aboutsummaryrefslogtreecommitdiff
path: root/src/runtime/NEON/functions
AgeCommit message (Collapse)Author
2020-01-06COMPMID-2801: Add support for QASYMM8_SIGNED in ↵Michele Di Giorgio
NEDirectConvolutionLayerOutputStageKernel Change-Id: Ib047dd1024b8ecac60e2d368cb161ca418c933ff Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-on: https://review.mlplatform.org/c/2503 Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2020-01-06COMPMID-2999: Nighliy Build Failure [655] - Bare metal build for armv7Michele Di Giorgio
Change-Id: I42f06fd2520a7efe5e6d40f7f9829e3d33c9e387 Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-on: https://review.mlplatform.org/c/2534 Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2020-01-02COMPMID-2998: NEON/GEMMConvolutionLayer/Quantized/QASYMM8_SIGNED failing on ↵Michele Di Giorgio
32-bit Change-Id: If8a705ba09f184e264c37453ffbee591f45e7cba Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-on: https://review.mlplatform.org/c/2524 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Marquez <pablo.tello@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-12-20COMPMID-2608: Enable quantization with multiplier greater than 1 on NEONMichele Di Giorgio
Change-Id: Ib2b0c9ac88fc2b645f478c9981f71ee28f2c77fd Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-on: https://review.mlplatform.org/c/2425 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-12-19COMPMID-2980 (Nightly) armv7a build failuresMichalis Spyrou
Change-Id: I8c2a20fc345694d1ad6e0fe63e4f22fb73e6c1df Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com> Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-on: https://review.mlplatform.org/c/2463 Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-12-11COMPMID-2855: CLReduceMean throws error for invalid configsPablo Tello
Change-Id: Icc198a6db7dc79230972e12f226fa6fa1d9de6cb Signed-off-by: Pablo Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/2452 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-12-11COMPMID-2754: Add support for QASYMM8_SIGNED in NE kernels/functions.Georgios Pinitas
Kernels/Functions extended support: - NEBatchToSpaceLayerKernel/NEBatchToSpaceLayer - NEChannelShuffleLayerKernel/NEChannelShuffleLayer - NECol2ImKernel/NECol2Im - NEConvertFullyConnectedWeightsKernel/NEConvertFullyConnectedWeights - NECopyKernel/NECopy - NEConvolutionLayerReshapeWeights - NEDepthToSpaceLayerKernel/NEDepthToSpaceLayer - NEFlattenLayerKernel/NEFlattenLayer - NEFillBorderKernel - NEFullyConnectedLayerReshapeWeights - NEGatherKernel/NEGather - NEGEMMInterleave4x4Kernel - NEGEMMTranspose1xWKernel - NEIm2ColKernel/NEIm2Col - NEMemsetKernel - NEPadLayerKernel/NEPadLayer - NEPermuteKernel/NEPermute - NEReverseKernel/NEReverse - NEReorgLayerKernel/NEReorgLayer - NEReshapeLayerKernel/NEReshapeLayer - NESplit - NESlice - NEStridedSliceKernel/NEStridedSlice - NESpaceToBatchLayerKernel/NESpaceToBatchLayer - NESpaceToDepthLayerKernel/NESpaceToDepthLayerKernel - NEStackLayerKernel/NEStackLayer - NETileKernel/NETile - NETransposeKernel/NETranspose - NEWidthConcatenateLayerKernel/NEHeightConcatenateLayer - NEHeightConcatenateLayerKernel/NEHeightConcatenateLayer - NEDepthConcatenateLayerKernel/NEDepthConcatenateLayer - NEBathConcatenateLayerKernel/NEBatchConcatenateLayer Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Change-Id: Ia070332ad4c4dbced2541dc46f7f2f3a86833b65 Reviewed-on: https://review.mlplatform.org/c/2442 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2019-12-04COMPMID-2305: NEDepthwiseConvolution 3x3: support for QUANT8_PER_CHANNEL_SYMMGiuseppe Rossini
Change-Id: I9a917cff6a089ce6ae16fb4e6066a4194e2e9487 Signed-off-by: Giuseppe Rossini <giuseppe.rossini@arm.com> Reviewed-on: https://review.mlplatform.org/c/2241 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Marquez <pablo.tello@arm.com>
2019-12-04COMPMID-2764: Add support for QASYMM8_SIGNED in NEConvolutionLayer.Georgios Pinitas
