aboutsummaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
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-415: New framework - base framework [1/5]Moritz Pflanzer
Change-Id: Icfbfb43321c3bbe6e2aa511dd03a613eed7734a5 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79760 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-09-17COMPMID-440, COMPMID-441 - Port CLConvolutionLayer and CLFullyConnectedLayer ↵Gian Marco Iodice
to support 16 bit fixed point Change-Id: I8d8ef2cb5ec453eb83fba8d8077550b96ed4bceb Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79837 Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@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: Increase NEON QS16 tolerance for reciprocalMoritz Pflanzer
Change-Id: Ice12e6205a9aec11f097cb9f82305e2cd3bd3f05 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79903 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Pablo Tello <pablo.tello@arm.com>
2018-09-17COMPMID-421: Fixed compiler error for arm-v8.2-a.Pablo Tello
Change-Id: If4703e58becd244ce15c91dab2192eeccebcb835 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79900 Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-417 Updating 3rd party folder with the new x86_64 libs (The correct ↵Anthony Barbier
ones this time). 3RDPARTY_UPDATE Change-Id: Ib4da455b6f5da612f7a2d66d57ca992e558f7b42 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79882 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@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-420, COMPMID-414 - Port CLConvolutionLayer and CLFullyConnectedLayer ↵Gian Marco Iodice
to use 8 bit fixed point Change-Id: I1cb1b4d7711ad7b569ee691e13a5df1b3430292b Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79565 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2018-09-17COMPMID-417 Updating 3rd party folder with the new x86_64 libsAnthony Barbier
Change-Id: I59d4df17dcd25392f8a262f3aa79ff57ccba4ff4 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79810 Reviewed-by: Anthony Barbier <anthony.barbier@arm.com> Tested-by: Anthony Barbier <anthony.barbier@arm.com>
2018-09-17COMPMID-417: Fix Dimensions::collapseMoritz Pflanzer
Number of dimensions was reduced to zero if more dimensions than available were requested to be collapsed. Correct result is to keep at least one dimension. Change-Id: I094c3f2f41411913a2ac417b42310416642fb5a8 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79801 Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-417 - Fixed bug in gemm_interleave_16bit and gemm_interleave_32_bit ↵Gian Marco Iodice
due to the non non representable numbers in half and float Change-Id: I1590699e788a2cd3ed6e72de4afff75e6346f6d8 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79800 Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@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-434 - Port CLGEMM to support 16 bit fixed pointGian Marco Iodice
Change-Id: I30aef3c7ecd1ee740c2a7f2ce65a63c7dcd66e49 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79630 Reviewed-by: Anthony Barbier <anthony.barbier@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@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>
2018-09-17COMPMID-417: Remove val_to_stringMoritz Pflanzer
The function was a duplicate of the to_string function we have in support/ToolchainSupport.h Change-Id: If1eff674333248f7d189824ced717310afec19b0 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79634 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-421: Fixed FP16 support in Neon GEMM.Pablo Tello
Fixed GEMM FP16 problem with matrices that are not multiple of 32. Added a new test suite NEON/GEMM/Float16/SmallGEMM. Implemented FP16 function to multiply vector by a matrix. Change-Id: Ie6c692885a48d0206bd6fe748332fa83bc286d67 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79118 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com>
2018-09-17COMPMID-417: Revert invsqrt validation implementation.Georgios Pinitas
Change-Id: I7da949aaddef8cbd25b3d05727839b7b82b08eb5 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79530 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com>
2018-09-17COMPMID-417: Auto configuration for Add/Sub/Mul Neon/CL.Georgios Pinitas
Change-Id: I3580de76bc53d342b53443d1077b1407d75a672a Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79570 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Michele DiGiorgio <michele.digiorgio@arm.com> Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com>
2018-09-17COMPMID-417: Auto initialization for PoolingLayer for NEON/CL.Georgios Pinitas
Change-Id: I2c399c5fe30a9d68fb84742771e7ef10beadb071 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79569 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-09-17COMPMID-417: Autoconfigure for BatchNormalization CL/NEON.Georgios Pinitas
Change-Id: I49a410ccd0102699543bfd23a4d9518d781df281 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79563 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-09-17COMPMID-417: Fix neon_cnn example.Georgios Pinitas
Add proper stride and paddding info to avoid incorrect tensor shapes. Change-Id: Ib906e43cf4ef09364acfb03201e765642dbc25af Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79555 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
2018-09-17COMPMID-417: Add autoconfigure in NormalizationLayer CL/NEON.Georgios Pinitas
Change-Id: I6a6ad0c7a92322776de6a4d4cceeb1365859ef4d Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79566 Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-408 Create OpenCL complex math functions for 8 bit fixed point ↵Michalis Spyrou
arithmetic. Logarithm, inverse square root, exponential and multiplication for 8 bit fixed point arithmetic in OPenCL. Change-Id: Ib976da7057242967c940df28ceebf39bc3ea3811 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/78273 Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-432 - Extended Convolution Layer to support rectangular kernelsGian Marco Iodice
Change-Id: I99be1efede4de6dd63ce103fb11196c413757621 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79252 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com>
