aboutsummaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
2022-02-01 Enable kernel selection testing (Phase #2)Yair Schwarzbaum
Resolves COMPMID-4987 Change-Id: I1201ca3eae107989d13b6a2c6d9560de24fe112d Signed-off-by: Yair Schwarzbaum <yair.schwarzbaum@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/7015 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2022-01-27Remove padding from CpuDirectConv2dKernelAdnan AlSinan
- Replaces NCHW kernels with a generic one - Removes padding Signed-off-by: Adnan AlSinan <adnan.alsinan@arm.com> Change-Id: Ia3d8f788b49a878085b296c11eb83d0152b49eef Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6996 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2022-01-26Fix s10plus NEON/PoolingLayer Nightly failureAdnan AlSinan
Resolves: COMPMID-5041 Signed-off-by: Adnan AlSinan <adnan.alsinan@arm.com> Change-Id: I27a856671ec6210ef7b28f894124a79ca6a4281c Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/7009 Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2022-01-25Add OpenBSD/arm64 support.Kevin Lo
Signed-off-by: Kevin Lo <kevlo@kevlo.org> Change-Id: I6f29bdb55caeec8893f128fdd50bdcc3d058cb3c Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6905 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-by: Pablo Marquez Tello <pablo.tello@arm.com> Comments-Addressed: Pablo Marquez Tello <pablo.tello@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2022-01-25Rework gemm_mm_reshaped_only_rhs_ kernels with new macrosGian Marco Iodice
- Rework gemm_reshaped_rhs_only with new TILE macros - Fuse post ops in gemm_reshaped_rhs_only Resolves COMPMID-4890 Change-Id: I944948ecec6d08deaf3545b80cd3eeac26e44205 Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6944 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Sheri Zhang <sheri.zhang@arm.com>
2022-01-24Fix CPU detection code on MacosPablo Marquez Tello
* Query the number of cpus on the system via sysctl and set CpuInfo accordingly * Query support for fp16 via sysctlbyname and use this to update isainfo * Resolves MLCE-746 Change-Id: I4ad802435c21ee238bff0344af1d7a1ce2479ac5 Signed-off-by: Pablo Marquez Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6989 Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Reviewed-by: SiCong Li <sicong.li@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2022-01-24Select kernel decouplingAnton Vainer
Resolves COMPMID-4614 Signed-off-by: Anton Vainer <anton.vainer@arm.com> Change-Id: I19476d43b8e685de2eed973425d5d31b9cdb84ca Signed-off-by: Anton Vainer <anton.vainer@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6960 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
2022-01-24Fix nightly failure within the elementwise unary build.Dana Zlotnik
Resolves COMPMID-5059 Change-Id: Ic3a76114cc4b0149e80b09494c33c896b6fe9012 Signed-off-by: Dana Zlotnik <dana.zlotnik@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6990 Reviewed-by: SiCong Li <sicong.li@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2022-01-21DepthwiseConv reports full assembly kernel namePablo Marquez Tello
* Fixed the kernel name in CpuDepthwiseConv2dAssemblyWrapperKernel * Resolves MLCE-706 Change-Id: I01ddbe2c030e22e5ba6761ed32110a35c314ccae Signed-off-by: Pablo Marquez Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6787 Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2022-01-21Fix heuristic in ClConv2DGian Marco Iodice
Resolves COMPMID-5040 Change-Id: I12f4c56da71ccee502f5656a80059f903877e8b9 Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6983 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: SiCong Li <sicong.li@arm.com>
2022-01-21A73 Devices Regression 300% fixMohammed Suhail Munshi
- Currently regresses on A73 devices (tested on android hikey, inceptionv3), this patch solves this - Changed mws for all cores to use default values - Existing mws value for A73 tuned for hikey-linux, caused regression on hikey-android Resolves [COMPMID-5044] Change-Id: Ifd6faaa34a0b405d0c390015566f2c75436dfb07 Signed-off-by: Mohammed Suhail Munshi <MohammedSuhail.Munshi@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6973 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: SiCong Li <sicong.li@arm.com>
2022-01-19Decouple CpuElementwiseKernelDana Zlotnik
1- reorganize the folders struct according the new definition 2- separate between unary and binary implementations 3- decuple kernels - unary , binary op and binary comparision Resolves COMPMID-4634 Change-Id: I0195846cc372e74a63c659069a4508de53a22110 Signed-off-by: Dana Zlotnik <dana.zlotnik@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6860 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2022-01-13Decouple CpuSoftmaxKernelDana Zlotnik
