Gitiles
Code Review
Sign In
review.mlplatform.org
/
ml
/
ComputeLibrary
/
66c656a1d10831d8311f7797b285faa2c30bcb3f
/
src
/
core
/
CL
/
cl_kernels
1f9ca1d
COMPMID-935 Implementing Convolution with Winograd on OpenCL (part 3)
by Giorgio Arena
· 6 years ago
ae2af74
COMPMID-935 - Implementing Convolution with Winograd on OpenCL (Part 1)
by Gian Marco
· 6 years ago
5b2191e
COMPMID-765: Fix incorrect comma position in DepthwiseConv cl kernel.
by Georgios Pinitas
· 6 years ago
933fe86
COMPMID-927: Adding support for FP16 in CLDepthwiseConvolutionLayer3x3
by Michele Di Giorgio
· 6 years ago
9414f64
COMPMID-582: Add validation to channel_extract kernels.
by Ioan-Cristian Szabo
· 7 years ago
287b570
COMPMID-853 Use tile 2 for CL depthwise convolution QASYM8
by Giorgio Arena
· 6 years ago
72f39be
COMPMID-939 Fix mismatches and finalize CLSoftmaxLayer optimization
by Giorgio Arena
· 6 years ago
3cfd237
COMPMID-938: OCLgrind: Mismatches in depthwise convolution on Bifrost
by Georgios Pinitas
· 6 years ago
19835e5
COMPMID-882 - Optimizing GEMMLowp on OpenCL reshaping matrices
by Gian Marco
· 7 years ago
4402cb9
COMPMID-905 Optimize CLSoftmaxLayer for QASYMM8
by Giorgio Arena
· 6 years ago
a086a0a
COMPMID-765 Move direct convolution output stage to the right file
by Giorgio Arena
· 6 years ago
de5a1cc
COMPMID-856: CL Depthwise Convolution QASYMM8 support
by Georgios Pinitas
· 6 years ago
b99f00d
COMPMID-905 Asymm functions support for all vec sizes
by Giorgio Arena
· 6 years ago
a527e8c
COMPMID-828 - Add support for pool widths 4, 5 & 6 and for non square data sizes - Part 2 (CL)
by Isabella Gottardi
· 6 years ago
1167487
COMPMID-897 Merge batch normalization with bounded relu
by Giorgio Arena
· 6 years ago
4e1e7dc
COMPMID-892: OCLGrind failures on both validation and benchmark
by Georgios Pinitas
· 6 years ago
6232d04
COMPMID-907 Optimizing FixedPoint calculation in the output stage of GEMMLowp
by Giorgio Arena
· 6 years ago
25a340f
COMPMID-578: Implement FAST corners for CL/NEON
by Abe Mbise
· 7 years ago
54f18c4
COMPMID-901 - Optimizing CLCol2ImKernel
by Gian Marco
· 6 years ago
c799ed8
COMPMID-895 - Optimizing CLDepthwiseConvolution3x3Kernel
by Gian Marco
· 6 years ago
76faef8
COMPMID-855 - Optimizing im2col on OpenCL (DCHW)
by Gian Marco
· 7 years ago
f6402dd
COMPMID-834 Fix arm_compute_nightly_validation getting killed
by Michalis Spyrou
· 7 years ago
36a0a46
COMPMID-748 - Integrating optimized SGEMM for bifrost
by Gian Marco
· 7 years ago
7b4d547
COMPMID-816 - Optimizing CLGEMMLowpMatrixMultiplyCore - Part1
by Gian Marco
· 7 years ago
a1f7e33
COMPMID-841: Add CL QASYMM8 RELU Activation
by Michele Di Giorgio
· 7 years ago
5237e01
COMPMID-838 Implement CLPermute
by Michalis Spyrou
· 7 years ago
652bde5
COMPMID-674 - Create Google InceptionV3 example
by Georgios Pinitas
· 7 years ago
a0d1183
COMPMID-751 QASYMM8 ActivationLayer optimisation: don't requantize if not necessary
by Giorgio Arena
· 7 years ago
944d3f7
COMPMID-751 Processing 8 elements makes computation up to 80us faster on MobileNet QASYMM8 dwc layers
by Giorgio Arena
· 7 years ago
780db4e
COMPMID-471 Implement Deconvolution on OpenCL
by Michalis Spyrou
· 7 years ago
25f2368
COMPMID-589: Port HOGDescriptor to new validation
by John Richardson
· 7 years ago
1d08a31
COMPMID-765: Collapse execution window in CL kernels.
