review.tizen.org Git - platform/upstream/armcl.git/log

Fix packaging error

This fixes:

=== the following packages failed to build due to missing build dependencies (1) ===
nnstreamer:
  have choice for libarm_compute.so()(64bit) needed by nnfw-devel: libarmcl-devel libarmcl-release
  have choice for libarm_compute_core.so()(64bit) needed by nnfw-devel: libarmcl-devel libarmcl-release
  have choice for libarm_compute.so()(64bit) needed by nnfw: libarmcl-devel libarmcl-release
  have choice for libarm_compute_core.so()(64bit) needed by nnfw: libarmcl-devel libarmcl-release

by
- Make files exclusively installed between subpackages (release and devel)
- Add dependency between relead and devel.

Change-Id: I1b738124c9d62c663e99316eba68cbcda7cd033f
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>

packaging: add gbs build support for Tizen

Change-Id: I961315e7bb17e071deae469fc26b3a151eb1b0e4
Signed-off-by: Inki Dae <inki.dae@samsung.com>

move ndrange.hpp header into arm_compute directory

ndrange.hpp file is needed by arm_gemm_compute_iface.hpp when building
armnn.
In a cross compilation environment like Yocto Openembedded, the installed
include files comes from either /arm_compute, /include or /support
directories.
So to avoid armnn build failure, ndrange.hpp need to be moved in
arm_compute/core/NEON/kernels/arm_gemm directory.

Other files that need ndrange.hpp to build are updated accordingly.

Signed-off-by: Vincent ABRIOU <vincent.abriou@st.com>
Change-Id: Icc8e18b23dd06c56e0c49a84d4a51bfd85fe2290
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3271
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>

COMPMID-3508: issues with the document in the 20.05ACL RC

Generated documentation is no longer part of the release on the master
branch. It will only be part of gh-pages.

Change-Id: I326d7f3d0653f880d919cfebd70ccb4b67bd5e09
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3272
Reviewed-by: Manuel Bottini <manuel.bottini@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

Update release_repository pin - p2

Change-Id: Ia7ddf4527e0a44401b13328834b1800de8730dfe
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>

Update release_repository pin

Change-Id: I49ed0a376201767ee9c869e16dcfced364111bb7
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>

COMPMID-3424: Update Release Notes

Change-Id: I59c24900a62d8ec27f99e9fd3a5c1955b6d98783
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3250
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Tested-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3499: Fix integer overflow for large GEMM on NEON

Change-Id: Id9eef3abc8a902b52ba61772f716f2ba2b97f7d4
Signed-off-by: Sang-Hoon Park <sang-hoon.park@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3245
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3423: Update functions list in Doxygen

Change-Id: Ied6352783f2413ed817d80dcadf502169f84d3bf
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3252
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3069: Removing deprecated functions and classes from 20.05 release

Change-Id: Ic4d20995d6c6bb76d07113e86247bad2722e4e83
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3244
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

Remove blocks of codes / files which are marked as INTERNAL_ONLY and Disable validation and benchmark tests - p2

Change-Id: I9c458f87e34bcb2fa509fedfdc1654888d009be3
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>

Update SONAME_VERSION in SConscript

Change-Id: I3219426e2978555883f97f7788341a2df6c0e532
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>

COMPMID-3069: Fix CLQLSTM configure argument order

Change-Id: I5f56d65df38e9ac035a27dd2b15925c576cc3932
Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3242
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

Update release version in Doxygen

Change-Id: I8d18c1117cb6b3948c6699331c1cb68408cc123f
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>

Update Release VERSION in SConscript

Change-Id: I5983972af6e0268673fa3ab6faedc3b48ad1c7ca
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>

Remove blocks of codes / files which are marked as INTERNAL_ONLY and Disable validation and benchmark tests

Change-Id: Ic45589643bebfb6db748d25db5bca48def297193
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>

Remove 3dparty folder and submodule

Change-Id: Ia196f38a575b84c64ed1d5d1c2f727146aea4068
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>

COMPMID-3491: ReduceMean not properly validating quantizations

Change-Id: Ia2563da7271592c4c55dde7fcecaedda6a46a781
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3236
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3490 - Nightly fails: android armv7 microtask error - part2

Change-Id: I6555728a9d39485c4d9fb9b75024744f3454c8e5
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3234
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3497: Fix build LSTMParams<TensorInfo>

Copy of missing parameters is added to the build function.

