aboutsummaryrefslogtreecommitdiff
path: root/src/core/CL/cl_kernels
AgeCommit message (Collapse)Author
2021-06-24Rework gemmlowp reshaped_only_rhs using the new macrosGiorgio Arena
Resolve COMPMID-4416 Change-Id: I83cdf0de7adaf4d465ffebd494ab913182072485 Signed-off-by: Giorgio Arena <giorgio.arena@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5788 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-06-24Rework OpenCL Depthwise ConvolutionGian Marco Iodice
- Remove dedicated kernels for NCHW. Now we only use NHWC with permute - Remove specialized kernels for 3x3 NHWC - Simplify CLDepthwiseConvolutionLayer.cpp to call just the native implementation for both floating-point and quantized data types - Develop two parametric opencl kernels for depthwise convolution layer NHWC (floating-point and quantized) - Add support to export the weights to cl_image - Extend test for depthwise convolution on opencl Resolves COMPMID-4417 Change-Id: I253dd5d959a70783c82e62b1771a5e9f91621cb0 Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5806 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
2021-06-22Add FP16 support to CLRemapFreddie Liardet
Add FP16 support to CLRemap when data layout is NHWC. Add relevant tests for FP16 and validation. Update NERemap function level to be consistent with CLRemap. Add depreciation notice for uint_8 only function level methods. Resolves: COMPMID-4335 Signed-off-by: Freddie Liardet <frederick.liardet@arm.com> Change-Id: If05f06801aef7a169b73ff1ebe760a42f11ca05c Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5816 Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-06-15Add NHWC support to CLRemapFrederick Liardet
Add NHWC support to CLRemap, also add relevant tests. Partially resolves COMPMID-4335. Change-Id: I119bea99be497fb85d5cd83a10f8d4e8e1f97f17 Signed-off-by: Freddie Liardet <frederick.liardet@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5773 Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-06-07Revert "Add optimization for global pooling in pooling_layer.cl"Pablo Tello
* Resolves COMPMID-4544 This reverts commit 50929ef951880469b9d579323d2f9c9f5025327d. Signed-off-by: Pablo Tello <pablo.tello@arm.com> Change-Id: Ibc40054cecc2fe99bff48154f11fe5b602b4a64a Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/c/VisualCompute/ComputeLibrary/+/330380 Tested-by: bsgcomp <bsgcomp@arm.com> Reviewed-by: Sheri Zhang <sheri.zhang@arm.com> Comments-Addressed: bsgcomp <bsgcomp@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5774 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-06-02Fix bug in PReluLayer when input is 1xN sizeFreddie Liardet
Fix issue where gpu elementwise operation kernel would not compile when input is 1xN size and prelu is the chosen operator. Add relevant tests for 1xN input. Signed-off-by: Freddie Liardet <frederick.liardet@arm.com> Change-Id: If0651cfa399ca1d9c65f2632b75536c7931f27d4 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5760 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-06-01Fuse activation in ClDirectConv2dKernel for float typesGeorgios Pinitas
Resolves: COMPMID-4430 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Change-Id: I9a40033e09223d601460a7e52cc297c58c9a2737 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5757 Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-05-20Enable unroll through pragma based on DDK versionGiorgio Arena
Change-Id: Id98a107d512369d3799961011a84e9cc4d99e775 Signed-off-by: Giorgio Arena <giorgio.arena@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5679 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-05-19clCreateKernel failure of CL/ChannelShuffle/U8/Manuel Bottini
Create right vector for size of 16 elements Resolves: COMPMID-4532 Change-Id: I74dc89d158a6aa3f50db710cbc002a92156e2263 Signed-off-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5673 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Sheri Zhang <sheri.zhang@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-05-19Port DepthConvert to new ApiGeorgios Pinitas
- Renames DepthConvert to Cast - Ports both NEDepthConverLayer and CLDepthConvert variants - Removes legacy shift capability from DepthConvert, allowing only shifts of 0 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Change-Id: I806a0f8eb23d23502b632c529fda7edde19c8176 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5565 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-05-17Add macro to manually unroll loops in OpenCLGiorgio Arena