Resolves COMPMID-4633 Change-Id: I9f93b28fbc3b18ccaeb453596dc8e0eddfe06b6a Signed-off-by: Dana Zlotnik <dana.zlotnik@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6861 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Marquez Tello <pablo.tello@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2022-01-12Enabled support for QASYMM8 in ClCastKernelPablo Marquez Tello
* Partially resolves MLCE-736 Change-Id: I07140a21d9cc1f804cdf647ce863bb5e2eb10229 Signed-off-by: Pablo Marquez Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6918 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2022-01-12Decouple NEMeanStdDevNormalizationKernelDana Zlotnik
Resolves COMPMID-4617 Change-Id: Ic8793aaf64c6137f848f39c62e33b44ae79ad21d Signed-off-by: Dana Zlotnik <dana.zlotnik@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6870 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2022-01-12Decouple NEInstanceNormalizationLayerKernelDana Zlotnik
Resolves COMPMID-4620 Signed-off-by: Dana Zlotnik <dana.zlotnik@arm.com> Change-Id: I22c285339840493c9cfd4c1abfbc3768ad4db824 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6871 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2022-01-12Enable kernel selection testing (Phase #1)Giorgio Arena
Change-Id: I1d65fb9d3a7583cf8d4163ca7c0fbee27dc52633 Signed-off-by: Yair Schwarzbaum <yair.schwarzbaum@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6767 Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2022-01-11Decouple NEBoundingBoxTransformKernelDana Zlotnik
Resolves COMPMID-4622 Signed-off-by: Dana Zlotnik <dana.zlotnik@arm.com> Change-Id: I18acd03e323f7734635284a763442d2cb4ded177 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6872 Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2022-01-11Decouple NEROIAlignLayerKernelDana Zlotnik
Resolves COMPMID-4624 Change-Id: Ib8d24c44ae62c4f8272310c9305d863d8447eafa Signed-off-by: Dana Zlotnik <dana.zlotnik@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6873 Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2022-01-11Decouple NEGenerateProposalsLayerKernelDana Zlotnik
Resolves COMPMID-4621 Change-Id: I3d89fa6d8273cc5f61a5cc0470fd730919bcd432 Signed-off-by: Dana Zlotnik <dana.zlotnik@arm.com> Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/c/VisualCompute/ComputeLibrary/+/385363 Tested-by: bsgcomp <bsgcomp@arm.com> Comments-Addressed: bsgcomp <bsgcomp@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6867 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2022-01-06Decouple NEMaxUnpoolingLayerKernelDana Zlotnik
Resolves COMPMID-4619 Change-Id: I9c43dcd3fb3a688e1c0ccc858a02376741381ba7 Signed-off-by: Dana Zlotnik <dana.zlotnik@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6874 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2022-01-06Add signbit support for f16 typeSiCongLi
Resolves COMPMID-5037 Change-Id: I648e8810dfd86e2ebcc16554f3f4cc58a5fd8c32 Signed-off-by: SiCongLi <sicong.li@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6899 Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-12-25Add tests for FP Cpu Pooling where pool region is completely outside the inputSiCongLi
* Add floating point validation tests for this configuration * Fix reference implementation to return -inf for this configuration * Prohibit this config in Cl, as well as non-float cases in Cpu * Direct this config to non-asm path Resolves COMPMID-4998 Change-Id: If88025c51b14ea337aea2441c548f858e95e5819 Signed-off-by: SiCongLi <sicong.li@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6857 Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-12-24Fix test validation methodSiCongLi
* Allow non-finite values to be equal (inf == inf, -inf == -inf) in validate * Fix SpecialPoolingLayerValidationFixture Partially resolves COMPMID-4998 Change-Id: I3fba1ccee74c1af419a3b6088ddac68c79aa243a Signed-off-by: SiCongLi <sicong.li@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6856 Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-12-24Replacing non-inclusive terms with proper termsramelg01
Partially-Resolves: COMPMID-4854 Signed-off-by: Ramy Elgammal <ramy.elgammal@arm.com> Change-Id: Ic9757c89878b9b5a89680b5344de657f676c7bf2 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6859 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com>
2021-12-24Fix typo in SConstructSiCongLi
Resolves: COMPMID-5012 Signed-off-by: SiCongLi <sicong.li@arm.com> Change-Id: I7df7098b464f564cf7f3d83e10fe57e2788531a0 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6839 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-12-23Rework gemm_reshape_lhs_ with new macrosAdnan AlSinan
Resolves COMPMID-4892 Signed-off-by: Adnan AlSinan <adnan.alsinan@arm.com> Change-Id: I52f23ca293506fc693ae829daccc6e889a050752 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6833 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: SiCong Li <sicong.li@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-12-23Fix SCons linking error when building for Multi ISA.Motti Gondabi