Change-Id: I06d665729a61f2f01dcfab21c6e6d55359fc81cb
Signed-off-by: Sang-Hoon Park <sang-hoon.park@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3226
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-by: James Conroy <james.conroy@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3490 - Nightly fails: android armv7 microtask error

- Removing OpenMP from failing references

Change-Id: I3a90347dccf55fa22893b42b5236e45f15da19e3
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3219
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3289: Test improvement CLGEMMMatrixMultiplyReshapedKernel.

Change-Id: Ia6c2f115849889baceafaf716477456e41f96037
Signed-off-by: morgolock <pablo.tello@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3186
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Sang-Hoon Park <sang-hoon.park@arm.com>

COMPMID-3457: vexp failure in QASYMM8_SIGNED overflowing

Change-Id: Ied11a4a3e9d04615a1a1f0bfa552f3dd8293a170
Signed-off-by: morgolock <pablo.tello@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3178
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3069: Align graph convolution implementation for CL and NEON.

Enables fast-math on Neon backend for convolution

Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Change-Id: Ia072f0fd2db1f0814562049b290cffc91cbbd9a8
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3201
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3295: Static input for ActivationLayer test suite

- A member function added to AssetsLibrary to fill tensors with
static values.
- ActivationLayerFixture has been modified use the new function.
- Redundant nightly tests are removed.

Change-Id: Ib2a1103a1e438e808183170dc9d097599523c6ec
Signed-off-by: Sang-Hoon Park <sang-hoon.park@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3188
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3290: Test improvement for CLGEMMMatrixMultiplyReshapedOnlyRHSKernel

Signed-off-by: Sheri Zhang <sheri.zhang@arm.com>
Change-Id: I7335ee07f777087e06ca26f762b2b5e3668362ab
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3175
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Sang-Hoon Park <sang-hoon.park@arm.com>

COMPMID-3483: Refactor NEDepthwiseConvolutionLayerNativeKernel

Removed is_per_channel template arguments since it wasn't used
anywhere and also made has_biases a runtime parameter. The total
size reduction from this change is 28.6kb.

Change-Id: I292ac27ae3ea2885b8438f613390486323982664
Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3189
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3069: Fix CL nightly builds due to unused variable

Change-Id: I97c51b957817bf2699900f5947a0d1bc3f0bbf76
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3192
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3484: Regression in Transpose convolution Android R CTS test.

- Removing _flip_axis from _memory_group management

Change-Id: Ica3039fa7b545b34de231bcd7620d76914681808
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3191
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3128: Test improvement for GEMMConvolutionLayer on CL and NEON(Function-level)

Cleaning up GEMMConvolutionLayer tests by doing the following:

- Remove unnecessary configuration tests
- Remove redundant tests
> Redundant shapes
> For large shapes there are already tests for each internal kernel/function
- Test NHWC 1x1 kernel 1x1 stride to stress _skip_im2col
- Stimulate gemm3d to skip col2im
- Test asymmetric padding
- Test batch size equal to one and different than one
- Test fully connected convolution
- Test with a few different padding values
- Test 1D kernel
- Test with FLOOR rounding policy

Change-Id: I88e7009b8e9c991994ed264476c16a79a0de4a68
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3150
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Sang-Hoon Park <sang-hoon.park@arm.com>

Introduce CL kernel cache feature

This patch introduces CL kernel cache feature for
NN runtimes such as ARMNN or other in-house versions.

Originally, this feature came from utils/Utils.cpp and utils/Utils.h
files but was used just for testing.

Caching CL kernel binaries is mandatory and required for real product
because inference performance at first run is one of very important
factors. This patch exposes two API - save_program_cache_to_file
and restore_program_cache_from_file so that NN runtimes can use
these API to cache the compiled CL kernel binaries.

Changelog v1.
- Reordered including headers suggested by Manuel Bottini.
- Updated description.
Changelog v2.
- Drop function descriptions.
Changelog v3.
- Fix build error with opencl = 0. As for this, it moves two functions,
  save_program_cache_to_file and restore_program_cache_from_file,
  to src/runtime/CL/Utils.cpp and arm_compute/runtime/CL/Utils.h,
  and set ARM_COMPUTE_CL flag only when opencl = 1.
  This change makes CL kernel cache feature to depend on OpenCL support.
Changelog v4.
- Fix coding_standard_checks error.
Changelog v5.
- Add a new cpp file to Android.bp to fix coding_standard_checks error.
Changelog v6.
- rebased on top of master, and dropped _new_built_programs_map object
  so it adds compiled CL kerne binaries per model to a given file.
Changelog v7.
- Correct descriptions.
- Correct arm_compute/runtime/CL/Utils.h file description.
- Correct src/runtime/CL/Utils.cpp file description.
- Drop ARM_COMPUTE_CL ifdef ~ endif from examples/cl_cache.cpp
- Do not add 'ARM_COMPUTE_CL' flag at SConscript.