Change-Id: I092d10534816f5b3717325952033c351b8231380 Signed-off-by: Giorgio Arena <giorgio.arena@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5643 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-05-17Fix MeanStdDevNormalizationLayer reference outputting nan for FP16Giorgio Arena
- Bring the epsilon up to 1e-3 for FP16 (both backends) since it was causing the reference's variance being negative and its square root being NaN - Bring the epsilon up to 1e-7 for FP16 NEON test for the same problem on the NEON kernel - Adjust the CL kernel's vec_size when input tensor's width < 16 and use macros agnostic of vector size for sum reduction - Add previously mismatching tensor shapes Resolve COMPMID-4354 Change-Id: I823c871aacb72326f90c86b24cb16c3e2d4bd15e Signed-off-by: Giorgio Arena <giorgio.arena@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5630 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2021-05-13Remove padding from CLChannelShuffleLayerKernelManuel Bottini
Only for NHWC Instead of starting from the input vector and insert the values in the output, we now create an output vector by taking from the input tensor. This makes it easier to store the result in the output with border awareness Resolves: COMPMID-4445 Change-Id: I77cf2d2d5ff30a383cddb8a3c81c462d7a39fd2e Signed-off-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5627 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
2021-05-12Fix GEMMLowp output stage validation crash when input's first dimension == 1Giorgio Arena
- Change vectors' fixed length of 4 in gemmlowp_output_stage_quantize_down_float with its kernel's adjusted vec_size Resolve COMPMID-4355 Signed-off-by: Giorgio Arena <giorgio.arena@arm.com> Change-Id: I371ecf114356fb6a8d18c5c3727f09ae247484bd Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5631 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-05-12Fix bug in Select operator when input is 1xNFreddie Liardet
Fix issue where gpu select kernel would not compile when input was 1xN size. Also add relevant tests for 1xN input. Resolves: COMPMID-4357 Signed-off-by: Freddie Liardet <frederick.liardet@arm.com> Change-Id: Ib5f339379e9cc7ef05605312889dbad34cbfe5dd Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5620 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-05-12Fix 'ARM_DOT_K0XN0' macro redefinedGiorgio Arena
Resolve COMPMID-4339 Change-Id: Ieacd288ca8f69e04d88692a093325a24198ebe3e Signed-off-by: Giorgio Arena <giorgio.arena@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5623 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-05-07Fix missing DATA_TYPE in DOT_PRODUCT4_INTEGER8 OpenCL macroGian Marco Iodice
- DOT_PRODUCT8_INTEGER8 and DOT_PRODUCT16_INTEGER8 are calling DOT_PRODUCT4_INTEGER8 without passing DST_DATA_TYPE Resolves COMPMID-4491 Change-Id: I394bd2f9208489e820885e49ed40e607d6470620 Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5594 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-05-07Remove TODOsSheri Zhang
Remove TODO/FIXME either already done or won't do. Partially resolve: COMPMID-4471 Signed-off-by: Sheri Zhang <sheri.zhang@arm.com> Change-Id: Iec8f25b9bf1b2a70c072dd17d44625fa93e84ed1 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5591 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Georgios Pinitas <georgios.pinitas@arm.com>
2021-05-05Adding S32 support to CLPixelWiseMultiplicationSuhail Munshi
Partially resolves : COMPMID-3793 Signed-off-by: Suhail Munshi <MohammedSuhail.Munshi@arm.com> Change-Id: Id82e00c784f0a039017fd896f11671bdda2dd4ab Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5530 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-05-04Fix bug on CLReductionOperationGiorgio Arena
Execution window along the X axis needs to be collapsed on the 3rd axis (rather than the 2nd) since there could be implicit padding added along the Y Resolve COMPMID-4425 Change-Id: I9623a31749b737fea7c623cabdcfbf77cbe8f6dc Signed-off-by: Giorgio Arena <giorgio.arena@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5551 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: TeresaARM <teresa.charlinreyes@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-04-30Update operator list documentation. Part 2.Teresa Charlin
