlock<8, 4>(src_map, lhs_sums_of_each_slice, r + 8, c); UnpackResultBlock( src_map, output_pipeline_executor_8x4, dst, lhs_sums_of_each_slice, rhs_sums_of_each_slice, lhs_offset, rhs_offset, depth, r, c, global_row, global_col, global_row, global_col); } for (; r <= dst_block.rows - 4; r += 4) { const int global_row = r + dst_block.start_row; UnpackResultBlock( src_map, output_pipeline_executor_4x4, dst, lhs_sums_of_each_slice, rhs_sums_of_each_slice, lhs_offset, rhs_offset, depth, r, c, global_row, global_col, global_row, global_col); } for (; r < dst_block.rows; r++) { const int global_row = r + dst_block.start_row; UnpackResultBlock( src_map, output_pipeline_executor_1x4, dst, lhs_sums_of_each_slice, rhs_sums_of_each_slice, lhs_offset, rhs_offset, depth, r, c, global_row, global_col, global_row, global_col); } } for (; c < dst_block.cols; c++) { const int global_col = c + dst_block.start_col; PrefetchResultBlock<8, 1>(src_map, lhs_sums_of_each_slice, 0, c); int r = 0; for (; r <= dst_block.rows - 8; r += 8) { const int global_row = r + dst_block.start_row; PrefetchResultBlock<8, 1>(src_map, lhs_sums_of_each_slice, r + 8, c); UnpackResultBlock( src_map, output_pipeline_executor_8x1, dst, lhs_sums_of_each_slice, rhs_sums_of_each_slice, lhs_offset, rhs_offset, depth, r, c, global_row, global_col, global_row, global_col); } for (; r <= dst_block.rows - 4; r += 4) { const int global_row = r + dst_block.start_row; UnpackResultBlock( src_map, output_pipeline_executor_4x1, dst, lhs_sums_of_each_slice, rhs_sums_of_each_slice, lhs_offset, rhs_offset, depth, r, c, global_row, global_col, global_row, global_col); } for (; r < dst_block.rows; r++) { const int global_row = r + dst_block.start_row; UnpackResultBlock( src_map, output_pipeline_executor_1x1, dst, lhs_sums_of_each_slice, rhs_sums_of_each_slice, lhs_offset, rhs_offset, depth, r, c, global_row, global_col, global_row, global_col); } } } } // end namespace gemmlowp #endif // GEMMLOWP_INTERNAL_UNPACK_H_