Change-Id: I8fbbd2e399f48968337a60147098d04f27c2d1c0 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/2402 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2019-12-04COMPMID-2962 Fix compiler warnings for g++8Michalis Spyrou
Change-Id: I410c4c54577c1f253597821b1ec6cc40d936b96f Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-on: https://review.mlplatform.org/c/2406 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2019-11-27COMPMID-2763 convert functions and helpers for QASYMM8_SIGNEDSang-Hoon Park
Helpers and functions are modified to add capabilities to convert from/to QASYMM8_SIGNED. Change-Id: Iadd94f9a8469da77421d6987ce67738cf16a1c52 Signed-off-by: Sang-Hoon Park <sang-hoon.park@arm.com> Reviewed-on: https://review.mlplatform.org/c/2288 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2019-11-27COMPMID-2805: Add QASYMM8_SIGNED support in NEGEMMLowpOutputStageGeorgios Pinitas
Add support from requantizing down from S32 to Int8 with fixed point requantization. This involves the following: - Compute fixed point multiplication between each entry of input by result_fixedpoint_multiplier - Add bias to final result if bias tensor is not a nullptr - Round to nearest division by a power-of-two using result_shift - Add offset to each result - Clamp the value between the specified min and max bounds - Cast to int8 data type Change-Id: I641b3fac0833c568d8565ccb859bbc561a24c17d Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/2340 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-11-22COMPMID-2855: NEReduceMean throws error for invalid configsPablo Tello
Change-Id: I600507d0de19d7da6c1a13edcfff0a11ea6b5264 Signed-off-by: Pablo Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/2254 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Reviewed-by: Giuseppe Rossini <giuseppe.rossini@arm.com> Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-by: SiCong Li <sicong.li@arm.com>
2019-11-21COMPMID-2920: NEInstanceNormalization fails on NHWC validationsManuel Bottini
Improved TensorInfo to accept DataLayout, useful to test the validate functions Removing nighlies tests Moving all vpadds instructions in add.h Change-Id: I96290a6f26272eae865dba48bbc3c6aee4bc0214 Signed-off-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-on: https://review.mlplatform.org/c/2287 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2019-11-21COMPMID-2152: Implement Fill for NEONKurtis Charnock
Exposes NEMemsetKernel as a function to allow a tensor to be filled with a constant value. Signed-off-by: Kurtis Charnock <kurtis.charnock@arm.com> Change-Id: Iaf1914077afeab2f8cd9f61f7a167efe108fa0ca Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/c/VisualCompute/ComputeLibrary/+/212389 Tested-by: bsgcomp <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/2318 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2019-11-18COMPMID-2853: Fix VGG16 for fp16.Georgios Pinitas
Half the number of granules for fp16 compaired to fp32. Change-Id: Id9a39d29a5b08295fade8271b93ef80a19a536ce Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/2298 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2019-11-13COMPMID-2921 Performance regression on NEON Int8Giorgio Arena
Change-Id: I60ede3e8aeec678a9fb0d8d0f05a8048ab1ced7c Signed-off-by: Giorgio Arena <giorgio.arena@arm.com> Reviewed-on: https://review.mlplatform.org/c/2285 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
2019-11-12COMPMID-2671 use Signed32 for default output data type of ArgMinMaxSang-Hoon Park
Signed32 is used as data types before and after reshaping of ArgMinMax. Change-Id: I230af43a931d4e106de6c72f716ced1dab511084 Signed-off-by: Sang-Hoon Park <sang-hoon.park@arm.com> Reviewed-on: https://review.mlplatform.org/c/2262 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
2019-11-11COMPMID-2853: VGG16 regression for fp32Georgios Pinitas
* Caps dynamic scheduling granule to a max number Change-Id: I35a9239bc9984dbc1b416c40c4c1b4ac7f5808bd Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/2223 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