by Georgios Pinitas
· 7 years ago
5124be5
COMPMID-661: Convolution quantized (#32)
by Chunosov
· 7 years ago
fcd52fb
COMPMID-661: Vectorize im2col and add lws heuristics for convolution kernels #46
by Anthony Barbier
· 7 years ago
58c5794
COMPMID-706 - Add GEMMLowp output stage for scaling by a fixed point number
by Gian Marco
· 7 years ago
0162436
COMPMID-684: 2D In-Map normalization support for CL
by Georgios Pinitas
· 7 years ago
45bcc3a
COMPMID-661: QASYMM8 support for fully connected layer.
by Georgios Pinitas
· 7 years ago
6fdfaa8
COMPMID-713: Address failures in OCLGrind for CLDirectConvolution
by Georgios Pinitas
· 7 years ago
47b5603
COMPMID-712: OCLGrind CLSoftmaxLayer quantized failures
by Georgios Pinitas
· 7 years ago
c000fb8
COMPMID-714: Resolve failures in OCLGrind for CLPhase
by Georgios Pinitas
· 7 years ago
05288a2
COMPMID-697 - Rework GEMMLowp interface on OpenCL
by Gian Marco
· 7 years ago
f202e50
COMPMID-556 Improved indentation and error handling in format_doxygen.py
by Anthony Barbier
· 7 years ago
c809712
COMPMID-556 Add saturation to 8-bit activation. This prevents undefined overflow
by Rob Hughes
· 7 years ago
02bf80d
COMPMID-661: Fix scale border issue (#38)
by Daniil Efremov
· 7 years ago
3e80c7f
COMPMID-661: Optimize FC layer with 2 new Bifrost kernels and LWS tuning (#33)
by Anton Lokhmotov
· 7 years ago
d7295b7
COMPMID-661: Add QASYMM8 support (and basic tests) to CLDepthwiseConvolution3x3 kernel (#28)
by Dmitry Savenko
· 7 years ago
540d008
COMPMID-556: Fixes bias in CLDirectConvolutionLayer to be int32.
by Georgios Pinitas
· 7 years ago
7a49c79
COMPMID-661: issue# 23 Scale border fix (#26)
by Daniil Efremov
· 7 years ago
f1f3ebd
APPBROWSER-298, APPBROWSER-306: Reimplement the common code of compute shader
by Joel Liang
· 7 years ago
624b778
COMPMID-556: Fix CLNormalization issues.
by Georgios Pinitas
· 7 years ago
2f18579
COMPMID-661: Fix rounding in average pooling for uint8.
by Georgios Pinitas
· 7 years ago
f450caa
COMPMID-661: softmax-uint8 implementation (#16)
by Chunosov
· 7 years ago
7068f99
COMPMID-631: Merge branches/gles_compute branch
by Anthony Barbier
· 7 years ago
4df76c9
COMPMID-661: Fix beta in softmax_layer kernel
by Georgios Pinitas
· 7 years ago
af6204c
COMPMID-661: Add avgpool-uint8 support. Optimize avgpool-fp32 for Bifrost. (#13)
by Anton Lokhmotov
· 7 years ago
d6afedc
COMPMID-661: softmax-fp32 optimisation (#14)
by Chunosov
· 7 years ago
d621bca
COMPMID-661: directconv-uint8 (#20)
by Chunosov
· 7 years ago
6f31f8c
Allow running without cl_khr_fp16
by Matthew Bentham
· 7 years ago
adaae7e
COMPMID-647: Exclude padding pixels from averaging factor.
by Georgios Pinitas
· 7 years ago
0063380
IVGCVSW-619: Support for Cl u8 bounded Relu
by Michel Iwaniec
· 7 years ago
00f4d00
COMPMID-648: Fix initial value for MAX pooling for QS types in CL.
by Georgios Pinitas
· 7 years ago
81a26ad
COMPMID-643: Add bias to CLDepthwiseConvolution.