All data type and data layout information for the operators are store in the function header files Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com> Change-Id: I30b564f7eda6bbd99bf3ad36ddb6639ac118eb8b Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/c/VisualCompute/ComputeLibrary/+/319829 Tested-by: bsgcomp <bsgcomp@arm.com> Reviewed-by: Michele DiGiorgio <michele.digiorgio@arm.com> Comments-Addressed: bsgcomp <bsgcomp@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5531 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: SiCong Li <sicong.li@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-04-30Add optimization for global pooling in pooling_layer.clGian Marco Iodice
- Simplify the implementation when the pooling size has the same spatial dimensions of the input tensor - Rework the heuristic for F32/F16 - Add test for validating the global pooling path - Fix compare_dimensions in validation. The validation fails because we have different number of dimensions for NCHW and NHWC (e.g. 1,1,2,1(NCHW) -> 2,1,1,1(NHWC) Resolves COMPMID-4426 Change-Id: Ia53ee659a9fbc3d011f286a8150d1be9d6d2cd05 Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5533 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-by: TeresaARM <teresa.charlinreyes@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-04-29Remove Global pooling optimizationMichalis Spyrou
It doesn't take into consideration the padding. Change-Id: Ie7a2a0c7aee93fd7007deb6e46f6bec0e7f06066 Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5529 Reviewed-by: TeresaARM <teresa.charlinreyes@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-04-29Remove OpenCL padding: CLReductionOperationKernelGiorgio Arena
Change the parallel implementation across the X, now every thread computes one row Add missing test for MEAN_SUM Make reduction on any axis != 0 work with num_channels > 1 Resolve COMPMID-3917 Signed-off-by: Giorgio Arena <giorgio.arena@arm.com> Change-Id: Ib0f99540104e3c253bcd1ea637833db533f5e76e Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5522 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-04-29Remove stale/solved TODOsMichele Di Giorgio
Change-Id: I5c440f4c6ca4186adcfa926e6b7d924086671f29 Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5520 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Manuel Bottini <manuel.bottini@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-04-29Fix Global Pooling failuresMichalis Spyrou
Correctly calculate the filter size when we exclude padding. Resolves: COMPMID-4408 Change-Id: Ia96d6ddbda58d23a5a0e02854ecb22089d538f94 Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5523 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Sheri Zhang <sheri.zhang@arm.com>
2021-04-28Add explicit cast on GEMMLowp kernelMichalis Spyrou
Oclgrind flags this as a warning. Resolves: COMPMID-4338 Change-Id: Id5a075894027d867bdd21937626f052acb05b531 Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5512 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Manuel Bottini <manuel.bottini@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-04-27Add optimization for global pooling in pooling_layer.clGian Marco Iodice
- Simplify the implementation when the pooling size has the same spatial dimensions of the input tensor - Rework the heuristic for F32/F16 - Add test for validating the global pooling path - Fix compare_dimensions in validation. The validation fails because we have different number of dimensions for NCHW and NHWC (e.g. 1,1,2,1(NCHW) -> 2,1,1,1(NHWC) Change-Id: Iba680cb30bf2a5d0952265a4cc9794f368549ca5 Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5510 Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-04-27Fixed CTS failures CLInstanceNormPablo Tello
* Resolves COMPMID-4400 Change-Id: I54c33a017c735194fbf4437d1c7df465208bc0ca Signed-off-by: Pablo Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5505 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Sheri Zhang <sheri.zhang@arm.com>
2021-04-23[Nightly #1129] CL/Winograd/ConvolutionLayer/F16 mismatch on Mate9Manuel Bottini
