aboutsummaryrefslogtreecommitdiff
path: root/src/core/NEON
AgeCommit message (Collapse)Author
2018-11-02COMPMID-424 Implemented reference implementation, new output valid region ↵Isabella Gottardi
and validation tests (NEON and CL) for Scale Change-Id: I056fa3588b807a97cacf0b8afaec56e37ffc92af Reviewed-on: http://mpd-gerrit.cambridge.arm.com/83872 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-358 Implement OpenCL ROI PoolingSiCong Li
* Implement OpenCL ROI Pooling * Add CLROIPoolingLayer benchmarks Change-Id: I8786d01d551850a1b4d599a48fabe3925e0a27d0 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79833 Reviewed-by: Anthony Barbier <anthony.barbier@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-11-02COMPMID-417: Port PoolingLayer to new validation.Georgios Pinitas
Change-Id: I7f2f5f5f81ad9932661fc4c660bf90614288bc96 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/85270 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-439: Implement NEON Dequantization Layer.Michele Di Giorgio
Change-Id: I2f4f9d0d3437e9d8142f0f82b330233d31ffd552 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/80086 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-345: Added support for 5x5 kernels in NEDirectConvolutionPablo Tello
Change-Id: I25cd8f057566b59ce40e2acf14714e83a286ae4e Reviewed-on: http://mpd-gerrit.cambridge.arm.com/83791 Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-11-02COMPMID-439: Implement NEON Quantization Layer.Michele Di Giorgio
Change-Id: Iefbb421915e56d880d6a3e20c113913560f6ca10 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79934 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2018-11-02COMPMID-345: Optimization for NEFillBorder kernel.Pablo Tello
It's about 0.8 faster than the old code for the special cases where left and top borders are both of size 1. This should improve a bit the performance of many kernels but specially in DirectConvolution where the kernel size is 3. Change-Id: I7d150cac4b1d9bf3bbf897ef6151e139fc34b39c Reviewed-on: http://mpd-gerrit.cambridge.arm.com/83403 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-417: Cleanup NEON FullyConnectedLayerMoritz Pflanzer
Change-Id: Ie02a0a1a28ca2771e29a5e6552242caf0f6db1cf Reviewed-on: http://mpd-gerrit.cambridge.arm.com/83555 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-11-02COMPMID-417: Add in-place support for batch-normalization.Georgios Pinitas
Change-Id: I4b0c9348f3bc2addc198a76fadd1b583abf42b60 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/84434 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
2018-11-02COMPMID-424 NEON/CL Harris Corners validation tests.Giorgio Arena
Change-Id: I82d2a73f515a8d45d16b9ddb702fea51ae05c82e Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79687 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com>
2018-11-02COMPMID-412: Port PoolingLayer to use fixed point 16.Michalis Spyrou
Change-Id: I2005de4c7c14526996309826d33a0ec8e732d2d5 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/78720 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Steven Niu <steven.niu@arm.com>
2018-11-02COMPMID-417 - Bug Fix WarpPerspective kernelIsabella Gottardi
Change-Id: Ic55f9d739f5f8dca65963fdefa6e2528421042b1 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/83550 Reviewed-by: Steven Niu <steven.niu@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-11-02COMPMID-477 - Optimizing CLDirectConvolution 3x3 on OpenCL and added the ↵Gian Marco Iodice
auto configuration Change-Id: I3c8384dcbc9d7786943134bb658dafb35356d90d Reviewed-on: http://mpd-gerrit.cambridge.arm.com/83253 Reviewed-by: Steven Niu <steven.niu@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-424 Implemented reference implementation and validation tests (NEON ↵Isabella Gottardi
and CL) for Warp Perspective Changed the behaviour in NEWarpKernel for border mode replicate and constant to stick with the VX specs. When the new coords are out of the valid region, the output will be computed using the values from the border. In the validation tests the validate will be called with tolerance_value 1 and tolerance_number 0.2%, due to some float arithmetic related mismatches. Change-Id: Id4f9d0ef87178f8f8fd38ee17fee0e6f4beb85cd Reviewed-on: http://mpd-gerrit.cambridge.arm.com/80283 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com> Reviewed-by: Steven Niu <steven.niu@arm.com>
2018-09-17COMPMID-417: Fix invalid read in GEMMMoritz Pflanzer
Change-Id: Ia817dc4415197c310e7e9f9717ce9679294ddfcf Reviewed-on: http://mpd-gerrit.cambridge.arm.com/83109 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-: Implement L2NormalizationLayer.Georgios Pinitas