by Georgios Pinitas
· 7 years ago
96880cf
COMPMID-640: FullyConnectedLayer failures on both NEON/CL
by Georgios Pinitas
· 7 years ago
13fc22c
COMPMID-556: Fix CLPoolingLayer checks
by Georgios Pinitas
· 7 years ago
a1ed41f
IVGCVSW-601: support for asymetric padding in cl conv and depthwise conv
by Jaroslaw Rzepecki
· 7 years ago
48a60f9
IVGCVSW-632 CL support for Softmax beta parameter
by Pablo Palmier
· 7 years ago
744b5ed
COMPMID-606 - Fix for S8 failures
by Gian Marco Iodice
· 7 years ago
040bffe
COMPMID417 - Fix illegal scalar access in color_convert.cl
by Gian Marco Iodice
· 7 years ago
f01f9de
COMPMID-545 add CL printf support
by steniu01
· 7 years ago
349feef
COMPMID-417 - Added validation for FP16 CLBatchNormalizationLayer
by Gian Marco Iodice
· 7 years ago
83be745
COMPMID-424 Implemented reference implementation and tests for WarpAffine
by Isabella Gottardi
· 7 years ago
54f366a
COMPMID-417: Fix CL compiler warnings
by Moritz Pflanzer
· 7 years ago
4726fdf
COMPMID-541: Fix padding in CLMinMaxLocationKernel
by Moritz Pflanzer
· 7 years ago
cdf5145
COMPMID-515: L2 Pooling for FP32/FP16 in CL.
by Georgios Pinitas
· 7 years ago
f81652d
COMPMID-516 Increase tolerance rate of Scale, Conv, fully connected and GEMM
by steniu01
· 7 years ago
9fe4144
COMPMID-452 CL Generic Depthwise Convolution implementation.
by Giorgio Arena
· 7 years ago
bf17955
COMPMID-522 - Added support for GlobalPooling in CLPoolingLayer and CLFlattening for 3D tensor
by Gian Marco Iodice
· 7 years ago
52f8b39
COMPMID-417: Fix CLNonLinearFilter
by Georgios Pinitas
· 7 years ago
5ee66ea
COMPMID-462: Implement TensorReshape for NEON and CL.
by Georgios Pinitas
· 7 years ago
56dd726
COMPMID-448: Implement CL Quantization/Dequantization Layer.
by Michele Di Giorgio
· 7 years ago
1c8409d
COMPMID-477 - Optimized CLDirectConvolution1x1 for Bifrost
by Gian Marco Iodice
· 7 years ago
cfb6553
COMPMID-417 Fix ROIPooling
by SiCong Li
· 7 years ago
64ebe5b
COMPMID-519: Add support for Lower and Upper Bounded RELU for CL/NEON
by Georgios Pinitas
· 7 years ago
1fab09f
COMPMID-424 Implemented reference implementation, new output valid region and validation tests (NEON and CL) for Scale
by Isabella Gottardi
· 7 years ago
04f089c
COMPMID-476 L2 Normalization for CL
by Michalis Spyrou
· 7 years ago
3e36369
COMPMID-358 Implement OpenCL ROI Pooling
by SiCong Li
· 7 years ago
edfa9f4
COMPMID-477 - Optimized batched case in CLConvolutionLayer
by Gian Marco Iodice
· 7 years ago
93a690e
COMPMID-452 CL Depthwise Separable Convolution Layer kernel implementation, validation and benchmarking for 3x3xC depthwise filter and DataType::F32.
by Giorgio Arena
· 7 years ago
d60a6b9
COMPMID-477 - Optimized CLNormalizationLayer
by Gian Marco Iodice
· 7 years ago
0c7614f
COMPMID-431 Port OpenCL pooling layer to use fixed point
by steniu01
· 7 years ago
cb29283
COMPMID-477 - Optimizing Pooling 3x3 with stride_x <= 3 on OpenCL
by Gian Marco Iodice
· 7 years ago
1246b63
COMPMID-477 - Optimized Direct Convolution 3x3 and 5x5 (f32) for Bifrost.
by Gian Marco Iodice
· 7 years ago
409ee0a
COMPMID-417: Add in-place support for batch-normalization.
by Georgios Pinitas
· 7 years ago
1e5c157
COMPMID-450 Add YOLOV2 benchmark tests
by SiCong Li
· 7 years ago
db00668
COMPMID-478 Implemnt CL direct convolution 5x5
by steniu01
· 7 years ago
def665a
COMPMID-474 - Add support for QS8/QS16 DirectConvolution CL
by Michalis Spyrou
· 7 years ago
2eac5bd
COMPMID-417 - Fixed bug in CLCol2ImKernek related to the stride passed during the configuration
by Gian Marco Iodice
· 7 years ago
868e541
COMPMID-459 Collapse CL Im2col's higher dimensions
by steniu01
· 7 years ago
5cb4d6a
COMPMID-477 - Optimizing CLDirectConvolution 3x3 on OpenCL and added the auto configuration
by Gian Marco Iodice
· 7 years ago
d8e765b
COMPMID-472 : Implement Floor for CL and NEON.
by Georgios Pinitas
· 7 years ago
Next »