aboutsummaryrefslogtreecommitdiff
path: root/src/cpu
AgeCommit message (Collapse)Author
2022-11-30Fix build error for unused variables in data type specific buildsGunes Bayir
The fp32 mws selection variables are guarded because they are flagged as unused when data type support does not include fp32. Resolves: COMPMID-5761 Change-Id: I7ac783e3d5ca51868b562ab879d03e02140b51a1 Signed-off-by: Gunes Bayir <gunes.bayir@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8707 Reviewed-by: Jakub Sujak <jakub.sujak@arm.com> Reviewed-by: Pablo Marquez Tello <pablo.tello@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2022-11-23ONCPUML-1072: Remove double definition of get_mws for Mul kernelfadara01
Signed-off-by: fadara01 <fadi.arafeh@arm.com> Change-Id: Ieaa2fa4a6a69e7e0a48633967dabe91c786b42b7 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8682 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-11-22ONCPUML-1072: Tuned MWS values (for N1, V1) for binary operators used by oneDNNFadi Arafeh
Added approximate values for MWS for the following binary operators: Add, Sub, Mul, Min, Max, Div Change-Id: I5c4c75511129982a3f44c038ee272f09598469de Signed-off-by: Fadi Arafeh <fadi.arafeh@arm.com> Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/c/VisualCompute/ComputeLibrary/+/459609 Tested-by: bsgcomp <bsgcomp@arm.com> Reviewed-by: Viet-Hoa Do <viet-hoa.do@arm.com> Comments-Addressed: bsgcomp <bsgcomp@arm.com> Signed-off-by: fadara01 <fadi.arafeh@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8392 Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-11-15Fix regression caused by mws in ActivationLayerMohammed Suhail Munshi
- Regression is caused by the small default mws in ActivationLayer - Syncronization of threads takes longer than the workload on small sized tensors. - Size 1536 is chosen arbitrarily based on the size of tensors in benchmarked networks Resolves: [COMPMID-5655] Signed-off-by: Mohammed Suhail Munshi <MohammedSuhail.Munshi@arm.com> Change-Id: I02e865b578399e75484f471e67806dd4cf7502c0 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/c/VisualCompute/ComputeLibrary/+/468454 Comments-Addressed: bsgcomp <bsgcomp@arm.com> Tested-by: bsgcomp <bsgcomp@arm.com> Reviewed-by: Jakub Sujak <jakub.sujak@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8615 Benchmark: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2022-11-15Fixed Arm NN unit test failure caused by quantised multiplication patch.Omar Al Khatib
Resolves : [COMPMID-5698] Signed-off-by: Omar Al Khatib <omar.alkhatib@arm.com> Change-Id: I21ce976473e0e8807c14989e98e68aca69c7f1f3 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8603 Benchmark: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2022-11-09Fix CPU multiplication layer threading overheadViet-Hoa Do
* When the tensors are reinterpreted as 1D, any thing smaller than 10KB won't be splitted into different thread. Resolves: COMPMID-5630 Signed-off-by: Viet-Hoa Do <viet-hoa.do@arm.com> Change-Id: Icff7089e37c85c8b325f099008a080a5805d36a2 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8581 Benchmark: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2022-11-08SVE Hard-Swish via Lookup table for quantized inputPablo Marquez Tello
* Datatype supported qasymm8 and qasymm8_signed. * Resolves COMPMID-5211 Change-Id: Ib161af3af5ad11ec6b46de0bf4fbac172d2525fb Signed-off-by: Pablo Marquez Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/7729 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Reviewed-by: Viet-Hoa Do <viet-hoa.do@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-11-07Optimize CPU mul layer on quantized dataOmar Al Khatib
Resolves : [COMPMID-5461] Signed-off-by: Omar Al Khatib <omar.alkhatib@arm.com> Change-Id: I89b99d267c32b00ef44f9bb6e7c714dfe4a0d29d Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8420 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-11-01Fix fixed-point quantized additionViet-Hoa Do
* The range check condition is incorrect. The maximum value of 8-bit quantized data is 255, not 1023. * Use 256 to make the check slightly stricter than it should. Resolves: COMPMID-5458 Signed-off-by: Viet-Hoa Do <viet-hoa.do@arm.com> Change-Id: I5c071680531574b409a7ce71a732f6480caa10d8 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8537 Reviewed-by: Jakub Sujak <jakub.sujak@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2022-11-01Updateable weights in depthwise convolutionMilos Puzovic