Change-Id: I8c640fb1809ca79191daf5b1ff72a50dd492c36d
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/2914
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3108: Add Winograd 3x3,4x4 FP16 support for NEON

Change-Id: I20680dc74a3d709297539e2132417308a7aecc9d
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3159
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3463 Refactor NECropKernel

Removed most of the templates used in this kernel. This
resulted in a reduction of 35Kb.

Change-Id: I0d50aa769b361790d47a8017d795f2c92c6d8a6f
Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3158
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Sang-Hoon Park <sang-hoon.park@arm.com>

COMPMID-3464: Address NESoftmaxLayer failures for QASYMM8_SIGNED

Normalization with the maximum value was causing results to wrap-around
As a work-around we use saturating intrinsics to perform the operation

Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Change-Id: I719b7ac7ad274dc2ae339bc4a055f9200134ed97
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3184
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Sang-Hoon Park <sang-hoon.park@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3439 Fix CLQLSTMLayer

* Apply conditional data type check to CLGEMMLowpMatrixMultiplyNativeKernel

Signed-off-by: SiCong Li <sicong.li@arm.com>
Change-Id: Ife9f85ec91ca38d079b9b4e045fbd4585dd1c8dd
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3180
Reviewed-by: Sang-Hoon Park <sang-hoon.park@arm.com>
Reviewed-by: Sheri Zhang <sheri.zhang@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3439: Fix peephole and projection in CLQLSTMLayer

The followings are essential to make it work

- QSYMM16 is added as supported data type in CLGEMMLowpOutputStage
- Internal TensorCopyKernel is added similar to NEQLSTMLayer

The followings are fix for related things.

- Projection is modified to remove copy of projection_bias from
NEQLSTMLayer.
- Fix wrong argument for validate_mm()
- validate_mm() now returns on error.

Change-Id: Icbd04e9fdb8821eb41dd3e0a6a0980965b779714
Signed-off-by: Sang-Hoon Park <sang-hoon.park@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3177
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>

COMPMID-3442: Add support of negative axis in NESoftmaxLayer and reference code

Signed-off-by: Sheri Zhang <sheri.zhang@arm.com>
Change-Id: I285cc3b74ac0a45f0ad5830baed5237cea568f15
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3147
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3286: Test improvement for CLGEMMMatrixMultiplyKernel

Change-Id: Idef4065ae98b1f41bd38d1fdef3c5f7eff45d940
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3169
Reviewed-by: Sang-Hoon Park <sang-hoon.park@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3470: Modify NE/CLQLSTMLayer interface to provide 3 outputs

Change-Id: I895b697c89c9a7509d48a54ac1effb7fbd8cca19
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3174
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Sang-Hoon Park <sang-hoon.park@arm.com>

IVGCVSW-4744 CTS: RandomGraphTest LargeGraph failures in CpuAcc HAL 1.2 and 1.3

* Retain data layout for SpaceToBatchLayerKernel and BatchToSpaceLayerKernel

Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: I5bd0c3ca90a4964eda9f838c97c784aced0f4579
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3166
Reviewed-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3316: NEDeconvolutionLayer failing for a big input

- Using NEDirectConvolution for big shapes since the memory
required explodes for 9x9 kernel
- Adding test cases
- Fix enables only the NEON Deconvolution for NHWC

Change-Id: I8a541346428e5686818f8ecb7f69e2a9106cbceb
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3135
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Sang-Hoon Park <sang-hoon.park@arm.com>

COMPMID-3467: Zero-initialize std::array in ndrange.hpp

Fix uninitialized variable warning.

Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Change-Id: Ifeabc68e2ed3bc35b6cb68890716cceea6b519db
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3170
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3239: Fix projection and peephole in NEQLSTMLayer

- Peephole and projection has been fixed to be working
- Small internal kernel copying data between tensors to
  cover the case where num_units and output_size is different
  is added.

