aboutsummaryrefslogtreecommitdiff
path: root/src/core/NEON/kernels
AgeCommit message (Collapse)Author
2019-07-01COMPMID-2407: Add (logistic and tanh) activation support for QSYMM16 for NEONgiuros01
Change-Id: Ib89c9cfe12975e51d1710af736c73ce79e667363 Signed-off-by: giuros01 <giuseppe.rossini@arm.com> Reviewed-on: https://review.mlplatform.org/c/1412 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Manuel Bottini <manuel.bottini@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2019-07-01COMPMID-2426: Logical dead code fixGeorgios Pinitas
Removes unneeded conditional Change-Id: I42f7b33bdd9ba2a9e67fb21ab3861fb3fc137018 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/1436 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: VidhyaSudhan Loganathan <vidhyasudhan.loganathan@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2019-06-28COMPMID-2234 : Add support for axis 3 in NE/CLConcatenateLayerVidhya Sudhan Loganathan
Change-Id: Ic86f89ece3afe72809bc69c6de6fee7d21daa1d4 Signed-off-by: Vidhya Sudhan Loganathan <vidhyasudhan.loganathan@arm.com> Reviewed-on: https://review.mlplatform.org/c/1440 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-06-25COMPMID-2245: Extend NEFuseBatchNormalization to support ↵Manuel Bottini
DepthwiseConvolution weights Change-Id: I2ee4aebfd69865290ed6c78dd17ff1299353317e Signed-off-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-on: https://review.mlplatform.org/c/1371 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giuseppe Rossini <giuseppe.rossini@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
2019-06-25COMPMID-2406: Create a new GEMMLowpQuantizeDownInt32ToInt16ScaleKernel for NEONGian Marco Iodice
Change-Id: I3f3e247728fd6dafca066e41835f0ef9442d9b7a Signed-off-by: giuros01 <giuseppe.rossini@arm.com> Reviewed-on: https://review.mlplatform.org/c/1379 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>
2019-06-24COMPMID-2235: Extend type support for CL/NEON DequantizationLayer.Georgios Pinitas
Adds support for: - QSYMM8 Change-Id: Ia0b839fc844ce0f968dad1b69a001f9a660dbcd5 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/1378 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
2019-06-20COMPMID-2408: Add QSYMM16 support for ElementwiseAddition for NEONManuel Bottini
Change-Id: I22991e9369ffba9b51a94522ff4977933e887b94 Signed-off-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-on: https://review.mlplatform.org/c/1352 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: Giuseppe Rossini <giuseppe.rossini@arm.com>
2019-06-20COMPMID-2383: DepthwiseConv use output qinfoPablo Tello
There was a problem in depthwise_conv where when needing to run permute it was calling auto_init to initialise the temporary tensor with input's qinfo instead of output's qinfo. This patch fixes this and adds some tests to exercise the cases where we have different qinfos Change-Id: I396fbb4d73f6aa5efe03dc2dd727b3e13154d6f5 Signed-off-by: Pablo Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/1270 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giuseppe Rossini <giuseppe.rossini@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2019-06-18COMPMID-2336: Account for padding in NEIm2ColKernel for NHWC.Georgios Pinitas
Change-Id: I494c4acc95cb431b1718ae62c1504522a115ba10 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/1312 Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giuseppe Rossini <giuseppe.rossini@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-06-18COMPMID-2387: Add support for NEMeanStdDevNormalizationLayerMichele Di Giorgio
Change-Id: If5a9558b78f9c696dfc72633f73e9e21e5c2970f Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-on: https://review.mlplatform.org/c/1351 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Marquez <pablo.tello@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2019-06-13COMPMID-2336: Fix armv8.2-a build on AndroidMichele Di Giorgio
Change-Id: I073f38bebfa732bab63cdb6c2a8aed984521abdf Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com> Reviewed-on: https://review.mlplatform.org/c/1334 Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-06-11COMPMID-2380: Create utility functions for is_one and is_zero with floatGian Marco Iodice
Change-Id: If5b968e19cf830d5472395a1b43bf72a456fd331 Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Reviewed-on: https://review.mlplatform.org/c/1322 Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-06-11COMPMID-2321: PRELU support in NEActivationLayergiuros01