Check whether weights are defined as constant, if they are not constant then do not pack them if they are already packed so that they can be updated. Signed-off-by: Milos Puzovic <Milos.Puzovic@arm.com> Change-Id: I73447e31e3660b05f8f40e04ea4ea2003eb9b802 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8539 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-11-01Add threshold for floating-point SOFT_RELU activationMilos Puzovic
Added missing threshold for calculating SOFT_RELU when SVE and CL implementations are used. As a result removed from the testing bounds for input values that were set to be in the interval [-40, 40]. Resolves: COMPMID-5658 Signed-off-by: Milos Puzovic <Milos.Puzovic@arm.com> Change-Id: I3d14df60125e36e4eb85aeb222f4fb0cc5741521 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8536 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Viet-Hoa Do <viet-hoa.do@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-11-01Add check for Batch Matmul in GemmAssemblyDispatchMohammed Suhail Munshi
Relates to : COMPMID-5507 Change-Id: Ia2c4ea153ac2524ffa6b2a9c10f3a0318a8a67a1 Signed-off-by: Mohammed Suhail Munshi <MohammedSuhail.Munshi@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8509 Benchmark: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: SiCong Li <sicong.li@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2022-10-27Fix fixed-point quantized additionViet-Hoa Do
* Use the same rounding function for the left-over part with the vectorized part. Resolves: COMPMID-5640 Signed-off-by: Viet-Hoa Do <viet-hoa.do@arm.com> Change-Id: I07450b2a43390b77539b78cd5d3e6772bdc38548 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8520 Benchmark: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2022-10-20Update reinterpret tensor as 1D for CPU addViet-Hoa Do
* Use the same implementation as other layers. Resolves: COMPMID-5108 Signed-off-by: Viet-Hoa Do <viet-hoa.do@arm.com> Change-Id: I5a50259b398b71ca1f61b5ee8daa539bf8263fac Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8501 Benchmark: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2022-10-20Add test in GEMMLowp for batch matmulMohammed Suhail Munshi
- Adds tests for batched matrix multiplication - Bugfix for issue : 3d tensors input tensors with offsets in GemmLowp results in mismatches Resolves : [COMPMID-5507] Signed-off-by: Mohammed Suhail Munshi <MohammedSuhail.Munshi@arm.com> Change-Id: I68e036fbca642c1841dd4321033045aadc8f5636 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/c/VisualCompute/ComputeLibrary/+/461298 Comments-Addressed: bsgcomp <bsgcomp@arm.com> Tested-by: bsgcomp <bsgcomp@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8482 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Viet-Hoa Do <viet-hoa.do@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-10-19Fix FFTConvolutionLayer testViet-Hoa Do
* Do not change the tensor info after configure stage. * By fixing this, the 1D optimization for activation layer can be applied to all data types and tensor layout. Resolves: COMPMID-5644 Signed-off-by: Viet-Hoa Do <viet-hoa.do@arm.com> Change-Id: I557f9bb84e5e456c28d6b423584887d7a3648ad4 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8470 Benchmark: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Mohmun02 <MohammedSuhail.Munshi@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2022-10-12Optimize Neon™ Logistic ActivationMohammed Suhail Munshi
- Use a 1d execution window to improve memory access pattern. Resolves: [COMPMID-5465] Signed-off-by: Mohammed Suhail Munshi <MohammedSuhail.Munshi@arm.com> Change-Id: Ida30669ffa06eb002ca43a6edf15e25a6eaad2f6 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8344 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-10-12Adding documentation section explaining how BF16 is usedRamy Elgammal
