- 15d0c44 Fix performance regression in Transposed Convolution by Gunes Bayir · 1 year, 5 months ago
- ae72a46 Add new operator AddMulAdd for Neon™ backend for Float/Quantized types by Gunes Bayir · 1 year, 5 months ago
- 464ed20 Remove fixed format strides hack by Jonathan Deakin · 1 year, 6 months ago
- b609c93 Skip upsampling for deconvolution when not needed by Annop Wongwathanarat · 1 year, 5 months ago
- a6a1538 Fix performance regression when stride equal to kernel size by Gunes Bayir · 1 year, 5 months ago
- 11f7d7e Add enable_fast_math for NEDeconvolutionLayer by Annop Wongwathanarat · 1 year, 6 months ago
- a68df8d Revert "Update the heuristic for CLDepthwiseConvolutionNative kernel" by Gian Marco Iodice · 1 year, 5 months ago
- bc67208 Update the heuristic for CLDepthwiseConvolutionNative kernel by Gian Marco Iodice · 1 year, 6 months ago
- 9d3bd41 Move DWC native heuristic into the heuristic folder by Gian Marco Iodice · 1 year, 6 months ago
- a5cb79f Update the ClConv2d heuristic by Gian Marco Iodice · 1 year, 6 months ago
- e317baf Optimize MeanReduce by integer acc. and removing upfront dequant. by Omar Al Khatib · 1 year, 7 months ago
- a0ae8d2 Optimize Transposed Convolution for CL backend (Quantized) by Gunes Bayir · 1 year, 7 months ago
- df6a3b0 Use heuristics for setting dynamic fusion direct conv2d tile sizes by Ramy Elgammal · 1 year, 7 months ago
- 76335eb Implement the OpenCL kernel to compute the indirect convolution by Gian Marco Iodice · 1 year, 7 months ago
- 31df05a Remove dynamic fusion prototype with tests and examples by SiCong Li · 1 year, 8 months ago
- d2475c7 Add num_threads_to_use to OMPScheduler based on workload size by cfRod · 1 year, 8 months ago
- ec0113d Optimize Transposed Convolution for CL backend (FP32/16) by Gunes Bayir · 1 year, 8 months ago
- ad9a7ed Rework DepthwiseConvolution heuristic on OpenCL by Gian Marco Iodice · 1 year, 10 months ago
- c4f2743 Optimize Quantized/Integer Bilinear Scale for Neon™ by Gunes Bayir · 1 year, 10 months ago
- 0eed305 Optimize FP32/16 Bilinear Scale Kernel for Neon™ by Gunes Bayir · 1 year, 10 months ago
- 211a55d Optimize depthwise convolution on OpenCL by Gian Marco Iodice · 1 year, 10 months ago
- cd083cf Use parent buffer in CLSubTensor. This avoids calling enqueueMapBuffer repeatedly when mapping multiple children of the same parent. by Murray Kornelsen · 2 years ago
- 8bab588 Fix validation problem in CLQLSTMLayer by Pablo Marquez Tello · 1 year, 11 months ago
- 52695f5 Fix CTS/SLTS failure related to Depthwise Convolution by Gunes Bayir · 1 year, 11 months ago
- 13b623e [ONCPUML-968] Fixed format kernel support in additional APIs by Milos Puzovic · 1 year, 11 months ago
- 385dad2 Fix for OpenMP scheduler work breakdown by Milos Puzovic · 1 year, 11 months ago
- 9178002 Fix for inclusion of "arm_gemm" from src into "Types.h" from core by Ramy Elgammal · 2 years ago
- 553f695 [ONCPUML-951] Variable weight support for Convolution. by Francesco Petrogalli · 2 years ago
- a1f7851 Integrate new winograd APIs from MLTech by ramelg01 · 2 years ago
- 4bfc70e Add Gemm MMUL Reshaped Only Rhs Support for FP32/FP16 by Gunes Bayir · 2 years, 7 months ago
- 22dd8b9 Add missing flag when building cl graph examples and fix by Michalis Spyrou · 2 years ago
- f28abef Fix compilation error in 32-bit with logging build by Viet-Hoa Do · 2 years, 1 month ago
