/******************************************************************************* * Copyright (c) 2019 Konduit K.K. * This program and the accompanying materials are made available under the * terms of the Apache License, Version 2.0 which is available at * https://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. * * SPDX-License-Identifier: Apache-2.0 ******************************************************************************/ // Created by Abdelrauf (rauf@konduit.ai) 2020 #include #include #include #include #include "armcomputeUtils.h" namespace sd { namespace ops { namespace platforms { ////////////////////////////////////////////////////////////////////// PLATFORM_IMPL(deconv2d, ENGINE_CPU) { auto input = INPUT_VARIABLE(0); // [bS, iH, iW, iC] (NHWC) or [bS, iC, iH, iW] (NCHW) auto weights = INPUT_VARIABLE(1); // [kH, kW, oC, iC], [iC, oC, kH, kW], [iC, kH, kW, oC] auto bias = block.width() > 2 ? INPUT_VARIABLE(2) : nullptr; // [oC] auto output = OUTPUT_VARIABLE(0); // [bS, oH, oW, oC] (NHWC) or [bS, oC, oH, oW] (NCHW) REQUIRE_TRUE(input->rankOf() == 4, 0, "CUSTOM DECONV2D ARMCOMPUTE OP: rank of input array must be equal to 4, but got %i instead !", input->rankOf()); REQUIRE_TRUE(weights->rankOf() == 4, 0, "CUSTOM DECONV2D ARMCOMPUTE OP: rank of weights array must be equal to 4, but got %i instead !", weights->rankOf()); int kH = INT_ARG(0) > 0 ? INT_ARG(0) : static_cast(weights->sizeAt(0));// filter(kernel) height int kW = INT_ARG(1) > 0 ? INT_ARG(1) : static_cast(weights->sizeAt(1));// filter(kernel) width int sH = INT_ARG(2); // strides height int sW = INT_ARG(3); // strides width int pH = INT_ARG(4); // paddings height int pW = INT_ARG(5); // paddings width int dH = INT_ARG(6); // dilations height int dW = INT_ARG(7); // dilations width int paddingMode = INT_ARG(8); // 0-VALID, 1-SAME bool isNCHW = block.getIArguments()->size() > 9 ? !INT_ARG(9) : 1; // INT_ARG(9): 0-NCHW, 1-NHWC int wFormat = block.getIArguments()->size() > 10 ? INT_ARG(10) : 0; // 0 - [kH, kW, iC, oC], 1 - [oC, iC, kH, kW], 2 - [oC, kH, kW, iC] // Calculate individual paddings unsigned int padLeft, padTop, padRight, padBottom; int bS, iC, iH, iW, oC, oH, oW; // batch size, input channels, input height/width, output channels, output height/width; int indIOioC, indIiH, indWoC, indWiC, indWkH, indOoH; // corresponding indexes ConvolutionUtils::getSizesAndIndexesConv2d(isNCHW, wFormat, *input, *output, bS, iC, iH, iW, oC, oH, oW, indIOioC, indIiH, indWoC, indWiC, indWkH, indOoH); std::vector expectedWeightsShape = ConvolutionUtils::expectWeightsShape(wFormat, kH, kW, oC, iC); REQUIRE_TRUE(weights->isSameShape(expectedWeightsShape), 0, "CUSTOM DECONV2D ARMCOMPUTE OP: wrong shape of weights array, expected is %s, but got %s instead !", ShapeUtils::shapeAsString(expectedWeightsShape).c_str(), ShapeUtils::shapeAsString(weights).c_str()); if (bias) REQUIRE_TRUE(bias->rankOf() <= 2 && oC == bias->lengthOf(), 0, "CUSTOM DECONV2D ARMCOMPUTE OP: wrong shape of array with biases, expected rank, length: <=2, %i, but got %i, %i instead !", oC, bias->rankOf(), bias->lengthOf()); if(paddingMode){ //Note: we're intentionally swapping iH and oH, to calculated the padding for a"normal" conv (not