aboutsummaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
2018-11-02COMPMID-1079: Removed tests asking for too much memory.Pablo Tello
Change-Id: Ia835e164c98513bac6e9858ffb3be89a689bf841 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/129238 Reviewed-by: Anthony Barbier <anthony.barbier@arm.com> Tested-by: Jenkins <bsgcomp@arm.com>
2018-11-02COMPMID-1077 Fixed SVM memory tracking instrumentAnthony Barbier
Change-Id: Iae8f29eb38efce686cd032ab779200d3b4dda4c5 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/129227 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Jenkins <bsgcomp@arm.com>
2018-11-02COMPMID-1041 NEON Winograd: update function to use GEMM functionMichalis Spyrou
Change-Id: I1ecdf10e02193de7f47a72b75cce0d58a1fa1a1c Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/128411 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Tello <pablo.tello@arm.com>
2018-11-02COMPMID-1023: Import memory for OpenCLGeorgios Pinitas
Change-Id: I201bc00a1261814737e6b6878ecfe9904bae0cc1 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/128212 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-1080 - Fix CLFullyConnectedLayer validate() tests failingGian Marco Iodice
is_interleaved_transposed was set to true on Midgard when m == 1 Change-Id: I991be804073c12f6be894abbfdd010744559772d Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/128678 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Jenkins <bsgcomp@arm.com>
2018-11-02COMPMID-959 : Disables HogDetector test on CLGeorgios Pinitas
Change-Id: Ief1d7060a224a05805a7fc9b409ba17fc90e6502 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/128982 Reviewed-by: Anthony Barbier <anthony.barbier@arm.com> Tested-by: Jenkins <bsgcomp@arm.com>
2018-11-02COMPMID-718 : Winograd: add validate method and testsVidhya Sudhan Loganathan
Changed API's to use winograd_info struct instead of individual params Modified validation to test Validate API Change-Id: I796650092165069e2067e02ace3f42a43f545779 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/128991 Reviewed-by: Anthony Barbier <anthony.barbier@arm.com> Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2018-11-02COMPMID-1070: Rewrote im2col nhwc referencePablo Tello
The new reference computes directly the results without the need of using temporary tensors and perform permutations which is problematic for big 4k tensors as we get bad_alloc execptions. Change-Id: I1fb0a495c4e9ca3356de76c9298832db8ace794a Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/128683 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-718 : Winograd: add validate method and testsVidhya Sudhan Loganathan
Validate methods added to Winograd kernels and function. Renamed validation test suit Change-Id: I0a88df436aff0bbaf4fd82213eeda089b87ac5bf Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/127781 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2018-11-02COMPMID-803: Add NHWC data format support for CL batch normalisationMichele Di Giorgio
Change-Id: Ie37588f60b9cfc7b1d09b1e8628fcfb4b17e0717 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/123834 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-1043: Rework GCGEMMMatrixMultiplyKernel interface and allow auto ↵Michele Di Giorgio
initialization of the tensors This patch also: - removes support for already reshaped weights in GCConvolutionLayer - makes GCConvolutionLayer similar to CLGEMMConvolutionLayer - enables usage of the GCGEMM function in GCConvolution instead of calling the GEMM kernels directly Change-Id: I3e4a64335555e86e18585d38d8fda4bfdb44e265 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/127696 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-804: Add NHWC data format support for NEON batch normalisationMichele Di Giorgio
Change-Id: I04892e7be3f5aa58cd95917a4f90a6b4ffcf6efc Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/122897 Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-959: Sets memory manager to CLWinogradGeorgios Pinitas
-Sets memory manager to Winograd functions -Marks CLGEMM inputs as unused if needed Change-Id: I425a3f864c756e0e2b4da895e1730b8822149ba8 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/128891 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-1077 Add OpenCL Memory Usage instrumentAnthony Barbier
Change-Id: Ic5a24f22bec9d6ca71486097f6ad70fb0d83da2b Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/128674 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Jenkins <bsgcomp@arm.com>
2018-11-02COMPMID-1081: Introduced test-wide instrumentsAnthony Barbier
Change-Id: I5831241f3fc503717cc51136453c2bf96d4b420b Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/128484 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2018-11-02COMPMID-1078 Fix CL DepthwiseConvolutionLayer QASYMM8 failing validationGiorgio Arena
Change-Id: Id540490e5faf11c466ff039a20880eeedd6e5ec7 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/128612 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-by: Isabella Gottardi <isabella.gottardi@arm.com>
2018-11-02COMPMID-802 Fix NEIm2Col NHWCGiorgio Arena
Change-Id: I513e0199b6fa665c4a7d2a739f4871b4575ef347 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/128490 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Tello <pablo.tello@arm.com>
