Age | Commit message (Collapse) | Author |
|
- Update heuristic for GEMM reshaped RHS only
- Fix left-over block size in CLGEMMMatrixMultiplyReshapedOlyRHSKernel
Change-Id: I34c738821ed2e4a537da4a15058eec164cb6b61f
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4305
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
CLGEMMLowpMatrixMultiplyReshapedOnlyRHSKernel
Change-Id: I45d26d5f565f9a55f6b5e8d7652b14283ae616f7
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4299
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
|
|
- Since the GEMM kernel can now work without padding, the heuristic
requires to be fine-tuned to exploit this feature
- The heuristic affects Mali-G76 FP16 only
Change-Id: Ia430627f02131ad956ce2219b80c83c8e7cabaf2
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4284
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-by: SiCong Li <sicong.li@arm.com>
|
|
This patch solves the following issues that arose from nightly tests:
- The accumulated result of gemmlowp_mm_native can be either uint or
int and in order to be stored in memory we need to convert it to int.
- The RHS matrix still needs padding on the X dimension. Hence, revert
few changes to add the necessary padding elements.
- Replace zero padding validation tests with assertion in the configure
method of the kernel.
Change-Id: Ib6614a91bd0e98f2b850f52eef14d4fbf55517c8
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4259
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
|
|
Change-Id: Ifcb4bfb357f342c8f45665e089d22677e6f8fe36
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4275
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
Change-Id: Ie70ba877f0356661a055f026124904bbf2181a33
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4251
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-by: SiCong Li <sicong.li@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
Signed-off-by: Sheri Zhang <sheri.zhang@arm.com>
Change-Id: I0a3f0a989a4db9a2abc9c89429e94af4c6b6b366
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4274
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
Remove padding from CLWidthConcatenate2TensorsKernel
Remove padding from CLWidthConcatenate4TensorsKernel
Change-Id: I2142618e87bf11f831fe3b9375c4a7efda8d3a21
Signed-off-by: Sheri Zhang <sheri.zhang@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4266
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
|
|
Signed-off-by: Sheri Zhang <sheri.zhang@arm.com>
Change-Id: Iee1d4655012ce4cb699535697aeefec673f0bc63
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4157
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
Signed-off-by: Sheri Zhang <sheri.zhang@arm.com>
Change-Id: I705044a9429bb9a08268368b09463c2af85616d5
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4253
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Manuel Bottini <manuel.bottini@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
Change-Id: I1a82315dfa9b021f72e0b687da658e2e02c9bb34
Signed-off-by: Giorgio Arena <giorgio.arena@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4257
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
- Add has_pad_y flag in GEMMKernelInfo
- Skip reinterpret as 3D in CLGEMMMatrixMultiplyReshapedOnlyRHSKernel if
has_pad_y = false
- Add test to validate CLGEMMMatrixMultiplyReshapedOnlyRHSkernel with
had_pad_y = false/true
- Configure two variants of CLGEMMMatrixMultiplyReshapedOnlyRHSKernel to
run with has_pad_y = false/true in CLGEMM
- Check if the lhs/dst tensors have pad y. If not, run
CLGEMMMatrixMultiplyReshapedOnlyRHSKernel without padding requirement
Change-Id: I68bb43389789736d676b899ac7c77fd9138babaf
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4248
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
- Refactor the OpenCL kernels for Winograd output transform NHWC to
avoid padding requirement
- The kernel adopt the reverse store approach to avoid out-of-bound
writes
Change-Id: If9aad20354ff2146f57ead07ba0aaadb3df919f9
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4222
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
Signed-off-by: Giorgio Arena <giorgio.arena@arm.com>
Change-Id: I004128fdcc1207c25d2b959f17f04f9e1a8b4cb5
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4247
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
|
|
Signed-off-by: Giorgio Arena <giorgio.arena@arm.com>
Change-Id: I448f2b52e2089da1b859301eade7b988d816a816
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4245
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
Change-Id: Id68d1d02a05025098bc87e080909e9efbf2e7820
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4238
Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
|
|
* Remove the user-supplied padding from CLCopyKernel
Note that this padding is different from the internal "padding" in the
original task, as it is user-supplied instead of internal.
This user-supplied padding interface is removed simply because it has
been replaced by a more capable CLPadLayerKernel, and is not used
anywhere else.