Below is strictly outside of this patch's scope but are changes
helping this patch working (directly or indirectly) or making
NEQLSTM more complete.

- Consideration for layer normalization is added to InfoHelpers
- QSYMM8 data type is added to helper function to
  print out tensors.
- NE/CLLSTMLayer::validate() logic has been modified to use correct
  value for shape validation.

Change-Id: I40b4e71dfdbe8432caa2fe4a9af60a725362cc33
Signed-off-by: Sang-Hoon Park <sang-hoon.park@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3157
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

[ONCPUML-7] arm_compute support for ND parallelism

Currently 1D ranges of work are specified by the scheduler
via two integers, start and end. This limit opportunities
for advance parallelism and scheduling

This patch expands the interfaces to allow for ND parallism.

`GemmCommon::get_window_size` now returns an `NDRange` specifying the work
in N-dimensions rather than with the single integer it used prior (1D)

Execute now takes an `NDCoordinate` which specifies an `NDRange` with a start
position for that work along with an `NDCoordinate` to specify the thread location

In addition to expanding the interface to enable this functionality,
we have added the capability to SGEMM when the number of threads is high
this has the effective of allowing a much greater degree of parallelism
where te problem dimension would previously have limited the number of threads.

Change-Id: I3e1a8b7276216627bec4ff6f24ac2147552ea9fb
Signed-off-by: Joseph Dobson <joseph.dobson@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/2962
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3459: Updated heuristic in CLGEMMReshapedOnlyRHSKernel

Change-Id: Ib958649d0193c148f45d72404da21d3f02dc5183
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3154
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>

COMPMID-3459: Update the heuristic for GEMV for F16 on Mali-G77

Change-Id: I10020cd5c88ec1926087531c4c4ce0a1cfaf43c9
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3151
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3069: Update gcc version to 6.3.1 and NDK to r18b

Change-Id: I109e9322ee153fb65ca4402d858bb4d9328198d2
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3114
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>

COMPMID-3434 Add SIMD support in gemmlowp_mm_native

* Add SIMD support in gemmlowp_mm_native for Midgard arch
* Extend the validation test to include the corner case of m0=1, n0=1 and
k0=1

Change-Id: I1f5e384b206a20e974932b8aa10cd628c69e5efc
Signed-off-by: SiCong Li <sicong.li@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3142
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>

COMPMID-3443: Android R CTS 1.3 QASYMM8_SIGNED failure with MEAN on CpuAcc

- Properly perform the division of the sum of the elements by the interested dimension of the tensor

Change-Id: I0a30be4e5e6cfc1bd24a17eb4307dd2acf98db3e
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3145
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>

COMPMID-3428: Fixed Mali-G77 heuristic for GEMMReshapedOnlyRHS - f16

Change-Id: Ie1f021235edf3d60171a252d519c53df2d7fb0c6
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3146
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3288: Test improvement for CLGEMMReshapeRHSMatrixKernel (Kernel-level)

Removed configuration tests as they don't add any real value.
Since this is a memory copy kernel, I removed all the large shapes
and created all possible configurations across the 3 data layouts.
From a total of 5957 nightly and 2285 precommit tests now we have
233 in both cases.

Change-Id: I219f771a0ee5b07e5de3d5df715fe5841b4ea5fe
Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com>
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/c/VisualCompute/ComputeLibrary/+/237083
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3106
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Sang-Hoon Park <sang-hoon.park@arm.com>

COMPMID-3444: Fix CpuAcc Hal 1.3 PRelu failure issue
Fix CpuAcc Hal 1.3 Max & Min failure issue

Signed-off-by: Sheri Zhang <sheri.zhang@arm.com>
Change-Id: Ib519e6c6717e6608106aebc8c1cd751a731428a3
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3139
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

MLCE-166: Add support for extracting indices in NEPoolingLayer 2x2 NHWC

* Added support for pooling indices in NHWC Poolsize 2x2

Change-Id: Ib2a3468e794f58bbf2c03aba9f6b184b9d76b183
Signed-off-by: morgolock <pablo.tello@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/2997
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>

COMPMID-3441: Nightly failed due to NEScale QASYMM8_SIGNED output wrong result

Change-Id: I4a97523408f44c509ba0f6bf180c56ea33b964cf
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3138
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Sheri Zhang <sheri.zhang@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3379: Add support for QASYMM8_SIGNED in NEScale
Update QASYMM8_SIGNED support document

Signed-off-by: Sheri Zhang <sheri.zhang@arm.com>
Change-Id: Ia0cb0d2eed13e7e38b165dc317af06e7f2df4285
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3132
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>

IVGCVSW-4743: Fix CpuAcc Hal 1.3 Comparison Failures

Broadcast for QASYMM8_SIGNED was not handled.