Fixing Conv5x5, Conv5x1, Conv1x5 Resolves: COMPMID-4380 Change-Id: I5206d9b85b1d73f6010f02c119aae91266395ba7 Signed-off-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5485 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Aleksandr Nikolaev <aleksandr.nikolaev@arm.com> Reviewed-by: Sheri Zhang <sheri.zhang@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-04-22Fix bug on CLPixelWiseMultiplication broadcasted for Quantized typesGiorgio Arena
Resolve COMPMID-4396 Change-Id: I9b16791f84d60bc4a5303a6393cdbe9db3a4f0e9 Signed-off-by: Giorgio Arena <giorgio.arena@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5483 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: TeresaARM <teresa.charlinreyes@arm.com>
2021-04-21Remove unused yolo_layer OpenCL kernelMichele Di Giorgio
Change-Id: Ibab2095a1a5b525c8513f924cd5bcecd5172ed48 Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5467 Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-04-20CLDepthwiseConvolutionLayer rework - Part 1Giorgio Arena
Remove the reshaped variant for CLDepthwiseConvolutionLayer 3x3 NHWC Quantized - Remove kernel selection by GPUTarget - Remove unused quantized support from the NHWC kernel - Remove CLDepthwiseConvolutionLayerReshapeWeightsKernel - Remove OpenCL kernels for reshaped dwc 3x3 quantized and weights reshape - Remove the "_bifrost" suffix in common OpenCL kernel - Remove the ICLDepthwiseConvolutionLayer3x3Kernel common interface Resolve COMPMID-3864, COMPMID-3907 Change-Id: Icfac0fb6c00e214985beb05dad7c0cdbbee7d830 Signed-off-by: Giorgio Arena <giorgio.arena@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5447 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-04-20Remove OpenCL padding: CLPixelWiseMultiplicationKernelGiorgio Arena
- Change kernel's vec_size to 16 / sizeof(output) - Change ICLKernel.cpp to handle broadcast without padding Resolve COMPMID-3913 Signed-off-by: Giorgio Arena <giorgio.arena@arm.com> Change-Id: I03e884b250ef5784dc109bff8cf2c96b345d119f Signed-off-by: Giorgio Arena <giorgio.arena@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5450 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
2021-04-20[Nightly #1129] CL/Winograd/ConvolutionLayer/F16 mismatch on Mate9Manuel Bottini
Computing the activation in FP32 and then converting in FP16 Resolves: COMPMID-4380 Change-Id: I8a857af65967c8017fb60a358b4f8f0d9fc2e1c2 Signed-off-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5457 Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-04-19Added S32 Integer support to DIV operator in CLElementWiseOperations with TestsSuhail Munshi
Partially Resolves : COMPMID-3793 Signed-off-by: Suhail Munshi <MohammedSuhail.Munshi@arm.com> Change-Id: I14d6884c34f33a6caee11fc1230f9d2d3ae6c4c1 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5425 Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Reviewed-by: Manuel Bottini <manuel.bottini@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-04-19CLInstanceNormalizationLayer NHWC optimisationPablo Marquez Tello
* Make changes to split the workload into two kernels. One kernel precomputes mean and variance and the second kernel just loads these precomputed values. * The new approach runs %30 faster than the original code for NHWC workloads like 32x192x256. * Resolves MLCE-337 Change-Id: I8356fcefa2d131ab4dcb32268ce7142421d073e4 Signed-off-by: Pablo Marquez Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5355 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
2021-04-19Remove padding from CLNormalizePlanarYUVLayerKernelSheri Zhang
Resolve: COMPMID-3911 Signed-off-by: Sheri Zhang <sheri.zhang@arm.com> Change-Id: Id5615b6a8b52030fb611a1a04bcd4664b8232e90 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5451 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-04-14Remove OpenCL padding: CLNormalizationLayerKernelManuel Bottini
Only for NHWC data layout Resolves: COMPMID-3910 Change-Id: Ie2d71482b3e3b55ac155e9af152032a5de8bbd50 Signed-off-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5388 Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-04-13Fix TILE initialization in direct convolution and winograd transformsGian Marco Iodice