2019-11-08COMPMID-2452: Fix 32-bit per-channel convolution for NEON.Georgios Pinitas
Rearrange the kernels in run to ensure type conversion takes place before the matrix transformations. Change-Id: Ibf47788fe71a84fd7549f8667549552e15ca8aab Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/2251 Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-11-08COMPMID-2576: Fuse activation in Winograd output transform.Pablo Tello
Change-Id: I26dd1307847adeaaefae0a7374b9858c07d71372 Signed-off-by: Pablo Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/2172 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
2019-11-06COMPMID-2308: NEConvolutionLayer: support QUANT8_SYMM_PER_CHANNEL filtersGeorgios Pinitas
Change-Id: Ic1bf5f0d21ccd525f84213a360f7e199d7f50577 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/2177 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2019-11-06COMPMID-2862: remove forward declaration of ReductionOperationSang-Hoon Park
For simpler client code development using ArgMinMax, the forward declaraion of the class is removed and the required header file is included. Change-Id: I4e7671cb9c9db7fb31c228c6a986bd45ed726478 Signed-off-by: Sang-Hoon Park <sang-hoon.park@arm.com> Reviewed-on: https://review.mlplatform.org/c/2214 Reviewed-by: Manuel Bottini <manuel.bottini@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-10-30COMPMID-2707: add keep_dims parameter to Reduction OperationSang-Hoon Park
The added parameter is used to decide whether or not to keep the target dimension of reduction operation. ArgMinMax operations will always remove the reduced dimension. Following things are updated to support the parameter. - [CL/NEON] functions and reference kernel - [CL/NEON] ArgMinMax function to use ReductionOperation function - [CL/NEON] validation test suite for Reduction and ArgMinMax operations to validate the added parameter - ReductionOperationFixture is modified NOT to pre-populate output tensor and now relies on underlying kernel/function. - Adjust CL validation test suite for Reduction operation to remove excessive test cases with axis values beyond input tensor's dimension. Change-Id: I3e24d276ed469a4201f323001708f0c525f11c4f Signed-off-by: Sang-Hoon Park <sang-hoon.park@arm.com> Reviewed-on: https://review.mlplatform.org/c/2167 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2019-10-26COMPMID-2588: Optimize the output detection kernel required by MobileNet-SSD ↵Giuseppe Rossini
(~27% improvement) Change-Id: Ic6ce570af3878a0666ec680e0efabba3fcfd1222 Signed-off-by: Giuseppe Rossini <giuseppe.rossini@arm.com> Reviewed-on: https://review.mlplatform.org/c/2160 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-10-24COMPMID-2501: Support multiplier > 1 during QASYMM8 requantization for ↵Manuel Bottini
Quantized LSTM Change-Id: I7eddbdf77881f313b707b9e59428245f1330a2cf Signed-off-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-on: https://review.mlplatform.org/c/2119 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Marquez <pablo.tello@arm.com>
2019-10-23COMPMID-2515: Merge optimized depthwise convolution to the generic depthwise ↵Manuel Bottini
convolution function 3RDPARTY_UPDATE Change-Id: Iff9e915c5329c617527b6f5042979f4e21a8b2b8 Signed-off-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-on: https://review.mlplatform.org/c/2022 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2019-10-23COMPMID-2577: Fuse bias addition and activation in gemm assembly kernelsGeorgios Pinitas
Change-Id: I7f52112d2d05b1ea3d3f3d4b19b8eafab05d6c44 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/2141 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Marquez <pablo.tello@arm.com>
2019-10-21COMPMID-2708 NEDepthwiseConvolution Generic: support for QUANT8_PER_CHANNEL_SYMMGiorgio Arena