Signed-off-by: SiCong Li <sicong.li@arm.com>
Change-Id: Ib53e76efd7d043ee79dcd47ca734c6dc685da43e
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4194
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
Signed-off-by: Giorgio Arena <giorgio.arena@arm.com>
Change-Id: Ic43aba8a6a0a106fc4c1f665ff5cc3ccb31f403d
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4235
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
|
|
* Remove padding only for when user-supplied padding is empty
* Vectorize the case where output_window is not null and the output
window is narrow in x (smaller than vec_size_x)
Change-Id: I313089fe309e87e8529ecfd00542fcfa4dc44862
Signed-off-by: SiCong Li <sicong.li@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4193
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
The kernel was not using the preprocessor arguments needed avoiding the
use of padding.
Change-Id: I6b5fdf4f3f14edbef60b9d5b60179d619700bc00
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4232
Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
|
|
In gemmlowp_matrix_b_reduction kernel the accumulator data type might be
set to uint if the input data type is unsigned quantized. However, the
output of this kernel is always a signed integer, hence we need to
convert the result before storing in memory.
Change-Id: I9b936fbbcb8cd64319c42872648f5058f686b228
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4233
Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
|
|
COMPMID-3723: Remove OpenCL padding: CLGEMMLowpOffsetContributionOutputStageKernel
Change-Id: Iac265c2ac4c5749352daa311279a3b8c60ac3b3d
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4228
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
Added utility functions developed by Giorgio for checking that padding
remains unchanged after configure.
Change-Id: I6862e74baf9b8792991e3f25e176c672c0a46836
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4208
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
Signed-off-by: Giorgio Arena <giorgio.arena@arm.com>
Change-Id: I5f77356bff6c6ab513ed3555466c8c5bf5f4c4e3
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4227
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
|
|
FP16/32
Removed unused N from partial block loading macro
Created utility to assert change in padding
Signed-off-by: Giorgio Arena <giorgio.arena@arm.com>
Change-Id: Ifdd30c66dbf5f2842c6b2d939000613d5011708e
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4192
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
Guard the kernel with all required compile-time arguments, otherwise
the kernel might be wrongly included when compiling for other kernels
which don't have the required compile-time arguments, resulting in
mysterious kernel build errors.
Change-Id: Ib45b46a5ab14e6dc6a415c0466cf9a5963452364
Signed-off-by: SiCong Li <sicong.li@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4224
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
|
|
Change-Id: I8a685d4ac5de747a0f775bd10be9c411cf394953
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4140
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
Reviewed-by: Sang-Hoon Park <sang-hoon.park@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
|
|
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Change-Id: If9d6fa8c900b68c4b6fd373f2fc1f9abb83ea917
Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4145
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Sang-Hoon Park <sang-hoon.park@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
- Remove padding requirement from the OpenCL kernels
- Extend test to validate zero padding requirement
Change-Id: I1ddf04eba783721858792efb08a2c97f11f7297e
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4206
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
- Refactor pooling layer kernels on OpenCL (F32/F16/QASYMM8) to avoid
padding and improve performance
- Add test for checking zero padding requirement
- Fix issue with extracting the index. The issue was caused by the
padding passed at compile time
- auto_init indices tensor in CLPoolingLayerKernel
Change-Id: I1ae5a2ef8c4ce787c80dcd73e35c17bb34623cb5
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4188
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
CLGEMMLowpQuantizeDownInt32ScaleByFloatKernel
COMPMID-3725: Remove OpenCL padding: CLGEMMLowpQuantizeDownInt32ScaleKernel
Change-Id: Idea5974a56861efae3bc255f1224c7f1e88f3650
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4182
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
Change-Id: Iee28abcbba1e7b9e2f3aaa55685936dce815d5a3
Signed-off-by: morgolock <pablo.tello@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4141
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
remove padding from related OpenCL kernels
Change-Id: I0b0be8fcccf511c7214e83ba6aa8d0e901bc4f3c
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4146
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
Signed-off-by: Sheri Zhang <sheri.zhang@arm.com>
Change-Id: If077a245156be69f34834cbfbd0a36e570ee4149
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4107
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
|
|
Change-Id: I09f557b5cecafc669e12764e8592457212168d62
Signed-off-by: Giorgio Arena <giorgio.arena@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4131
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
|
|
COMPMID-3709 Remove OpenCL padding: CLDepthConcatenateLayerKernel
Signed-off-by: Giorgio Arena <giorgio.arena@arm.com>
Change-Id: Iaea4fafd5d0f081fd5b45b0f6945302dc3365bd9
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4105
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
|
|
macro
Change-Id: I73edadc7299247e7bc51ae37c00d3709023da44a