- b63b119 Integrate Dynamic Fusion patches by SiCong Li · 2 years, 5 months ago
- 5873358 QLSTM add support for different qinfo by Mike Kelly · 2 years, 2 months ago
- 1bae0f9 NEQLSTM: perform type conversion in prepare method. by Pablo Marquez Tello · 2 years, 2 months ago
- 0a486cf Enable dynamic cl tuning for dynamically fused kernels by SiCong Li · 2 years, 3 months ago
- 16c5697 Add DirectConvolution2D kernel component for dynamic fusion by Gunes Bayir · 2 years, 3 months ago
- 171fc3d Add CPU Pool3d FP16/32 implementation by Adnan AlSinan · 2 years, 4 months ago
- a5d61bf NEQLSTM: Add support for QASYMM8_SIGNED for input_to_forget_weights by Pablo Marquez Tello · 2 years, 3 months ago
- bfd10a1 QLSTM add support for different qinfo in weights by Pablo Marquez Tello · 2 years, 3 months ago
- 3751569 Implementation of ClPooling3d by ramelg01 · 2 years, 4 months ago
- 4e66d70 Added windows native build support by Pablo Tello · 2 years, 4 months ago
- 232c452 Merge kernel prototype patch by Giorgio Arena · 2 years, 4 months ago
- 9454cf7 Fixed threshould argument order in NE/CL/LSTM by Pablo Marquez Tello · 2 years, 4 months ago
- 149203b Port MaxUnpoolingLayer kernel and add KernelSelect vaidation test by Dana Zlotnik · 2 years, 5 months ago
- 6863fa0 Remove deprecated remap functions. by Adnan AlSinan · 2 years, 5 months ago
- 7195f71 Add OpenBSD/arm64 support. by Kevin Lo · 2 years, 6 months ago
- d7154db Implement 1D Adaptive Workload Splitting in CPPScheduler by Dana Zlotnik · 2 years, 8 months ago
- e7a5b0e Initialise quantization info in NEPadLayer by Pablo Marquez Tello · 2 years, 8 months ago
- 3177861 Fix PostOp dependency by SiCongLi · 2 years, 8 months ago
- 579ca84 Add PostOp support to GEMM and CLGEMM operators and functions Part 2 by SiCongLi · 2 years, 8 months ago
- 1af5416 Add experimental PostOp interface to ClGemmMatrixMultiplyReshapedKernel Part 1 by SiCongLi · 2 years, 9 months ago
- 5dda217 DirectConv3d support refine by Sheri Zhang · 2 years, 9 months ago
- c9cecc0 Remove legacy GeMM kernels on OpenCL by Gian Marco Iodice · 2 years, 9 months ago
- 6d9c982 Conv3d support by Sheri Zhang · 2 years, 9 months ago
- 945ae9e Implement CLDirectConv3D f32/f16 by Giorgio Arena · 2 years, 9 months ago
- b6af482 Per-operator build dependencies by Georgios Pinitas · 2 years, 10 months ago
- 8229e5a Fix of linking errors to find CPPSplit symbols when compiling android nn driver by ramelg01 · 2 years, 9 months ago
- 4a6d9e8 Provide logging for configure functions in all CPP functions by ramelg01 · 2 years, 9 months ago
- 6d89157 Provide logging for configure functions in all CL functions by ramelg01 · 2 years, 9 months ago
- cbbb038 Provide logging for configure functions in all NEON functions by ramelg01 · 2 years, 10 months ago
- 8afe6c7 OpenCL job chaining support by Giorgio Arena · 2 years, 10 months ago
- 7891a73 Move CPU/GPU files from Core/Runtime to the respective backend folders by Georgios Pinitas · 2 years, 10 months ago
- 87a74ef Re-use auxiliary memory withing CpuWinogradConv2d operators by Georgios Pinitas · 2 years, 10 months ago
- 511771f Remove padding from ClScaleKernel by Giorgio Arena · 2 years, 10 months ago
- 1988463 Rename [Cl|Cpu]GemmConvolution to [Cl|Gpu]GemmConv2d by Georgios Pinitas · 2 years, 11 months ago