COMPMID-2470 Implement a new and generic depthwise convolution for NEON QASYMM8 NHWC COMPMID-2477 Enable FP16 data type for the new generic convolution on NEON for NHWC COMPMID-2625 Remove old implementation files for the generic NEDepthwiseConvolution Change-Id: I8f6deda4fc69dd7e472fba3228b1ed5dad172f3e Signed-off-by: Giorgio Arena <giorgio.arena@arm.com> Reviewed-on: https://review.mlplatform.org/c/2094 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-10-16COMPMID-2265 add support for Log Softmax to NEONSang-Hoon Park
Kernel (NEON/reference), validation tests, function and fixture are updated to add support for Log Softmax Change-Id: I641dbf1552f4128c691af8875949ebf88da71ee8 Signed-off-by: Sang-Hoon Park <sang-hoon.park@arm.com> Reviewed-on: https://review.mlplatform.org/c/2075 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-10-16COMPMID-2486: Remove disabled compiler warningsMichalis Spyrou
Removed the following flags: -Wno-format-nonliteral: This had a side effect on Error.h that resulted in rewriting most of the macros. Since I was at it I removed all the va_args in order to comply with DCL50-CPP. -Wno-deprecated-increment-bool -Wno-vla-extension -Wno-mismatched-tags -Wno-redundant-move Change-Id: I7c593854ecc3b7d595b8edcbd6a86d3c2563c6bd Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-on: https://review.mlplatform.org/c/2069 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2019-10-08COMPMID-2486: Remove disabled compiler warningsMichalis Spyrou
Removed -Wno-unused-parameter and -Wno-deprecated-declarations compilation flags. Plus, 3RDPARTY_UPDATE. Change-Id: I43098c7af527d5651aad3c597b508a56f8813dda Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-on: https://review.mlplatform.org/c/2041 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-10-02COMPMID-2661: Implement complex function dynamic tensor support.Georgios Pinitas
Change-Id: I80772cb25514009b030e5ade28cbb71ed352da67 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/2019 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2019-10-02COMPMID-2257: Add support for QASYMM8 in NEGenerateProposalsMichele Di Giorgio
Change-Id: I7d9aa21ecac97847fce209f97dff0dea6e62790a Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-on: https://review.mlplatform.org/c/2020 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Marquez <pablo.tello@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2019-10-01COMPMID-2699: Add support for QASYMM16 in NEQuantizationLayerMichele Di Giorgio
Change-Id: Icb968e37551a9048040e9aaff5329e874c53a2ee Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-on: https://review.mlplatform.org/c/2016 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2019-10-01COMPMID-2693 Weights manager to mark the initial weights as unsuedMichalis Spyrou
Change-Id: Iaa964e71d0aa80fff0a465127ac3716563bbe85f Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-on: https://review.mlplatform.org/c/2017 Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-10-01INFPRF-609:Performance Issue of the Latest ArmCLgiuros01
We were creating too many small GEMM workloads. It affects performance when the number of thread is small and the matrices are bigger (especially when more single-threaded process are running on the same machine) Change-Id: I807019a7b2d043ca72b4bca11eb0b1960da00694 Signed-off-by: giuros01 <giuseppe.rossini@arm.com> Reviewed-on: https://review.mlplatform.org/c/2012 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-09-26COMPMID-2161 [NEON] Create IWeightManager classMichalis Spyrou
Change-Id: I1a9a46da2f98e896b825099151b56d1d8271dd31 Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-on: https://review.mlplatform.org/c/1915 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-09-20COMPMID-2681: Use NE/CL/Pad in NE/CL/GenerateProposalsLayerMichele Di Giorgio
Change-Id: Idf1c64224b0ddd4bdac1120cac4437eb2578bb2b Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-on: https://review.mlplatform.org/c/1963 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Marquez <pablo.tello@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-09-20COMPMID-2257: Implement NEGenerateProposals.Pablo Tello