2018-09-17COMPMID-417: Fix broken build after rebase gone wrongMoritz Pflanzer
Change-Id: I3e7037f2f690130feb6acc6be3935ea93254a183 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79484 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Pablo Tello <pablo.tello@arm.com>
2018-09-17COMPMID-411 - Port CLGEMM to support 8 bit fixed pointGian Marco Iodice
Change-Id: I6c8bd69ae9715e4d83d128b2162fc15aa5561afb Reviewed-on: http://mpd-gerrit.cambridge.arm.com/78804 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com>
2018-09-17COMPMID-417: Use a common create_tensor functionMoritz Pflanzer
Change-Id: I6b0511484a5b433ebec3fd62d778e64dcb4f89b5 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79362 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-417: Include std::vector where needed.Georgios Pinitas
Change-Id: I8a0c88b3c37a3d82f14493fa0b51108b3bdac852 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79407 Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-421: Fixed a problem in Convolution Layer reference values for FP16.Pablo Tello
All methods in std::numeric_limits<float16_t> return 0. Change-Id: I2289e01853e1b2c38afdec119ef6fc8af8a9752e Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79312 Reviewed-by: Anthony Barbier <anthony.barbier@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com>
2018-09-17COMPMID-359: Implement NEON ROIPoolingLayerGeorgios Pinitas
Change-Id: Ibffa738d4016d7221968bd43a4e6e1dab85baee8 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/78623 Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-417: Extract common toolchain support fileMoritz Pflanzer
The support file is needed as Android toolchains are missing some C++11 functions. It further includes functions that have been introduced in C++14 but which we already need. Eventually the support file might become unnecessary if the Android compilers improve and we switch to C++14. Change-Id: Ib5ec506f1a17c78149c315ab2096b901b852c180 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79247 Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Pablo Tello <pablo.tello@arm.com>
2018-09-17COMPMID-414 - Port CLConvolutionLayer to support 8 bit fixed point - ↵Gian Marco Iodice
CLWeightsReshapeKernel Change-Id: Ie32e6bdd557a8243eb9988aa7eab4e4ca2291e79 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/78701 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com>
2018-09-17COMPMID-400 Add support for 16 bit fixed point arithmetic.Michalis Spyrou
Change-Id: Iebfaef1b219d80d6362b7fd4b1357612b31e43cb Reviewed-on: http://mpd-gerrit.cambridge.arm.com/77749 Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-423: Port CLSoftmaxLayer to QS8Georgios Pinitas
Change-Id: I759b7585656d018d7c864425118cd3ec2ca9b0eb Reviewed-on: http://mpd-gerrit.cambridge.arm.com/78908 Reviewed-by: Michele DiGiorgio <michele.digiorgio@arm.com> Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-417: Move ActivationLayer tests to boostMoritz Pflanzer
This is only a quick&dirty solution to prevent the nightlies from failing. Clean up has to be done under COMPMID-426. Change-Id: Ife1a377efa6aaf389a36f1f8d913ad601b3fdb3c Reviewed-on: http://mpd-gerrit.cambridge.arm.com/78966 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Michele DiGiorgio <michele.digiorgio@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2018-09-17COMPMID-424 Clean up validation tests for 2D spatial filter functionsSiCong Li
* Remove shape_to_valid_region_undefined_border * Remove hardcoded border size Change-Id: I3c64148d8f7ce7e960f21d88d3af43bb23501683 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79106 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-417 - Adding support for rectangular kernelsGian Marco Iodice
Change-Id: I4dde0929bc689c83582b95856dd0253228125df2 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/78994 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-378 - Implemented reference implementation and tests (NEON and CL) ↵Isabella Gottardi
for Non Linear Filter Change-Id: I1b81e030a27cf01d098247a87d047099616e2a39 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/78531 Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-424 Add validation tests for Gaussian5x5SiCong Li
* Fix apply_2d_spatial_filter to use double as intermediate type * Fix tensor_elem_at to use random value if on border and border_mode is UNDEFINED Change-Id: I7feea23c4664cc63c5bab936566dc92b98c723b9 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/78905 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com>
2018-09-17COMPMID-417: Check for nullptr in mismatching data types check.Georgios Pinitas
Change-Id: Iff395b0d5f2858b3e24aad74c7cfd0a1fa6bb6d3 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/78999 Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-424 Add tests for Gaussian3x3SiCong Li
* Add reference implementation and NEON CL validation tests for gaussian3x3 * Add tests for different border modes * Incoporate PaddingCalculator Change-Id: Ia3cd5cb2cd0b4d870a217e6aedecc01369dc2541 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/78125 Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com> Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
2018-09-17COMPMID-417: Remove dependency on CLScheduler from coreMoritz Pflanzer
Change-Id: I62b46c3d7356572b4ef769d92693d1bc173d1c4e Reviewed-on: http://mpd-gerrit.cambridge.arm.com/78997 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Steven Niu <steven.niu@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2018-09-17COMPMID-424 Add CL validation tests for Box3x3SiCong Li
* Add tests for different border modes * Add padding calculator Change-Id: Ic4708faddfb1c8e6b59d349cf9cb48c9a181d717 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/78105 Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com> Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com>
2018-09-17COMPMID-378 NEON/CL Sobel 3x3 and 5x5 tests.Giorgio Arena
Change-Id: Ic90edd77c7f694f95663d8163623db3837a48616 Reviewed-on: http://mpd-gerrit.cambridge.arm.com/78284 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 2024-07-05 17:26:50 +0000