Signed-off-by: Giorgio Arena <giorgio.arena@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4073
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
- Update the heuristic (m==1) for CLGEMMReshapedOnlyRHS
Change-Id: I216c158f2802d3d331e23e0d9eb0127107ec8af0
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4092
Reviewed-by: Giorgio Arena <giorgio.arena@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Georgios Pinitas <georgios.pinitas@arm.com>
|
|
- Update heuristic for CLGEMMReshapedKernel - FP16
- Update heuristic for CLGEMMReshapedOnlyRHSKernel - FP16
Change-Id: I35aa73e59d8c2d1bc6b2dd318fd8eeb3e42c27a4
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4026
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: SiCong Li <sicong.li@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
- The change affects Mali-G71 GPUs and should improve the performance of
GEMM in case of m = 1
Change-Id: I6b0e217e93fe468ec1325a5da74684811519c42f
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4002
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: SiCong Li <sicong.li@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
CLGEMMMatrixMultiplyReshapedKernel
Resolves: COMPMID-3671, COMPMID-3672
- Extend cl image support to f16 in CLGEMMMatrixMultiplyReshapedKernel
- Extend cl image support to f16 in CLGEMMMatrixMultiplyReshapedOnlyRHSKernel
- Change the interface of create_image2d_from_buffer
- Extend test
Change-Id: I27363be71fa515fbf71aa4be5ed0d6c730f38f34
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3992
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
Treat bf16 memory on memset as raw memory by casting to void*. This
hides the class-memaccess warning and is safe for the current class
layout of arm_compute::bfloat16
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Change-Id: I5e242827d3737b4491d29abe7570eefee5b6edc1
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3928
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
|
|
Change-Id: Ic0569fe9ed99e61084b601ce84ddc7ef288d1899
Signed-off-by: SiCong Li <sicong.li@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3852
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
* Properly support "axis" in CL and NEON (and GC) SoftmaxLayer and
LogSoftmaxLayer in accord with mainstream frameworks. Axis now defines
the dimension on which softmax is performed, and supports the range
[-rank, rank)
* Extend validation tests to include valid and invalid axes
* Remove unnecessary LogSoftmaxLayer fixture, as it is only a
specialisation of the SoftmaxLayer fixture
* Change the validation fill value range from [-1000, 1000] to [-10,
10], as the former often results in sparse outputs with a single one and
zeros elsewhere
Change-Id: I8a0040453182b04ed88260de3ba434e98258d863
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3830
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
|
|
The issue concerned gemmlowp_mm_reshaped_only_rhs_t_fused_output_stage_fixedpoint.
In particular the issue was with the z index to access the elements from
the lhs reduced tensor used to calculate the offset contribution.
Change-Id: I74f6398fc08894fc323ccd04fda9220752652d31
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3726
Reviewed-by: TeresaARM <teresa.charlinreyes@arm.com>
Reviewed-by: Sang-Hoon Park <sang-hoon.park@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
|
|
CLGEMMMatrixMultiplyReshapedKernel
- Change the interface of STORE_BLOCK_BOUNDARY_AWARE passing the
conditions on Y and X rather than the X/ coordinates. This allows to
use the macro with both GEMM reshaped and GEMM reshaped rhs only
- Remove padding from the output tensor of
CLGEMMMatrixMultiplyReshapedKernel
- Add tests for validating the zero padding requirement
Change-Id: I13263cc71ce065c5be34ed198def320dd5823495
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3712
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: SiCong Li <sicong.li@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
Change-Id: I3f85ece08c9fc4bdbbfd72b5a872d4f2c4b76357
Signed-off-by: SiCong Li <sicong.li@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3709
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
- Remove padding requirement for the input tensor of
CLGEMMReshapeLHSMatrixKernel
- Add utility function to load a boundary aware 2d tensor from buffer
- Extend validation for validating the zero padding requirement
Change-Id: I0ac6b1b517d75fd56998f406e0cce97b40918ce1
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3701
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: SiCong Li <sicong.li@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
|
|
* Remove channel paddings from all nhwc kernels (im2col_3x3_nhwc,
im2col_9x9_nhwc, im2col_generic_nhwc)
* Validate that input total spatial dimensions (with x and y paddings)
are bigger than or equal to the kernel spatial dimension.
- Otherwise it would result in invalid memory reads.
- This problem likely existed before, but was made obvious with the
removal of implicit paddings
* Add zero padding validation tests
* Fix Im2ColValidationFixture by not permuting the input shape in case of
NHWC
Change-Id: I1f895e8938af0e9130cb516106f0b4b665531709
Signed-off-by: SiCong Li <sicong.li@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3696
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
|
|
For the elements that shouldn't contribute to the sum, zero
is used to compute the correct sum.
Change-Id: I5360534b5b0f81ee3d3aaaf5a046b99ecd943894
Signed-off-by: Sang-Hoon Park <sang-hoon.park@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3703
Reviewed-by: Pablo Marquez <pablo.tello@arm.com>
Reviewed-by: TeresaARM <teresa.charlinreyes@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
|