Change-Id: Ib320ee7772492cd1b86eba624438da826d47b984 Signed-off-by: giuros01 <giuseppe.rossini@arm.com> Reviewed-on: https://review.mlplatform.org/c/1224 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: Manuel Bottini <manuel.bottini@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
2019-06-07COMPMID-2397: 64bit bare metal build failureMichalis Spyrou
Change-Id: I003973ab658edff580335c47ebe658ca038602ac Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-on: https://review.mlplatform.org/c/1311 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
2019-06-06COMPMID-2336: Fix warnings present in latest gccGeorgios Pinitas
Change-Id: Icb69db76603341333a44cf121cc1c79035b54ae2 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/1295 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
2019-06-05COMPMID-2372: Add support for QASYMM8 for TanhGeorgios Pinitas
-Perform calculations in the floating point domain -Extends checks for Logistic as scale should be 1/256 and offset 0 Change-Id: I90ef4a042f053976936f5d28f8e09b54eec196a2 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/1287 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2019-06-03COMPMID-2225: Add interface support for new quantized data types.Georgios Pinitas
Add support for: -QSYMM8, 8-bit quantized symmetric -QSYMM8_PER_CHANNEL, 8-bit quantized symmetric with per channel quantization Change-Id: I00c4ff98e44af37419470af61419ee95d0de2463 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/1236 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2019-05-29COMPMID-2237Manuel Bottini
Implement SPACE_TO_DEPTH for NEON Change-Id: I9f427bceca6da52671e0096be08772612f4be152 Signed-off-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-on: https://review.mlplatform.org/c/1227 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
2019-05-24COMPMID-2160: Implement Round for NEONUsama Arif
Change-Id: Ie80e2ad294eaf95bc823d979842c320e8fb41f67 Signed-off-by: Usama Arif <usama.arif@arm.com> Reviewed-on: https://review.mlplatform.org/c/1215 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-05-24COMPMID-2336: Add clang-tidy checks in assembly subfolderMichalis Spyrou
Change-Id: I57fc21cfb8e2751e1ebb59f9106764775d09a00a Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-on: https://review.mlplatform.org/c/1222 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giuseppe Rossini <giuseppe.rossini@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2019-05-24COMPMID-2252: Saturate when converting from S32 to U8 IN ↵Usama Arif
NEDepthConvertLayerKernel Change-Id: Ic1f7500cf5d1ab669abf8a5540a8493e3b5ce4d3 Signed-off-by: Usama Arif <usama.arif@arm.com> Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/186336 Tested-by: bsgcomp <bsgcomp@arm.com> Reviewed-by: Manuel Bottini <manuel.bottini@arm.com> Comments-Addressed: bsgcomp <bsgcomp@arm.com> Reviewed-on: https://review.mlplatform.org/c/1220 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-by: Pablo Marquez <pablo.tello@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2019-05-24COMPMID-2240 Implement DEPTH_TO_SPACE for NEONMichalis Spyrou
Change-Id: I705aa0f804093c3628c691e46cca475f2819dc65 Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-on: https://review.mlplatform.org/c/1198 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giuseppe Rossini <giuseppe.rossini@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-05-23COMPMID-2252 NECast.Usama Arif
Change-Id: I7532aea6827a325eb8457132d4787ac527e93cd4 Signed-off-by: Usama Arif <usama.arif@arm.com> Reviewed-on: https://review.mlplatform.org/c/1149 Reviewed-by: Pablo Marquez <pablo.tello@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2019-05-23COMPMID-2336: Initialize zerobuf arrays in gemm transform kernels.Georgios Pinitas
Change-Id: I3e369295a7caece8142376b75796567242c1ee8d Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/1211 Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Marquez <pablo.tello@arm.com>
2019-05-22COMPMID-2279: Implement REDUCE_MAX operator for NEONUsama Arif
Change-Id: Iccd25b8aab1dd871c0d86ec3816b1cbf48370066 Signed-off-by: Usama Arif <usama.arif@arm.com> Reviewed-on: https://review.mlplatform.org/c/1193 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Marquez <pablo.tello@arm.com>
