aboutsummaryrefslogtreecommitdiff
path: root/src
AgeCommit message (Expand)Author
2021-09-09Remove padding from ClGemmMatrixMultiplyReshapedOnlyRhsKernelGiorgio Arena
2021-09-08Avoid checking on biases' constantness if nullptrGiorgio Arena
2021-09-08Fix vload_partial macros on OpenCLGiorgio Arena
2021-09-07Remove padding from ClGemmMatrixMultiplyReshapedKernelGiorgio Arena
2021-09-07Fuse pad layer with subsequent convolution layerGunes Bayir
2021-09-07Add support for non-constant weights and biases in CpuFullyConnectedMichele Di Giorgio
2021-09-07Update cpu depthwise kernelsFreddie Liardet
2021-09-06Revert "Remove padding from ClGemmMatrixMultiplyReshapedKernel"Pablo Marquez Tello
2021-09-06OpenCL job chaining supportGiorgio Arena
2021-09-03Remove padding from ClPool2dKernel NCHWGiorgio Arena
2021-09-03Fix CLNormalizationLayer NCHW border calculationSiCongLi
2021-09-01Remove padding from ClGemmMatrixMultiplyReshapedKernelMichele Di Giorgio
2021-09-01Printing operators parameters, currently for CpuAdd operator only.Ramy Elgammal
2021-09-01Fix strict overflow warningsMichalis Spyrou
2021-08-25Move CPU/GPU files from Core/Runtime to the respective backend foldersGeorgios Pinitas
2021-08-25Sync backend only if allocated rather than supportedGiorgio Arena
2021-08-24Remove map/unmap overhead for input/output accessor when using DummyAccessorGiorgio Arena
2021-08-24Re-use auxiliary memory withing CpuWinogradConv2d operatorsGeorgios Pinitas
2021-08-23Remove padding from ClScaleKernelGiorgio Arena
2021-08-20Rename [Cl|Cpu]GemmConvolution to [Cl|Gpu]GemmConv2dGeorgios Pinitas
2021-08-19Address comments on avoiding releasing weights if used by multiple functionsGiorgio Arena
2021-08-18Enable fast_math on CpuGemmConvolutionGeorgios Pinitas
2021-08-18Update the heuristic to call direct convolution in clConv2dGian Marco Iodice
2021-08-18Retain weights in ClGemm when reconfiguring the operator with retentionGeorgios Pinitas
2021-08-13Avoid releasing weights if they are used by multiple functionsGeorgios Pinitas
2021-08-13Ensure correct transformed matrices are used in CpuGemmConvolutionGeorgios Pinitas
2021-08-12Ensure that correct transformed matrices are used in CpuFullyConnectedGeorgios Pinitas
2021-08-11Fix performance regression due to clFinish()Gian Marco Iodice
2021-08-10Fix compiler error in CLActivationLayerPablo Marquez Tello
2021-08-06Fix compiler error in GCC 7.4 + Ubuntu 16Pablo Marquez Tello
2021-08-04Remove 21.08 deprecated codeFreddie Liardet
2021-08-04Report error for unsupported non-constant weights in CpuFullyConnectedMichele Di Giorgio
2021-08-04Fix depthwise convolution assembly kernelsFreddie Liardet
2021-08-04Avoid over-allocation of temporary buffers within CpuWinogradConv2dGeorgios Pinitas
2021-08-04Implement Operator APIGeorgios Pinitas
2021-08-02Add missing limits includeFreddie Liardet
2021-08-02Benchmark and set default LWS for GEMM, Direct convolution and WinogradGiorgio Arena
2021-08-02Port CLConvolutionLayerSheri Zhang
2021-07-30Port ClFullyConnected to new APIGeorgios Pinitas
2021-07-30Reintroduce implementation of NEConvolutionLayer::get_convolution_methodMichele Di Giorgio
2021-07-30Compilation issue: neon=1 armv8.2 on Android with NDKr18beta1Gian Marco Iodice
2021-07-29Fix A55 performance constant for fp16 hybrid gemm kernelGeorgios Pinitas
2021-07-29Port NEConvolutionLayerMichalis Spyrou
2021-07-28Create custom flags for enabling fp16 supportGeorgios Pinitas
2021-07-28Reduce binary footprint of CpuConvertFullyConnectedWeightsKernelMichele Di Giorgio
2021-07-28Fix bare metal build issuesFreddie Liardet
2021-07-28Fix cpu GEMM fp16 issueFreddie Liardet
2021-07-28Reorganize the kernels into nhwc, nchw and common foldersAdnan AlSinan
2021-07-28Remove generated kernels that overlap hand-written onesGeorgios Pinitas
2021-07-27Fix memory lifetime issueGeorgios Pinitas
generated by cgit v1.2.1 (git 2.23.0) at 2024-05-23 07:01:57 +0000