Change-Id: I2223850b84826a87650729d287c57c22bcab7462 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/82767 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-09-17COMPMID-415 Fixed bug in autoconfigure of NEPoolingLayerKernelAnthony Barbier
Change-Id: I5bcb030a3ceae8004671a73fb430499cc1581db3 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/82947 Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-472 : Implement Floor for CL and NEON.Georgios Pinitas
Change-Id: I675a4545b1fe9ab665a07c834720bfe7ff589cee Reviewed-on: http://mpd-gerrit.cambridge.arm.com/82527 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-09-17COMPMID-417 NEON/CL MeanStdDev bugfix using FillBorderKernelGiorgio Arena
Change-Id: Ic48ba7f69783d0e1e80611264e2bc67d1732436e Reviewed-on: http://mpd-gerrit.cambridge.arm.com/81293 Reviewed-by: Anthony Barbier <anthony.barbier@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-438: Add support for floating point Min-Max Location layer.Michele Di Giorgio
Change-Id: I84ae564a40fc7320a6f94a84d53906ba51404f51 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79797 Reviewed-by: Anthony Barbier <anthony.barbier@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-417: Port NEDirectConvolution 1x1 to QS16.Pablo Tello
Change-Id: Icae6a5091e836d0aca24375f43cca9e6d3a2090f Reviewed-on: http://mpd-gerrit.cambridge.arm.com/81662 Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-09-17COMPMID-413: Add support for QS8 and QS16 CLNormalizationLayer.Michele Di Giorgio
Change-Id: I1aaa9fb8d05796bbca9cfae584e084646552bb71 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/80155 Reviewed-by: Anthony Barbier <anthony.barbier@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-456: Add support for QS16 NEON Normalization Layer.Michele Di Giorgio
Change-Id: I1e542808cfd7774c67cc4e9a58e42449e4fb29aa Reviewed-on: http://mpd-gerrit.cambridge.arm.com/81735 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-09-17COMPMID-417 - Bug Fix WarpPerspective kernelIsabella Gottardi
Change-Id: Ic26fb3b1b60c1a1f4848d683862a25bd1ebc2cc8 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/82053 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Steven Niu <steven.niu@arm.com>
2018-09-17COMPMID-421: Added FP16 support in BatchNormalizationLayer.Pablo Tello
Change-Id: I7142e0e8466ef79e016ae56d285e8e9291573e52 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79814 Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-421: Added FP16 support in the Neon Locally Connected Layer.Pablo Tello
Change-Id: I4b52a209a5ce1a7e69494008538ed242b14b5593 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/81520 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-09-17COMPMID-447: Support scaling factors different than 1 for QS8/QS16 ↵Michele Di Giorgio
NEPixelWiseMultiplication. Change-Id: I6d90a18df861d53546bdca982192b4ffc0dbb3c2 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/80794 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Pablo Tello <pablo.tello@arm.com>
2018-09-17COMPMID-421: Added FP16 support to Softmax.Pablo Tello
Change-Id: If48178689e7cdadf1858556438c7292128be5b92 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/80436 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com>
2018-09-17COMPMID-421: Added FP16 support to the NEON Direct Convolution function.Pablo Tello
Change-Id: I3a1aa2ce985ecf95fc5f441a6e6d43b4935306ee Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79965 Reviewed-by: Anthony Barbier <anthony.barbier@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-421: Added FP16 support in Pooling LayerPablo Tello
Change-Id: I6b6119c8770051c1656da40aa073c539c15b493e Reviewed-on: http://mpd-gerrit.cambridge.arm.com/78985 Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-421: Added FP16 support in ActivationLayer.Pablo Tello
Change-Id: I7ba573b19d56e3c87996edb5218a00e5bfca451e Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79755 Reviewed-by: Anthony Barbier <anthony.barbier@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-421: Added FP16 support to Arithmetic Subtraction.Pablo Tello
Change-Id: I2043531e8e81f28354a208ff91024c3954389422 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/80304 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-09-17COMPMID-401: Implement FixedPointPosition conversion for NEON.Georgios Pinitas
Adds support of changing the fixed point position of a tensor in DepthConvert. Change-Id: Ic3b50a4628fac7497a0217d92941c9d6f64d21cb Reviewed-on: http://mpd-gerrit.cambridge.arm.com/80438 Reviewed-by: Anthony Barbier <anthony.barbier@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-410 Port BatchNormalization to use fixed point 16Michalis Spyrou