- The array initializer for the TILE object cannot always be utilized and so we do require to manually initialize the TILE with the LOOP_UNROLLING macro - Resolves COMPMID-4371 Change-Id: I2598354b9fae84c5e3bd11219fffdcdc297215e1 Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5417 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-04-13Fix T_LOAD too few parameters issueSheri Zhang
Resolve: COMPMID-4370 Signed-off-by: Sheri Zhang <sheri.zhang@arm.com> Change-Id: I4b2a8bf252405fe9006784fa1769ad5b6e708a71 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5414 Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-04-12Add support for cl_image in CLDirectConvolutionLayerGian Marco Iodice
- The cl_image object can be used for the weights - cl_image can only work for f32/f16 - Fix the implicit padding on the first dimension X Resolves COMPMID-4341 Change-Id: I04e0901c69e7765c42afceca38c4a840645b9123 Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5393 Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-04-12Fix OpenCL kernel compiling failure with array initializerSheri Zhang
The issue is related with clang version, clang 3.9 has the problem, clange 4.0 works. The workaround is to add an extra {} to make this work. Partial resolves: COMPMID-4348 Signed-off-by: Sheri Zhang <sheri.zhang@arm.com> Change-Id: Ia079cbb3c44d617b1b42cb2af758b5a8ba1a032e Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5399 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Manuel Bottini <manuel.bottini@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-04-09Fix bug on Implicit Padding for CL GEMMMatrixMultiplyInterleavedTransposedManuel Bottini
Resolves: COMPMID-4342 Change-Id: I468c6d68c0284e4ec76f22037a697fff7bc5638c Signed-off-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5391 Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-04-09Winograd Output transform 7x7 reworkGiorgio Arena
Resolve COMPMID-4140 Change-Id: I17db0ee596665598d08d4359a373160f21ab9acd Signed-off-by: Giorgio Arena <giorgio.arena@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5390 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-04-09Fix OpenCL kernel compiling failure with array initilizerSheri Zhang
The issue is related with clang version, clang 3.9 has the problem, clange 4.0 works. The workaround is to add an extra {} to make this work. Resolves: COMPMID-4348 Signed-off-by: Sheri Zhang <sheri.zhang@arm.com> Change-Id: I2d8fc6400f32af5406fbf2d2556127a53b2ce918 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5392 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-04-08Rewritten Winograd (4x4, 5x5) output transformationAleksandr Nikolaev
This patch takes advantage of tile_helpers.h and different data layout input and tmp matrices. Resolves: COMPMID-4142 Signed-off-by: Aleksandr Nikolaev <aleksandr.nikolaev@arm.com> Change-Id: I5d10bd3f08137414ee7520eef1e6d0aef8cbf160 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5382 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
2021-04-08Rework the OpenCL Winograd Input Transformations NHWCGian Marco Iodice
- Rework Winograd Input Transform 3x3 NHWC using the new macros - Rework Winograd Input Transform 5x5 NHWC using the new macros - Rework Winograd Input Transform 7x7 NHWC using the new macros - The new implementation is also faster than before - Winograd Input Transform 5x5/7x7 3x faster Resolves COMPMID-4139 Change-Id: Ia9c8af23a2d47d2db60ec4c44650a63a34ffa0d5 Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5358 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
2021-04-06Remove OpenCL padding: CLL2NormalizeLayerKernelManuel Bottini
Resolves: COMPMID-3909 Change-Id: I00a1705ed202002e2a6053702272181805fa6869 Signed-off-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5360 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-04-01Added Qasymm8 datatype support to CLROIPoolingLayer with TestsSuhail Munshi
Also fixes RoiPoolingLayer not matching reference with Float32 datatype Issue Tests added to check ROIPooling Layer against reference with both Float32 and Qasymm8 input. Resolves : COMPMID-2320 Change-Id: Ib86d2e6b3803e74f922a545ea573da02c28e54cc Signed-off-by: Suhail Munshi <MohammedSuhail.Munshi@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5332 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
generated by cgit v1.2.1 (git 2.23.0) at 2024-07-07 01:30:29 +0000