2019-05-22COMPMID-2280: Implement REDUCE_MIN operator for NEONUsama Arif
Change-Id: Iaa8d97e3328ce69dae7a97a7111120ecc61fb465 Signed-off-by: Usama Arif <usama.arif@arm.com> Reviewed-on: https://review.mlplatform.org/c/1192 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Marquez <pablo.tello@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-05-20COMPMID-2243 ArgMinMaxLayer: support new datatypesMichalis Spyrou
Change-Id: I846e833e0c94090cbbdcd6aee6061cea8295f4f9 Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-on: https://review.mlplatform.org/c/1131 Reviewed-by: Giuseppe Rossini <giuseppe.rossini@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-05-17Puts duplicate code into a function in NEBox3x3KernelAlan Kelly
Signed-off-by: Alan Kelly <me@alankelly.dev> Change-Id: Ida867196e57c5fffe7369e0607ae631003a49587 Reviewed-on: https://review.mlplatform.org/c/1046 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2019-05-17Optimizes NEBox3x3FP16Kernel by removing unnecessary int to float conversionsAlan Kelly
Signed-off-by: Alan Kelly <me@alankelly.dev> Change-Id: Icf54d5f37455fc5ba0fbc5f15d6e481a7a751d74 Reviewed-on: https://review.mlplatform.org/c/1044 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2019-05-17Optimizes NEBox3x3Kernel by replacing floating point computations with fixed ↵Alan Kelly
point Signed-off-by: Alan Kelly <me@alankelly.dev> Change-Id: Ic48a2db1053433f47eaa5a2d9dd1fa12e33c3953 Reviewed-on: https://review.mlplatform.org/c/1045 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2019-05-17COMPMID-2335: Crash in scale_bilinear_nhwc_corePablo Tello
NEScaleKernel crashes in scale_bilinear_nhwc_core trying to read the top padding but the configure() method sets the border to 0 for REPLICATE, so the crash can happen if REPLICATE+BILINEAR+USE_PADDING is used when configuring the function. Change-Id: I5f86da46c2525b67e41c7e1564b89bacac54cea0 Signed-off-by: Pablo Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/1141 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Giuseppe Rossini <giuseppe.rossini@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2019-05-16COMPMID-2254Manuel Bottini
Implement NEAbsLayer Change-Id: I88571010d727b2ac8d9fd3838a4d170cf66bf0ce Signed-off-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-on: https://review.mlplatform.org/c/1150 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-by: Giuseppe Rossini <giuseppe.rossini@arm.com>
2019-05-16COMPMID-2194: Add identity function to NEActivationLayerKernel.cppUsama Arif
Change-Id: Id9bc53722b7dfe277812e9e1d749ac4cea35260a Signed-off-by: Usama Arif <usama.arif@arm.com> Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/184712 Tested-by: bsgcomp <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Reviewed-on: https://review.mlplatform.org/c/1143 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Georgios Pinitas <georgios.pinitas@arm.com>
2019-05-15COMPMID-2269: Implement POW operator for NEONUsama Arif
Change-Id: I7135f665d89da3c24c9bbe00e991a64713a41d0e Signed-off-by: Usama Arif <usama.arif@arm.com> Reviewed-on: https://review.mlplatform.org/c/1128 Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-05-14COMPMID-2263: Implement NELogLayerUsama Arif
Change-Id: Ie2ae8f7a0b1803dae42873201cb643c71b26129f Signed-off-by: Usama Arif <usama.arif@arm.com> Reviewed-on: https://review.mlplatform.org/c/1122 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2019-05-14COMPMID-2248Manuel Bottini
L2NormalizeLayer: negative axis Change-Id: Ic164d7a9ddf1615a2e3b0e10430c34194a70f221 Signed-off-by: Manuel Bottini <manuel.bottini@arm.com> Reviewed-on: https://review.mlplatform.org/c/1127 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2019-05-13COMPMID-2268: Implement NEG for NEON.Usama Arif
Change-Id: I90c023dbea8ea12e9af677294ba576b2bfcc02a4 Signed-off-by: Usama Arif <usama.arif@arm.com> Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/184216 Tested-by: bsgcomp <bsgcomp@arm.com> Comments-Addressed: bsgcomp <bsgcomp@arm.com> Reviewed-by: Pablo Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/1099 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Marquez <pablo.tello@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2019-05-10COMPMID-2177 Fix clang warningsMichalis Spyrou