2018-11-02COMPMID-959 Refactor OpenCL interceptors to use lambda functionsAnthony Barbier
Change-Id: I29b73a311d7278255b77524f2a5eaaa4dccab711 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/128392 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Jenkins <bsgcomp@arm.com>
2018-11-02COMPMID-1012 : Add OpenCL benchmarks for WinogradVidhya Sudhan Loganathan
Change-Id: I1cc64535fcd87d500d3ee1ec9e9ff0caef5bcf9f Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/128282 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-by: Pablo Tello <pablo.tello@arm.com>
2018-11-02COMPMID-855: Get the library to work on non Mali GPUsAnthony Barbier
Change-Id: Ia6a7b7a9d8b10ebf6b3c6a0fffa10bdf5dd8d8ef Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/128381 Reviewed-by: Pablo Tello <pablo.tello@arm.com> Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2018-11-02COMPMID-1026 - Add support for 4x4 output tile in CLWinogradConvolutionLayerGian Marco Iodice
The performance achieved can be found at the following confluence page: https://confluence.arm.com/display/MLENG/GEMM-based+convolution+vs+Winograd-based+convolution+on+OpenCL Change-Id: I4b690cfdd4eb4ff0cd17b14fdd49ccaa1d1dc85c Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/127729 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2018-11-02COMPMID-1037 Add support for F(4x4, 5x5) in CLWinogradOutputTransformKernelGiorgio Arena
Change-Id: I0b126f03028f08687497b0d79d2e2764f7ed07c8 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/128001 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
2018-11-02COMPMID-797 Integrate Mobilenet QASYMM8 with new graph.Giorgio Arena
Change-Id: I4df63ec2f4eb27a8a6eec2bea27741bf8dec6910 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/126966 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-987: Make beta and gamma optional in BatchNormalizationMichele Di Giorgio
Currently we have beta and gamma compulsory in Batch normalization. There are network that might not need one or both of those. Thus these should be optional with beta(offset) defaulting to zero and gamma(scale) to 1. Will also reduce some memory requirements. Change-Id: I15bf1ec14b814be2acebf1be1a4fba9c4fbd3190 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/123237 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-577: Implement CL validation for GaussianPyramidSanghoon Lee
Change-Id: If879cbe15b14d97818c24d44b29fc69b6c8cb686 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/127601 Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-1056 - Optimizing CLGEMMMatrixMultiplyKernel refactoring the inner loopGian Marco Iodice
Results reported at: https://confluence.arm.com/display/MLENG/GEMM+FP32+performance%3A+ACL+18.05 Change-Id: I3246c4f19c4d21a7d6a44e4593bc5caffc016f81 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/127838 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-761: Add CL/NEON Histogram benchmark testsJohn Richardson
Change-Id: Ie77bbce265ace735a9d8ec9db5bad70c74800507 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/128035 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-959: Add accessors for the OpenCL program cacheAnthony Barbier
Change-Id: I7920ecdf6687341cbcf4d75aecc15c4164c64636 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/127722 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2018-11-02COMPMID-922 - CLGEMM FP16 optimizations - part1Gian Marco Iodice
This patch improves of ~20% GEMM fp16. The results has been reported at the following confluence page: https://confluence.arm.com/display/MLENG/GEMM+FP32+performance%3A+ACL+18.05 I am aware with few cases we have a bit of degradation. However this cases are memory bound anyway (Fully connected layer cases) Change-Id: I183cbb7fba55a0b5eb86532c4dca5efe096096b0 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/128044 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-811 Add NHWC data format support for CL depthwise convolution QASYMM8Giorgio Arena
Change-Id: I89de432f3fbcba7abf9e1d4f8396a4334b4fa2c2 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/118324 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
2018-11-02COMPMID-1037 Add support for F(4x4, 5x5) in CLWinogradInputTransformKernelGiorgio Arena
Change-Id: Iac26936f46d0f7cdd9d2f8393b0092cd5a223c45 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/127675 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
2018-11-02COMPMID-959 - Fixing subtensor validationGian Marco Iodice
Change-Id: I51c878ec9f19da9d05777271d17ecb740387cac8 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/127934 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2018-11-02COMPMID-1042: Graph / Stream not working on x86Georgios Pinitas
Avoid unspecified behavior in graph construction. This is fixed in C++17. Change-Id: I4ef45cb139bbd838103a9922441e32d2d16c33d2 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/127975 Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com> Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-959: Fixed order of init/destruction of CLSymbols / CLKernelLibraryAnthony Barbier