Resolves: COMPMID-5494 Signed-off-by: Ramy Elgammal <ramy.elgammal@arm.com> Change-Id: I8f512745855b8ca21181a9ab21323bfff6aeb866 Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/c/VisualCompute/ComputeLibrary/+/458884 Tested-by: bsgcomp <bsgcomp@arm.com> Reviewed-by: Viet-Hoa Do <viet-hoa.do@arm.com> Comments-Addressed: bsgcomp <bsgcomp@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8391 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Marquez Tello <pablo.tello@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-10-10Fix LUT-based activation layerViet-Hoa Do
* Use the window instead of the tensor shape to determine the number of elements in the x-dimension. * Remove the LUT implementation in 32-bit build. Resolves: COMPMID-5641 Signed-off-by: Viet-Hoa Do <viet-hoa.do@arm.com> Change-Id: I0a79aa38d8f6a105ad01785bd94571f5a2ecb348 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8380 Benchmark: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Marquez Tello <pablo.tello@arm.com>
2022-10-07Optimize Neon™ SUB operator by squashing execution windowJakub Sujak
Resolves: COMPMID-5462 Change-Id: I2c7151c8faf4016cc33592fff04d492d7cbc8fd6 Signed-off-by: Jakub Sujak <jakub.sujak@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8366 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-10-03Fix Batch Matmul nightly failureAdnan AlSinan
Resolves COMPMID-5601 Signed-off-by: Adnan AlSinan <adnan.alsinan@arm.com> Change-Id: I1baf92c4751d784d017c0b2f7de1fc09e42ce69c Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8309 Benchmark: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2022-10-03Optimize CPU add layer on quantized dataViet-Hoa Do
* Use fixed-point arithmetic where possible. * Various optimization for the FP32-based implementation. This implementation is kept as the fall-back solution in case of unrealistic quantization parameters that exceed the range of fixed-point solution. Resolves: COMPMID-5458 Signed-off-by: Viet-Hoa Do <viet-hoa.do@arm.com> Change-Id: I221d2d3801ecaae4fe0b7cf6ae8ef00ca3743665 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8317 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-09-26Add FP32 Neon™ swish activationJonathan Deakin
Change-Id: Id37b59adbc8c4cbe218d1652aeb02a0b4ce42c66 Signed-off-by: Jonathan Deakin <jonathan.deakin@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8256 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Pablo Marquez Tello <pablo.tello@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-09-22Fix unresolved symbol for target armv7a + AndroidPablo Marquez Tello
* Resolves COMPMID-5599 Change-Id: I4c1df48eda289c82ca567f6808fccd0b09065223 Signed-off-by: Pablo Marquez Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8302 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-09-16Fix bug in QASYMM8_SIGNED to F32 cast layerViet-Hoa Do
Resolves: COMPMID-5580 Signed-off-by: Viet-Hoa Do <viet-hoa.do@arm.com> Change-Id: Ia731560c23a6ab2e0ead5a857fbabb9cbc25154c Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/c/VisualCompute/ComputeLibrary/+/452428 Tested-by: bsgcomp <bsgcomp@arm.com> Reviewed-by: Pablo Tello <pablo.tello@arm.com> Comments-Addressed: bsgcomp <bsgcomp@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8268 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
2022-09-16Optimize Quantized/Integer Bilinear Scale for Neon™Gunes Bayir
This patch introduces several performance optimizations regarding the Bilinear Scale operator with REPLICATE Border mode. Changes apply only to NHWC. This patch - Reduces the memory footprint by disabling precomputation of indices and weights when they're not used - Rewrites the kernels for QASYMM8/QASYMM8_SIGNED/U8(Uint8) - Adds S8(Int8) Bilinear Scale for Border mode REPLICATE - Removes Bilinear Scale SVE kernels for Quantized and Integer types and adjust the heuristics to choose the Neon™ implementation - Adds new test cases where the input and output of the Bilinear Scale operator have different quantization scale and offset Resolves: COMPMID-5453, COMPMID-5454 Change-Id: I3d251e76e0c6978fd5a0a1795ec62ab536bec93c Signed-off-by: Gunes Bayir <gunes.bayir@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8250 Reviewed-by: SiCong Li <sicong.li@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-09-14Interpreting tensor as 1D for CPU multiplicationViet-Hoa Do
