1
I Use This!
Activity Not Available

Commits : Listings

Analyzed 12 months ago. based on code collected 12 months ago.
Nov 24, 2024 — Nov 24, 2025
Commit Message Contributor Files Modified Lines Added Lines Removed Code Location Date
Fix build error in CLROIPoolingLayer More... over 5 years ago
Rework the OpenCL Winograd Input Transformations NHWC More... over 5 years ago
Remove OpenCL padding: CLL2NormalizeLayerKernel More... over 5 years ago
Add Tensor related utilities to the new API More... over 5 years ago
Remove alignment FIXME More... over 5 years ago
Remove Computer Vision generic interfaces and types More... over 5 years ago
Fix bug on Implicit Padding for CL GEMMMatrixMultiplyInterleavedTransposed More... over 5 years ago
Update U8 mmla kernel
mpekatsoula
as Michalis Spyrou
More... over 5 years ago
NEGEMMConvolution and NEGEMMConv2d produce different results More... over 5 years ago
Implicit padding testing along the X axis on high priority operators More... over 5 years ago
Check biases pointer before referencing in CLDirectConvolutionLayer More... over 5 years ago
bf16_to_float: Fix strict aliasing violation More... over 5 years ago
Improve performance of Winograd Output Transform 3x3 More... over 5 years ago
Ensure OpenCL runtimes are initialized first More... over 5 years ago
Workaround for compiler error in gcc-9.2 and 9.3 More... over 5 years ago
Fix indexing of SVE ArithmeticAddition kernel More... over 5 years ago
Mixed data-layout testing on high priority operators More... over 5 years ago
Improve performance of Winograd Input transform 3x3 More... over 5 years ago
Added Qasymm8 datatype support to CLROIPoolingLayer with Tests More... over 5 years ago
Fix SVE build on newer compilers
mpekatsoula
as Michalis Spyrou
More... over 5 years ago
Create architecture documentation More... over 5 years ago
Remove usage of valid window region CL - NHWC
mpekatsoula
as Michalis Spyrou
More... over 5 years ago
Extend direct convolution (F32/F16/QASYMM8) More... over 5 years ago
Remove GLES-related code More... over 5 years ago
Increase tolerance for winograd input transform F32 NHWC More... over 5 years ago
New variant of OpenCL Winograd (4x4,5x5) input transformation More... over 5 years ago
Fixed compiler errors More... over 5 years ago
Add Queue support More... over 5 years ago
Make CpuPixelWiseMultiplicationKernel stateless More... over 5 years ago
Make ClPixelWiseMultiplicationKernel stateless More... over 5 years ago