Change-Id: Id5dbb0dce78838319218de94551bba52d697f4a4
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3131
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3069: Validate bias dimensionality in NEFullyConnectedLayer

Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Change-Id: Ib5add7951417806f89ce9445f8cdd723488193c3
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3119
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3069: Allow different quantization info in NEElementwiseMin/Max

Change-Id: I97f9f7e9f054b2a812a23b25cfb21033f76a9101
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3125
Reviewed-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

[ONCPUML-7] Improvement to Window::split_window

If the total passed to split window did not fit
nicely into the selected Dimensions size then
the size of of the window return will vary considerably
for different ids

This change means that the amount of work each id
will vary by the minimal amount.

For example:
If total was 10 and a Dimensions size was 19

With then with the old code :
* id 0 - 8 would get back 1,
* id 9 would get 10

With the new code:
* id 0-8 would get 2
* id 9 would get 1

Change-Id: I6b74b81d7ddcea06db7aa9fbaf8cb47a659994c1
Signed-off-by: Joseph Dobson <joseph.dobson@arm.com>
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/c/VisualCompute/ComputeLibrary/+/224448
Tested-by: bsgcomp <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: bsgcomp <bsgcomp@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/2961
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3287: Test improvement for CLGEMMReshapeLHSMatrixKernel (Kernel-level)

Removed configuration tests as they don't add any real value.
Since this is a memory copy kernel, I removed all the large shapes
and created all possible configurations across the 3 data layouts.
From a total of 21329 nightly and 1169 precommit tests now we have
545 in both cases with a larger test space.

Change-Id: I363caab64654064d2779a0ba847f522dfe388091
Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3118
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3294: Add AssetLibrary instance using a fixed seed

- Add fixed_seed instance for future use
- Make Copy fixture use the new instance to validate functionality

Change-Id: I7e8c1504b8db3203213d084b300f10ecd384f7fa
Signed-off-by: Sang-Hoon Park <sang-hoon.park@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3107
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3407: HAL 1.3 Driver TransposeConv2D VTS Failure

Upsample wasn't done with the correct zero value.

Change-Id: Icf55c0584342979ec4277a80832d29954f5f960c
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3115
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3292: Clean tests for Im2ColKernel.

Change-Id: I005ac542a94aeed8df806699d3d237ac8973d1e6
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3003
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Sang-Hoon Park <sang-hoon.park@arm.com>

COMPMID-3291: Test improvement for CLWeightsReshapeKernel

Signed-off-by: Sheri Zhang <sheri.zhang@arm.com>
Change-Id: I9956a3bdfa4ddbb205111ff2524ff40cf57644e0
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3052
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3240: Add support for layer normalization to CLQLSTMLayer

Signed-off-by: Sheri Zhang <sheri.zhang@arm.com>
Change-Id: I45359a4ddb46c059097a2d77c008f802e8f4c143
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3065
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-by: Sang-Hoon Park <sang-hoon.park@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3283: Update documentation for Android R operator support

Signed-off-by: Sheri Zhang <sheri.zhang@arm.com>
Change-Id: Iee5fb21e907b2807a5f8d4d40a7056d51e3a7b65
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3097
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3329: failures in nightly (GC/GEMM and CPP/DFT)

- Can't recreate GC/GEMM
- Fix on CPP/DFT/DFT2D/Complex using the appropriate tolerance
- Fix CPP/DFT/Conv/Real2Real by removing concurrency issue on faulty function

Looking at the proper fix for CPP/DFT/Conv/Real2Real since it is a concurrency issue

Change-Id: Ic846dc7925d4c1b47e14181656c905f90778cdeb
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3085
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>

COMPMID-3249: Remove non-public data submodule and mirror Github behaviour

This patch removes the `data` submodule and replaces it with the Github
version of the data folder.