* Also fix a bug in mul_U8_U8_U8. Resolves: COMPMID-5460 Signed-off-by: Viet-Hoa Do <viet-hoa.do@arm.com> Change-Id: Ie1edafeae7aaad91164caeeb04661a8974a7fc1b Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8244 Reviewed-by: SiCong Li <sicong.li@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-09-14Adding GELU activationMurray Kornelsen
OpenCL implementation uses built in erf. NEON implementation requires new vectorized erf. Uses the following approximation: erf(x) = 1 - 1 / (1 + a1x + a2x^2 + a3x^3 + a4x^4)^4 a1 = 0.278393, a2 = 0.230389, a3 = 0.000972, a4 = 0.078108 From https://en.wikipedia.org/wiki/Error_function#Numerical_approximations Signed-off-by: Murray Kornelsen <murray.kornelsen@mail.mcgill.ca> Change-Id: I2d3964b2c26a4334166b17135f9104bc6324fad2 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/7921 Reviewed-by: Viet-Hoa Do <viet-hoa.do@arm.com> Reviewed-by: Pablo Marquez Tello <pablo.tello@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Pablo Marquez Tello <pablo.tello@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-09-14INT8 Quantized MeanStdDevNorm (LayerNorm)Murray Kornelsen
Implements LayerNorm for qasymm8 tensors. Uses uint8x16 loads and stores. Summation is performed in integer arithmetic (vpaddl) Normalization is performed in float32 before requantizing back to int8. Signed-off-by: Murray Kornelsen <murray.kornelsen@mail.mcgill.ca> Change-Id: I2407c8b34717fb47adab98791bd76fb8a3c62f4a Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/7922 Comments-Addressed: Pablo Marquez Tello <pablo.tello@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Viet-Hoa Do <viet-hoa.do@arm.com> Reviewed-by: Pablo Marquez Tello <pablo.tello@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-09-12Add test for NEGEMM to test a batched matrix multiplication with variable ↵Adnan AlSinan
input tensors - Add a test for CPU to batched matrix multiplication with variable input tensors - Disable assembly kernel when using _reshape_b_only_on_first_run flag Resolves COMPMID-5501 Signed-off-by: Adnan AlSinan <adnan.alsinan@arm.com> Change-Id: If96b182584617806a9dfe597dbfaf05241b123c2 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8234 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-09-09Optimize FP32/16 Bilinear Scale Kernel for Neon™Gunes Bayir
This patch removes index and weight pre-computations where it's not used and reduces some calculations inside the inner-most loop of Scale. Resolves: COMPMID-5452 Change-Id: Ie149b1b76a90a8cb659ada0f97aef78caf69932f Signed-off-by: Gunes Bayir <gunes.bayir@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8220 Benchmark: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2022-09-08Disable Winograd on fp16 if fast-math = falseRamy Elgammal
- That would force CpuConv2d::get_convolution_method() choose GEMM_CONV2D or GEMM methods instead. Resolves: COMPMID-5531 Signed-off-by: Ramy Elgammal <ramy.elgammal@arm.com> Change-Id: I4dec8772e8c150da003d9a89c1d036057c4d28b0 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8233 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
2022-09-02F16 Specialization for MeanStdDevNormMurray Kornelsen
Ran into issues with f16 meanstddevnorm. Essentially, with large enough tensors and/or large values in tensors, output becomes all 0. This is due to the variance computation. In f16, it reaches infinity quite easily, then the division results in 0. This change modifies the OpenCL and NEON implementations to compute the sum of squares and the variance using f32, while other operations remain f16. Update: Found that the square operation also benefits from f32, rather than squaring in f16 and accumulating f32. Signed-off-by: Murray Kornelsen <murray.kornelsen@mail.mcgill.ca> Change-Id: Ide00afd84ec6d26fec4d53b073e295814f08ba46 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/7959 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Comments-Addressed: Pablo Marquez Tello <pablo.tello@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-08-24Fix add for tensors with non-matching stridesJonathan Deakin