deconv) forward pass ConvolutionUtils::calcPadding2D(pH, pW, iH, iW, oH, oW, kH, kW, sH, sW, dH, dW); } padLeft = pW; padTop = pH; padRight = (iW - 1) * sW - oW + kW - pW; padBottom = (iH - 1) * sH - oH + kH - pH; //deconv2dMKLDNN(input, weights, bias, output, kH, kW, sH, sW, pH, pW, dH, dW, paddingMode, isNCHW, wFormat); #if 0 nd4j_printf("deconv2d bS = %d, iH =%d, iW = %d, oH=%d, oW=%d kH=%d, kW=%d wformat=%d, iC =%d, , oC=%d\n", bS, iH, iW, oH, oW, kH, kW, wFormat, iC, oC ); nd4j_printf("deconv2d kH = %d, kW = %d, sH = %d, sW = %d , pH = %d , pW = %d, dH = %d, dW = %d, paddingMode = %d , isNCHW %d \n" , kH , kW , sH , sW , pH , pW , dH , dW , paddingMode,isNCHW?1:0 ); #endif auto dataLayout = isNCHW ? arm_compute::DataLayout::NCHW : arm_compute::DataLayout::NHWC; //check weight input datalayout match bool dataLayoutMatch = (isNCHW && wFormat == 1) || (!isNCHW && wFormat == 2); arm_compute::PermutationVector permuteVector; //unlike in cov2d for weights iC and oC permutted : for example {oC, iC, kH, kW}, {iC, oC, kH, kW} //but we need it normal way for arm if (!dataLayoutMatch) { //lets premute if (wFormat == 0) { if (isNCHW) { #if 0 nd4j_printf("perm choise %d\n", 0); #endif //reshape permuteVector = arm_compute::PermutationVector(2U, 3U, 0U, 1U); } else { #if 0 nd4j_printf("perm choise %d\n", 1); #endif //reshape permuteVector = arm_compute::PermutationVector(0U, 2U, 3U, 1U); } } else if (wFormat == 1) { #if 0 nd4j_printf("perm choise %d\n", 2); #endif permuteVector = arm_compute::PermutationVector(3U, 0U, 1U, 2U); } else { #if 0 nd4j_printf("perm choise %d\n", 3); #endif permuteVector = arm_compute::PermutationVector(1U, 2U, 3U, 0U); } } else { //fix weight if(isNCHW){ #if 0 nd4j_printf("perm choise %d\n", 4); #endif permuteVector = arm_compute::PermutationVector(0U, 1U, 3U, 2U); }else{ #if 0 nd4j_printf("perm choise %d\n", 5); #endif permuteVector = arm_compute::PermutationVector(3U, 1U, 2U, 0U); } } Arm_WeightsInfo wInfo(false, kW, kH, 1); arm_compute::PadStrideInfo pad(sW, sH, padLeft,padRight, padTop, padBottom, arm_compute::DimensionRoundingType::FLOOR); ArmFunctionWeighted deconv; deconv.configure( input, weights, bias, output, dataLayout, permuteVector, pad); deconv.run(); // run function return Status::OK(); } PLATFORM_CHECK(deconv2d, ENGINE_CPU) { auto input = INPUT_VARIABLE(0); auto weights = INPUT_VARIABLE(1); auto output = OUTPUT_VARIABLE(0); int dH = INT_ARG(6); int dW = INT_ARG(7); // Data types supported: QASYMM8/QASYMM8_SIGNED/F16/F32. auto dTypeInput = getArmType(input->dataType()); auto dTypeWeight = getArmType(weights->dataType()); auto dTypeOutput = getArmType(output->dataType()); bool isSupported = dW==1 && dH==1 && isArmcomputeFriendly(*input) && isArmcomputeFriendly(*weights) && isArmcomputeFriendly(*output) && (dTypeInput == Arm_DataType::F32 /*|| dTypeInput == Arm_DataType::F16*/) && (dTypeWeight == dTypeInput) && (dTypeOutput == dTypeInput); #if 0 nd4j_printf("deconv2d isSupported %d : isArmcomputeFriendly(*input) = %d , isArmcomputeFriendly(*weights) = %d, isArmcomputeFriendly(*output) %d\n", isSupported, isArmcomputeFriendly(*input),isArmcomputeFriendly(*weights),isArmcomputeFriendly(*output)); #endif return isSupported; } } } }