Change-Id: I6871c28db69e1580c2ece73a9294742586db81f0 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/127954 Reviewed-by: Michele DiGiorgio <michele.digiorgio@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Jenkins <bsgcomp@arm.com>
2018-11-02COMPMID-1031: Use LWS hints for G51, G51BIG, G51LIT, and TNOXSam Laynton
Change-Id: Ie07d9225faaef778bdcfdcb56ae42ec95962e48d Signed-off-by: Sam Laynton <sam.laynton@arm.com> Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/126735 Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com> Tested-by: Jenkins <bsgcomp@arm.com>
2018-11-02COMPMID-761: Add CL/NEON EqualizeHistogram benchmark testsJohn Richardson
Change-Id: I81a6fe93ce364e5e5530b54f9e81fd4d25804096 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/127664 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-1054 Update RSH's GEMM to add batch+multi supportMichalis Spyrou
Change-Id: Ib9d91b77f1d51976da4449fa1e6eeeffae307353 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/127876 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Tello <pablo.tello@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-812 Add NHWC data format support for NEON depthwise convolution ↵Giorgio Arena
(optimized case). Change-Id: Icdfd6c02ed526daf4f59a4b76c7bbc1bc48fde74 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/125938 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-1037 Add support for F(4x4, 5x5) in CLWinogradFilterTransformKernelGiorgio Arena
Change-Id: I6dd639bf5df9bc0c133996f75bdee767f70a6cfb Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/127469 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
2018-11-02COMPMID-959: Add FP32 support to GLES GEMMConvolutionMichele Di Giorgio
The following kernels were supposed to have FP32 support but this was not the case because of bugs and missing shaders: - GCCol2Im - GCIm2Col - GCWeightsReshape Change-Id: Ie6ea464db0612757c71c3d40874e7bb0d60f170a Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/127572 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-1051 - Fix validate method in NEGEMMConvolutionLayerGian Marco Iodice
Change-Id: I10e8e1267a09246cac77e677f1c087bb1d80a61b Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/127517 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-959: Disabled winograd on Arm Cortex-A53 while we're working on a fixAnthony Barbier
Change-Id: I29e6288ceb60002208bb9c9bb1a517e56e70c20e Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/127807 Reviewed-by: Pablo Tello <pablo.tello@arm.com> Tested-by: Jenkins <bsgcomp@arm.com>
2018-11-02COMPMID-959: Fixed clang-tidy formatting, made GLES builds fail for standalone=1Anthony Barbier
Change-Id: I746ef0b2f8e02349e6067139e90c2c34949cad03 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/127690 Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com> Tested-by: Jenkins <bsgcomp@arm.com>
2018-11-02COMPMID-584: Add validation to channel_combine kernelsIoan-Cristian Szabo
Change-Id: I67fe3fcea08704d9f4b04d22fe34db83b2697b87 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/110562 Reviewed-by: Pablo Tello <pablo.tello@arm.com> Tested-by: Jenkins <bsgcomp@arm.com>
2018-11-02COMPMID-802 Add NHWC data format support for NEON im2col.Giorgio Arena
Change-Id: I86e678179106a2b83d1c6a7cfe562df91b0f9eb2 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/124000 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Tello <pablo.tello@arm.com>
2018-11-02COMPMID-585: Port OpticalFlow to new validationJohn Richardson
Change-Id: Ia36bd11ca27420d3059eea15df81b237900149ec Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/125175 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: John Richardson <john.richardson@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-1045 - Fix shape_to_valid_region_gaussian_pyramid_half()Gian Marco Iodice
The utility function shape_to_valid_region_gaussian_pyramid_half() did not calculate correctly the valid regions for levels above 1 Change-Id: Id818efe3b95ce4e0ba1bc57f091d912357a8134a Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/127380 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Sanghoon Lee <sanghoon.lee@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-959: Number of threads was overwritten in benchmark_graph testsAnthony Barbier
RunExample would set the number of threads specified on the command line then the default graph config was overwriting it with the default hint. Now the default config of the graph only sets the number of threads if explicitly requested by the user. Change-Id: Ibc945e790abe8f16afd89e2a40cbe07bf0966e84 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/127301 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Tello <pablo.tello@arm.com>
2018-11-02COMPMID-959 Removed assert on cpuid because valgrind always passes '-1'Anthony Barbier
Change-Id: I648336df12308a0819c0e78d0b67d15e6193482b Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/127074 Tested-by: Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Tello <pablo.tello@arm.com>
2018-11-02COMPMID-959 - Fix doxygem comment in CLGEMMConvolutionLayerIsabella Gottardi
and CLFullyConnectedLayer. Change-Id: Idba56a4fabcd3a23d5f916f899d675ff28148328 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/127397 Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com> Tested-by: Jenkins <bsgcomp@arm.com>