Previously, the add_as_1d_array kernels were used on tensors with non-matching strides which caused the wrong elements to be added. The fix is to check that the strides are equal when selecting the addition kernel. Change-Id: I914ca2b95e5b8ed1875ec5ebe129bdfe2845496b Signed-off-by: Jonathan Deakin <jonathan.deakin@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8120 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-08-18Use Neon™ kernels for FP Bilinear Resize for SVEGunes Bayir
Removes FP Bilinear SVE kernels and uses Neon™ kernels instead Resolves: COMPMID-5449 Signed-off-by: Gunes Bayir <gunes.bayir@arm.com> Change-Id: I8e01de44bd884cb6578ca0b9358509b69bc31ca2 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8100 Benchmark: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Viet-Hoa Do <viet-hoa.do@arm.com> Reviewed-by: Pablo Marquez Tello <pablo.tello@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2022-08-17Add LUT for quantized sigmoid functionViet-Hoa Do
* Move LUT implementation to a seperate file. It will be used for both QASYMM8 and QASYMM8_SIGNED. * Fix wrong constant value related to QASYMM8_SIGNED leaky ReLU in 32-bit build. Resolves: COMPMID-5464 Signed-off-by: Viet-Hoa Do <viet-hoa.do@arm.com> Change-Id: I2b24d52409a38f1b66fd532f431eff8a9e4547b6 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8066 Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-08-08Fix for AI benchmark ResNet regressionViet-Hoa Do
* For 3x3 kernel, only choose the implementation with larger tile size if the input tensor is larger than the tile. Resolves: COMPMID-5467 Signed-off-by: Viet-Hoa Do <viet-hoa.do@arm.com> Change-Id: I2cf95ddb25f477cb05da3b3501e0afe9548fc33a Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8022 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-08-04[ONCPUML-970] Fast math mode for fixed format kernelsPablo Marquez Tello
Minor tweaks and test for running fixed format kernels with BF16 operations when specified by the user. Change-Id: Ic8167f67b86b1298da65e46cfebed9f3b86940e4 Signed-off-by: Milos Puzovic <milos.puzovic@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8000 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-08-03[ONCPUML-968] Fixed format kernel support in additional APIsMilos Puzovic
Implements required plumbing in order to be able to ask and execute fixed format kernels from NEFullyConnected, NEGEMM and NEGEMMConv2d. These APIs are used to accelerate oneDNN primitives (inner product, matrix multiplication and indirect GEMM respectively) and without changes it would not be possible to call fixed format kernels from those oneDNN primitives. Change-Id: I27534f0491ce28d0ccb98c19f318bd33dcdf2ff5 Signed-off-by: Milos Puzovic <milos.puzovic@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/7999 Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Reviewed-by: Pablo Marquez Tello <pablo.tello@arm.com> Reviewed-by: SiCong Li <sicong.li@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-08-01Optimize add layer by considering the input tensors as 1D arrayGunes Bayir
Resolves: COMPMID-5108 Change-Id: I544f8160fbe5b4ffbef348d1fbd3dd626a6e1bdb Signed-off-by: Gunes Bayir <gunes.bayir@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8002 Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2022-07-27Fix compilation error rasied in Nightly_NEWRamy Elgammal
- related to 91780021e2 Fix for inclusion of "arm_gemm" - arm_compute::WeightFormat was passed to a function expecting arm_gemm::WeightFormat Resolves: COMPMID-5415 Signed-off-by: Ramy Elgammal <ramy.elgammal@arm.com> Change-Id: Ib53fad2eab0148f466a5e2f11b931754569da3d6 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/7989 Reviewed-by: SiCong Li <sicong.li@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com>
2022-07-26Fix for inclusion of "arm_gemm" from src into "Types.h" from coreRamy Elgammal
- Added arm_compute::WeightFormat and converted to/from arm_gemm::WeightFormat when needed through two map function. - Moved to_string(WeightFormat) to TypePrinter.h Resolves: COMPMID-5415 Signed-off-by: Ramy Elgammal <ramy.elgammal@arm.com> Change-Id: I65f7942100bcd4dbf2c5cf6c07f26c8e1e3bf86e Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/c/VisualCompute/ComputeLibrary/+/438511 Tested-by: bsgcomp <bsgcomp@arm.com> Reviewed-by: Pablo Tello <pablo.tello@arm.com> Reviewed-by: Sicong Li <sicong.li@arm.com> Comments-Addressed: bsgcomp <bsgcomp@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/7985 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-07-25Enable march=armv8.6-a in non multi-isa buildsPablo Marquez Tello