When building for Multi ISA the link command throws: 'argument list too long' error since the list string being pass to the linker exceeds the ARG_MAX value. The solution is to split the object list into two distinct static libraries and have them link back together. Prohibit unknown variable use. The script now checks for unknown parameters. Resolves: COMPMID-4921 Change-Id: I97a27e86f90a7cc680a3c487da1c468cd3e1084e Signed-off-by: Motti Gondabi <motti.gondabi@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6847 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Reviewed-by: SiCong Li <sicong.li@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-12-23Remove Unusued arguments -pie and -static-libstdc++ in clang80 and clang80++Mohammed Suhail Munshi
Resolves: [COMPMID-4977] Change-Id: If44cb02dbb7e537b135c5ade529f41857133423a Signed-off-by: Mohammed Suhail Munshi <MohammedSuhail.Munshi@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6846 Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Comments-Addressed: Gunes Bayir <gunes.bayir@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-12-22Crop Kernel Decouplingalerah01
Change-Id: I1f39ba6a255847f4d21837a3dce4f867322203e6 Signed-off-by: alerah01 <alex.rahlis@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6799 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
2021-12-20Decouple CpuActivationKernelDana Zlotnik
1- Data types were already decoupled. This commit arrange the folder struct of the activation kernel. 2- Refactor NEON CpuActivationKernel for floating-point cases. Resolves COMPMID-4636 Change-Id: Ia4527244c84260dce1dd1d4bd4a9e3cfe2486d85 Signed-off-by: Dana Zlotnik <dana.zlotnik@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6739 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
2021-12-20Added support for filter size 8x8 NCHW DirectConvPablo Marquez Tello
* Allows NEDeconvLayer to reduce memory usage when workload has filter size 8x8 and NCHW * Resolves MLCE-696 Change-Id: Iaaf40c813376360f813d5babfb988d3e04e4bbc0 Signed-off-by: Pablo Marquez Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6806 Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-12-17Fix logging bug in graph librarySiCongLi
Resolves: COMPMID-5011 Signed-off-by: SiCongLi <sicong.li@arm.com> Change-Id: Ife9ab5492239ac967e6b8b1434548ec62d6e5d98 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6825 Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-12-17Redirect Pool2d unsupported cases to unoptimized pathGiorgio Arena
Unsupported case: Any of the convolution padding > pool size Resolve COMPMID-4999 Signed-off-by: Giorgio Arena <giorgio.arena@arm.com> Change-Id: I4279f3f14d865e0faf53e04847b52ef328970cf2 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6817 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
2021-12-15Unify Cpu Max pool 2d minimum value to be -inf for floating pointSiCongLi
Resolves: COMPMID-4998 Signed-off-by: SiCongLi <sicong.li@arm.com> Change-Id: Ia34c4b063ebcdb193db0a3d6c22f6769a8a902d5 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6822 Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-12-14Update A510 arm_gemm cpu Kernelsramelg01
Resolves: COMPMID-4910 Signed-off-by: Ramy Elgammal <ramy.elgammal@arm.com> Change-Id: I79b4aa51e07ad1fe81d9218ed8a8f34f0ec5ab06 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6803 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Sheri Zhang <sheri.zhang@arm.com>
2021-12-14Add Multi ISA support for SCons build System (part #1)Motti Gondabi
- Enhance the SCons build system to support V8 SVE/SVE2 achitecture in a single binary - Add additional filedefs.json to include build definitions Resolves: COMPMID-4921 Signed-off-by: Motti Gondabi <motti.gondabi@arm.com> Change-Id: Ie3c0ef444303270ba560ca3f43c6e22d50b86679 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6689 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-12-13Remove padding from ClDirectConv2dKernelAdnan AlSinan
- Delete old NCHW ClDirectConv2d kernels. - Merge all kernels on a single file. - Removed padding from ClDirectConv2dKernel Resolves COMPMID-4721 Signed-off-by: Adnan AlSinan <adnan.alsinan@arm.com> Change-Id: I624d218fb770e7b5f3c0acd4e85a21ae48470f55 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6779 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
2021-12-10Fix 300% Regression CPU - Change default mws value in Kernel filesMohammed Suhail Munshi
Resolves: COMPMID-5001 Change-Id: I13fbe859d2557be0459ba76da0136d0efb15f311 Signed-off-by: Mohammed Suhail Munshi <MohammedSuhail.Munshi@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6809 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Sheri Zhang <sheri.zhang@arm.com>