- 73df931 Address comments on avoiding releasing weights if used by multiple functions by Giorgio Arena · 2 years, 11 months ago
- df5bcb6 Enable fast_math on CpuGemmConvolution by Georgios Pinitas · 2 years, 11 months ago
- ebbd529 Update the heuristic to call direct convolution in clConv2d by Gian Marco Iodice · 2 years, 11 months ago
- f5d51f3 Retain weights in ClGemm when reconfiguring the operator with retention by Georgios Pinitas · 2 years, 11 months ago
- 45fbf9f Avoid releasing weights if they are used by multiple functions by Georgios Pinitas · 2 years, 11 months ago
- d4a5bc5 Ensure correct transformed matrices are used in CpuGemmConvolution by Georgios Pinitas · 2 years, 11 months ago
- fa1db17 Ensure that correct transformed matrices are used in CpuFullyConnected by Georgios Pinitas · 2 years, 11 months ago
- d761a3e Fix performance regression due to clFinish() by Gian Marco Iodice · 2 years, 11 months ago
- 288d3cb Fix compiler error in CLActivationLayer by Pablo Marquez Tello · 2 years, 11 months ago
- acb4d93 Fix compiler error in GCC 7.4 + Ubuntu 16 by Pablo Marquez Tello · 2 years, 11 months ago
- 3f74389 Remove 21.08 deprecated code by Freddie Liardet · 2 years, 11 months ago
- 0f6ca4b Report error for unsupported non-constant weights in CpuFullyConnected by Michele Di Giorgio · 2 years, 11 months ago
- 6634194 Avoid over-allocation of temporary buffers within CpuWinogradConv2d by Georgios Pinitas · 2 years, 11 months ago
- 4164814 Implement Operator API by Georgios Pinitas · 2 years, 11 months ago
- 06d1efd Port CLConvolutionLayer by Sheri Zhang · 3 years ago
- 529b5a2 Port ClFullyConnected to new API by Georgios Pinitas · 3 years ago
- a0efe69 Reintroduce implementation of NEConvolutionLayer::get_convolution_method by Michele Di Giorgio · 2 years, 11 months ago
- b55f8e8 Port NEConvolutionLayer by Michalis Spyrou · 3 years ago
- a86433a Reduce binary footprint of CpuConvertFullyConnectedWeightsKernel by Michele Di Giorgio · 3 years ago
- 9805583 Fix memory lifetime issue by Georgios Pinitas · 3 years ago
- d87aded Port CLGEMMConvolutionLayer by Manuel Bottini · 3 years ago
- 4718706 Dispatch Conv2d using the Direct method when necessary by Georgios Pinitas · 3 years ago
- e92c23e Fix allocation of prepare tensor on ClWinogradConv2d by Georgios Pinitas · 3 years ago
- 22f5ed5 Avoid allocation of auxiliary memory in CpuGemmConvolution by Georgios Pinitas · 3 years ago
- d9cdf14 Port NEFullyConnectedLayer to memory injecting interface by Michele Di Giorgio · 3 years ago
- a8297fb Pass fast math flag for correct GEMM3D validation support by Georgios Pinitas · 3 years ago
- 69a9ac4 Expose fast_math mode for GEMM through BFloat16 by Georgios Pinitas · 3 years ago
- 1fa27ad Inject temporary tensors to pack in they don't exist in CpuSoftmax by Georgios Pinitas · 3 years ago
- f4e84fb Port ClGemmLowp to new API by Georgios Pinitas · 3 years ago
- 4ee8b15 Update GEMM assembly kernels by Georgios Pinitas · 3 years ago
- 2b147ee Avoid multiple Rhs matrix transformation on ClGemm by Georgios Pinitas · 3 years ago
- d844c08 Port CLIm2ColKernel to ClIm2ColKernel by Manuel Bottini · 3 years ago
- 0b27133 Disabled DirectConv method for NCHW and kernel > 15 by Pablo Marquez Tello · 3 years ago
- 29599d0 Port NEGEMMConvolutionLayer by Manuel Bottini · 3 years ago