* scons arch=armv8.6-a translates to -march=armv8.6-a * scons arch=armv8.6-a-sve translates to -march=armv8.6-a+sve * scons arch=armv8.6-a-sve2 translates to -march=armv8.6-a+sve2 * Resolves COMPMID-5408 Change-Id: I0901e1de864d00109759509af7cc2b5c9ae1cd75 Signed-off-by: Pablo Marquez Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/7943 Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-07-19[ONCPUML-951] Variable weight support for Convolution.Francesco Petrogalli
API changes for NEGEMMConvolutionLayer and CpuGemmConv2d Built with: scons neon=1 opencl=0 os=linux arch=armv8.2-a multi_isa=1 \ build=native -j32 Werror=false validation_tests=1 build_dir=opt \ standalone=1 asserts=1 experimental_fixed_format_kernels=1 . Tested with: ./build/opt/tests/arm_compute_validation Hardware where the test executable was run: Neoverse N1 Test coverage: * NEGEMMConvolutionLayer, CpuGemmConv2d * NHWC (the only one supported by the fixed-format kernels) * F16, F32 * Shapes: RunSmall Change-Id: I4fd3e495a7cbf61210ea02d37440ba9652934e99 Signed-off-by: Francesco Petrogalli <francesco.petrogalli@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/7632 Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-07-14Integrate new winograd APIs from MLTechramelg01
Resolves: COMPMID-5400 Signed-off-by: Ramy Elgammal <ramy.elgammal@arm.com> Change-Id: Ib4428436dd7a6e40d8b2d8a2f8dac1b079154551 Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/7894 Reviewed-by: Pablo Marquez Tello <pablo.tello@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-07-04Fix build errors on armv8.6 SVE2 with NDK 23 and 24Michalis Spyrou
Extensive use of templates resulted in a compiler crash on NDK 23 and 24. This rework solves the issue and also reduces the library size by 101Kb. Resolves: COMPMID-5384 Change-Id: I9c5c68c5e36f236b0891e44d25478743417fb16d Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/7871 Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-06-30Wrong arguments for running activation function in CpuGemmDirectConv2dMichalis Spyrou
Resolves: COMPMID-5350 Change-Id: I2f5021ebe9fe4c71371f2cd6cb65cc9c0b2110b5 Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/7861 Reviewed-by: Francesco Petrogalli <francesco.petrogalli@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Benchmark: Arm Jenkins <bsgcomp@arm.com>
2022-06-29Add LUT-based leaky relu for QASYMM8 on CPUViet-Hoa Do
* Add LUT generation function for Leaky ReLU. * Some additional changes in the existing LUT implementation: + Bring back the NEON implementation of hard swish for 32-bit build. Library size of 64-bit build is not affected. + Add some extra #ifdef to remove unnecessary code in 32-bit build. Resolves: COMPMID-5386 Signed-off-by: Viet-Hoa Do <viet-hoa.do@arm.com> Change-Id: I1ea49611cc922765ee741e31138c888401d33e9b Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/7845 Benchmark: Arm Jenkins <bsgcomp@arm.com> Tested-by: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gunes Bayir <gunes.bayir@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
2022-06-24Improve LUT Neon Hard-SwishPablo Marquez Tello
* Changed window_step from 16 to tensor_shape().x() when calling into the assembly byte substitution code. * Resolve COMPMID-5211 Change-Id: I5c1f5273455999bb35f94c76a8afb4290e728858 Signed-off-by: Pablo Marquez Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/7843 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
2022-06-23Select neon LUT Hard-Swish kernel on all devicesPablo Marquez Tello
* Resolves COMPMID-5211 Change-Id: I560ab2992c6089774c7ebee3538847905521607d Signed-off-by: Pablo Marquez Tello <pablo.tello@arm.com> Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/7840 Tested-by: Arm Jenkins <bsgcomp@arm.com> Comments-Addressed: Arm Jenkins <bsgcomp@arm.com> Reviewed-by: Viet-Hoa Do <viet-hoa.do@arm.com>
generated by cgit v1.2.1 (git 2.23.0) at 2024-07-02 09:57:26 +0000