2021-12-10Use #if directive instead of regular condition in CLDirectConv2DGiorgio Arena
Resolve COMPMID-5004 Signed-off-by: Giorgio Arena <giorgio.arena@arm.com> Change-Id: Ib3e1b5a891234316c411ea9825ec10c68c4ab5a3 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6788 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Sheri Zhang <sheri.zhang@arm.com>
2021-12-09Graph Fusion With Post Ops Fixramelg01
- Fusing ConvolutionBatchNormalization Nodes with post ops (activation or element wise ops) Resolves: COMPMID-4982 Signed-off-by: Ramy Elgammal <ramy.elgammal@arm.com> Change-Id: I5b2d32cad00f710fd744cb5aa2d59fd7e5c97e0a Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6766 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Sheri Zhang <sheri.zhang@arm.com>
2021-12-01Improve start-up direct convolution on OpenCLGian Marco Iodice
- Pass arguments at runtime - Rework ClConv2D heuristic to select direct convolution when OFM < IFM also for small kernel sizes Resolves COMPMID-5000 Change-Id: I9b538e29093829bc366d24d1e904341c247fa22b Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6771 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-11-29Use loop unrolling only when the kernel height is less than 5Gian Marco Iodice
- In the dwc_native_fp_nhwc.cl, loop unrolling should only be enabled when kernel height is less than 5. - No performance regression experimented - The patch reduces the compilation time required for the kernel Resolves COMPMID-4887 Change-Id: I93188b9764cf7d1ad34ac164694f6f1fd37a90e8 Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6744 Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-11-28Decouple CpuAddKernelDana Zlotnik
1- NEON supported data types are : fp32, fp16, u8, s16, s32 , q8, q_s8 , q16 2- SVE supported data types are: fp32, fp16, u8, s16, s32 3- SVE2 supported data types are : q8, q_s8 , q16 4- Re-arange SVE folder sturct ** Need to remove gaurds and add testing after Multi ISA build system and validation tests will be avalible Resolves COMPMID-4635 Change-Id: I90e4f6a219478aa9ad5c4a6b9858496afa8af42d Signed-off-by: Dana Zlotnik <dana.zlotnik@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6711 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-11-26Update OpenCL headersPablo Marquez Tello
* Make changes to include CL/opencl.hpp * Update CL C++ header to v2.0.15 * Update CL Headers to v2021.06.30 * Resolves MLCE-665 Change-Id: Ie2896e213519003531ecff0889d2112838d72d1b Signed-off-by: Pablo Marquez Tello <pablo.tello@arm.com> Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/c/VisualCompute/ComputeLibrary/+/377282 Tested-by: bsgcomp <bsgcomp@arm.com> Reviewed-by: Sheri Zhang <sheri.zhang@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6751 Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-11-26Fix Cpu Conv3d gcc 8.3 build issuesFreddie Liardet
Resolves: COMPMID-4986 Signed-off-by: Freddie Liardet <frederick.liardet@arm.com> Change-Id: I54b682d377f3bcfc57fec54113debc5e8a1d75df Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6745 Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2021-11-26Rework gemm_reshape_rhs_(nt,t) with new macrosGian Marco Iodice
Resolves COMPMID-4891 Change-Id: Ifdf2a0eaed23347a1b4465ea8d58c11b72083952 Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6741 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
2021-11-25Fix node fusion mutator orderingSiCongLi
Partially resolves COMPMID-4966 Signed-off-by: SiCongLi <sicong.li@arm.com> Change-Id: I533bd29f416c923b71e7b753965a169253b00a52 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6694 Reviewed-by: Ramy Elgammal <ramy.elgammal@arm.com> Reviewed-by: Sheri Zhang <sheri.zhang@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-11-24Fixed build errorsPablo Marquez Tello
* Resolves problem in macOS Monterey + Clang 13 where there is no memalloc and memalign must be used * Resolves builld error when passing an empty list of files to the AR tool * Resolves MLCE-685 Change-Id: I862ff1dc7f74b2ba32479f6e8abaa32a88d47995 Signed-off-by: Pablo Marquez Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6706 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2021-11-23Decouple data type for NERangeKernelYair Schwarzbaum
- Decouple data type for CPU implementation, supported data types are: fp32, fp16, u8, u16, u32, s8, s16, s32 Resolves COMPMID-4612 Signed-off-by: Yair Schwarzbaum <yair.schwarzbaum@arm.com> Change-Id: Iec9aab9f59c5a344950c788281fc500290a19bbc Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/6686 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
generated by cgit v1.2.1 (git 2.23.0) at 2024-05-23 06:59:48 +0000