Change-Id: I8d751f8b09f842a214c305a4530a71d82f16db8f Signed-off-by: Pablo Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/1943 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
2019-09-17COMPMID-2314: Implement NEON INSTANCE_NORMALIZATION functionManuel Bottini
Change-Id: Ibaa574207aedf691953f8af8fa32b6408a1664ec Signed-off-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-on: https://review.mlplatform.org/c/1905 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
2019-09-09MLCE-129: NEPad 30x slower than TensorFlow's implementationManuel Bottini
Change-Id: I44770e6a3134c70c4bd58f890d06cb43c9bd8bff Signed-off-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-on: https://review.mlplatform.org/c/1853 Reviewed-by: Giorgio Arena <giorgio.arena@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-09-09COMPMID-2204: RuntimeContext interface for NEON functions.Georgios Pinitas
This patch creates the interfaces for the runtime context for NEON. Only the Neon backend implements the context which currently only holds an instance of the scheduler. The NEActivationLayer function has been updated to use the new context interface and the corresponding validation tests ported. Change-Id: I32e7e6aa888796dcbbfc5039b1e7f784a24f47da Signed-off-by: Pablo Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/1851 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
2019-09-04COMPMID-2246: Implement NEBoundingBoxTransform.Pablo Tello
Change-Id: I69faa21ccdc295f5369450a12e6e217d873349e3 Signed-off-by: Pablo Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/1854 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
2019-08-29COMPMID-2318: Implement NEROIAlignLayerPablo Tello
Change-Id: I2665c98b5d6523a9ddaef0b7c220aecd22559ce6 Signed-off-by: Pablo Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/1826 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
2019-08-28COMPMID-2605: Add asymmetric padding support for Deconvolution layerMatthew Jackson
Change-Id: I63b773bdce25f1342ccd3a08ded623a1508f70fe Signed-off-by: Matthew Jackson <matthew.jackson@arm.com> Reviewed-on: https://review.mlplatform.org/c/1797 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-by: Giuseppe Rossini <giuseppe.rossini@arm.com>
2019-08-24COMPMID-2336: Validate multiplier and offset calculation in LSTMQuantized ↵Michele Di Giorgio
and DepthwiseConvolution functions This patch also adds validation of internal functions in LSTMQuantizedLayer. Change-Id: Id8dbbfbb421f7d053410476b4bb4ef7d85e5f41e Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com> Signed-off-by: giuros01 <giuseppe.rossini@arm.com> Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-on: https://review.mlplatform.org/c/1794 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2019-08-23COMPMID-2317: Add new QASYMM16 data typeMichele Di Giorgio
Change-Id: I9fbbba633f10d542474a08b1898150b9799b7ae5 Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-on: https://review.mlplatform.org/c/1805 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-08-21COMPMID-2602: FullyConnected VTS failure on CL and Neon TENSOR_QUANT8_ASYMMGian Marco Iodice
Change-Id: I918d2b9e6d1297cf78c7e567b2a95406875f82ad Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Reviewed-on: https://review.mlplatform.org/c/1782 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-08-20COMPMID-2492 Fix Segfault in NEDeconvolutionLayer for NHWCSiCong Li
Manually manage the tensor _permuted_weights instead of letting memory group manage it. This is because the operation _permute_weights occurs inside the prepare stage. Signed-off-by: SiCong Li <sicong.li@arm.com> Change-Id: Idb639dffcc9942b4acf1bb8dab05f516bdec60d1 Reviewed-on: https://review.mlplatform.org/c/1769 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
generated by cgit v1.2.1 (git 2.23.0) at 2024-06-26 19:20:33 +0000