Change-Id: I78039db8c58d7b14a042c41e54c25fb9cb509bf7 Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-on: https://review.mlplatform.org/c/1092 Reviewed-by: VidhyaSudhan Loganathan <vidhyasudhan.loganathan@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-05-09COMPMID-1973: Implement FFTConvolutionLayer on NEONgiuros01
Change-Id: I2e667c0411bda0164a616ffe44473a78de6752c9 Signed-off-by: giuros01 <giuseppe.rossini@arm.com> Reviewed-on: https://review.mlplatform.org/c/1066 Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-05-01COMPMID-1963: Implement FFT (2D) on NEONgiuros01
Change-Id: I3b564be8d7949e00c6544071ef62dd51de838c96 Signed-off-by: giuros01 <giuseppe.rossini@arm.com> Reviewed-on: https://review.mlplatform.org/c/1048 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2019-05-01COMPMID-1995: Fix clang-tidy warningsMichalis Spyrou
- Remove VirtualCall checks - Fix some unused variables errors - Use std::array insted of C style arrays - Various fixes Change-Id: Ife6170b7102de42b8f04e298dcf8476bf90779f0 Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-on: https://review.mlplatform.org/c/1049 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
2019-04-26COMPMID-1961: Implement FFT (1D) on NEONgiuros01
Change-Id: I0bea3bfbc3b0cd9e8c9a0e0f6f430640573f08d1 Signed-off-by: giuros01 <giuseppe.rossini@arm.com> Reviewed-on: https://review.mlplatform.org/c/996 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2019-04-25COMPMID-2114: CLBatchToSpace missing validation testPablo Tello
Added test cases to exercise block_x != block_y path. Fixed CL back validation test to allow different block values for xy Change-Id: I8c354aceec16d9b16c2f7fcd2cae3f09b747ec0a Signed-off-by: Pablo Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/1017 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Usama Arif <usama.arif@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
2019-04-24COMPMID-2048: Add support for dilation in NEDepthwiseConvolution.Usama Arif
Change-Id: If9941e770779fbf918ba5ff0573da9378078b969 Signed-off-by: Usama Arif <usama.arif@arm.com> Reviewed-on: https://review.mlplatform.org/c/999 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Marquez <pablo.tello@arm.com>
2019-04-16COMPMID-2051 Refactor shape_calculator::calculate_concatenate_shapeMichalis Spyrou
Change-Id: Ibf316718d11fa975d75f226925747b21c4efd127 Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-on: https://review.mlplatform.org/c/974 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
2019-04-16COMPMID-1995: Improves DepthwiseConvolution for qasymm8.Georgios Pinitas
Improves NEON depthwise convolution for qasymm8 for both size and performance. Change-Id: If7eaecec60cd46949176f156fa7c76e59ab70c62 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/986 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Marquez <pablo.tello@arm.com>
2019-04-16COMPMID-2063: New Winograd implementationPablo Tello
Refactoring of winograd code reducing the size of the binaries about 8X. Change-Id: If8845bda324573e1a5cf436f354ac8603e88a92e Signed-off-by: Pablo Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/959 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Anthony Barbier <Anthony.barbier@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
2019-04-15COMPMID-1995: Fix NEPoolingLayer for quantized 3x3Georgios Pinitas
Quantized 3x3 pooling layer on NEON did not support different quantization information for the input and output. Change-Id: I38f8da6ec91c91ba37a21d9d0e1a14fd5bb99f86 Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/992 Reviewed-by: Isabella Gottardi <isabella.gottardi@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2019-04-11COMPMID-2012: Remove unnecessary templates from NEON kernelsMichalis Spyrou
Change-Id: I2deb26188c7de7c6ad10d2f51f83e729fed7e5e2 Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-on: https://review.mlplatform.org/c/961 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2019-04-09COMPMID-1995: Fix 32-bit NEDepthwiseConvolution errors.Georgios Pinitas
-Updates padding handling in assembly depthwise kernels. -Fixes 32-bit runs issues for depthwise convolution. Change-Id: I3fe6369397c1d13f5629dd34c068ce4af53c95cd Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com> Reviewed-on: https://review.mlplatform.org/c/939 Reviewed-by: Giuseppe Rossini <giuseppe.rossini@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
generated by cgit v1.2.1 (git 2.23.0) at 2024-06-11 22:04:45 +0000