Change-Id: I7d3e9ff70c717ef5e6de2bcfbfd277f39006702f Reviewed-on: http://mpd-gerrit.cambridge.arm.com/78956 Reviewed-by: Anthony Barbier <anthony.barbier@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-417: Add Leaky RELU support for both NEON/CL.Georgios Pinitas
-Adds parametrizable leaky relu (x>0) ? x : a*x. Change-Id: Ief19a435b5832a30b56f4aaaf55125787addee94 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/80575 Reviewed-by: Anthony Barbier <anthony.barbier@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-444: Add support for QS8/QS16 NEON Arithmetic Add/Sub/Mul.Michele Di Giorgio
Change-Id: Ia482498688ca1884272b5062e3415e736e03d36f Reviewed-on: http://mpd-gerrit.cambridge.arm.com/80448 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-417: Port DepthConcatenate to QS8/QS16 for NEON/CL.Georgios Pinitas
Change-Id: I3dddae63043c7aa18d908a4fc8abacf3c64f98ca Reviewed-on: http://mpd-gerrit.cambridge.arm.com/80081 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Steven Niu <steven.niu@arm.com>
2018-09-17COMPMID-421: Added FP16 suppot to NENormalizationLayer and ↵Pablo Tello
NEPixelWiseMultiplication. Change-Id: If174f8071502fc5cc94b27cd44a9b1d5e451a9e2 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79553 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2018-09-17COMPMID-421: Added FP16 support to Arithmetic Addition.Pablo Tello
Change-Id: I728f0a856e6581db5b61494a9c4850b963a61573 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/80280 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2018-09-17COMPMID-428: Port NESoftmaxLayer to 16-bit fixed point.Georgios Pinitas
Change-Id: I65122950bab9124b9758c27096c0f458b77aeabb Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79365 Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Steven Niu <steven.niu@arm.com>
2018-09-17COMPMID-421: Added F16 support in FC Layer.Pablo Tello
Change-Id: I9c3ab51ae024be69c7b1d83803b1a8f60a0cdbfd Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79326 Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2018-09-17COMPMID-409: Add support for QS8 and QS16 CLPixelWiseMultiplication.Michele Di Giorgio
Change-Id: I7f66d49d746ba9fb6e726ccab83d3a97b8ddef80 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/78491 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-427: Port NEActivationLayer in 16bit fixed point.Georgios Pinitas
Change-Id: Iebd61807f7b597c6bd990673bc7655c68ee16f4b Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79085 Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
2018-09-17COMPMID-417: Fix assert in GEMMTransposeMoritz Pflanzer
The assert was checking the wrong thing. Only if the window over the input is smaller than the number of processed elements, the output shape would be empty. However, the valid region will be empty if the input's first dimension is less than the number of elements processed. That required the changes in TensorShape. Change-Id: I36fed7893dfd502e26c5c776c9a2d774d6cd91c6 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79813 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
2018-09-17COMPMID-417: Fix output access window in ChannelExtract Kernels.Georgios Pinitas
Change-Id: I0349ef7205e316d85a01e83e86016310143f8886 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79820 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com>
2018-09-17COMPMID-417: Auto initialize for SoftmaxLayer NEON/CL.Georgios Pinitas
Change-Id: I6f35ac7a15fecab93deec4c6266e5c9632e599f0 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79628 Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
2018-09-17COMPMID-417: DepthConvert NEON for QS8/QS16.Georgios Pinitas
Change-Id: Ieb120bccf146045b3a0001ceb3893d4e67fd19df Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79763 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Steven Niu <steven.niu@arm.com>
2018-09-17COMPMID-436, COMPMID-437 - Port NEConvolutionLayer & NEFullyConnectedLayer ↵Gian Marco Iodice
to support 16 bit fixed point Change-Id: I69edf2dac242f941bac95c8479d921e7be6abca7 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79725 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Pablo Tello <pablo.tello@arm.com>
2018-09-17COMPMID-433 - Port NEGEMM to support 16 bit fixed pointGian Marco Iodice
Change-Id: I82de74d7027bbc8a00a4d6671e968785280d5f6c Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79498 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-09-17COMPMID-418 Add check and fix comments after preprocessor conditionsAnthony Barbier
Change-Id: I1353fd652ee180e3931e58b4ce13d651a48c7e2c Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79567 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com>
generated by cgit v1.2.1 (git 2.23.0) at 2024-07-03 20:41:18 +0000