Change-Id: I541880d7a823eb6eecdde8ee4fab28596efee02b
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3094
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3280: Make all ML primitives for CL use the new interface - Part 2

- CLFunctions have been updated

Change-Id: Ie3256a6c775bc12f3126482bd8e8a46da54b267c
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3053
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3394: Replace get_cl_type_from_data_type in All

Signed-off-by: Sheri Zhang <sheri.zhang@arm.com>
Change-Id: I978050182817c964779c775cdefd88d2c7df0ca5
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3069
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>

COMPMID-3247: Cleanup README file links

Removed release binaries links. The README now always points to the
GitHub Releases page.

Moved all documentation links to the GitHub Wiki and added a link to
that page in the README.

Change-Id: I9922cc74b14b4192766302f825b1b2af4aa02523
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3070
Reviewed-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3405: Fixed issue in gemmlowp_mm_native - OpenCL

The issue was related to the creation of the transpose macro
with N0 = 1

Change-Id: I1006bee583b99302d29346ffe7a54361f32d1ede
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3086
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3406: Fix build issue of gcc <= 5.4

Use double pairs of curly brackets to initizalie std::array.
(Single pair of brackets doesn't cause issue with gcc >= 6.3)

Change-Id: I1e49e92f44ffd5f359cce804d6c2c899b2f10f66
Signed-off-by: Sang-Hoon Park <sang-hoon.park@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3082
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3069: Requantize in int32_t in RELUs of NEActivationLayerKernel

Requantization must be in 32 bits as there is a multiplication by
scaling factor and an offset addition. In this case, clamp was
interpreting the input as an 8-bit value, therefore cutting the
remaining 24 bits.

Change-Id: I31a5837a94e229201a023b1617fc2df5c35f61e3
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3076
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3280: Make all ML primitives for CL use the new interface - Part1 - Fix3

- const fix in the CLKernels part 3

Change-Id: I9cfb896f334145249a97c9287fa00399b8319a8e
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3075
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3280: Make all ML primitives for CL use the new interface - Part1 - Fix2

- const fix in the CLKernels part 2

Change-Id: Ia12845e291b4137cbaf76eb8438e381c4fd0368a
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3071
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3241: Add Layer Normalization to NEQLSTMLayer

- Add output quantization calculation to Layer Normalization
- Add members for Layer Normalization to NEQLSTMLayer
- Add configure/validate/run of Layer Normalization to NEQLSTMLayer

Change-Id: I278c8e0edbb21212f3afa4d4a336df0f1a4c1bfb
Signed-off-by: Sang-Hoon Park <sang-hoon.park@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3059
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3280: Make all ML primitives for CL use the new interface - Part1 - Fix1

- const fix in the CLKernels part 1

Change-Id: I17340cb6ff26afd52b14b46645efedbe07aef1b6
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3067
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3069: Fix QASYMM8_SIGNED RELUs in NEActivationLayerKernel

Wrong boundaries were used for clamping the output when data type was QASYMM8_SIGNED.

Change-Id: Ie93bf64643fce95be2b4a5a63ef2d1dc83bbdd97
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3064
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3236: Implement CLQLSTMLayer

COMPMID-3081: Extend CLQLSTMLayer with enhancements

Change-Id: Idb7aaaacdba957e5ad61e94edeab2e898730a109
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3057
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Sang-Hoon Park <sang-hoon.park@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID: Fixed issue in the scheduler of NEConcatenateLayer

Change-Id: Ibee18503ae94da43811e8ad73760abf111c34ff7
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3063
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3241: Fix hidden scale in NEQLSTMLayer

- Fix wrong data types in LSTMParams
- Add logic to ignore epsilon for quantization multiplier computation
- Ignore epsilon for hidden gate scale computation

Change-Id: Ia0b2f523b1c2ad325f3523439a8eea051d81958c
Signed-off-by: Sang-Hoon Park <sang-hoon.park@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3058
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3304: Update OpenCL GEMM heuristic for Int8

Change-Id: I6b7ff678d8d0437a1639db2ff602ea1cdb155464
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3056
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3237: Implement NEQLSTMLayer

COMPMID-3082: Extend NEQLSTMLayer with enhancements

Change-Id: I88175b7bf69494a4eae510b74176fe8a0d6cd770
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/2969
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Sang-Hoon Park <sang-hoon.park@arm.com>
Reviewed-by: Sheri Zhang <sheri.zhang@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3314: Enable OpenMP in the reference tests

Change-Id: I05b5fedb998317144e0dd13a6377a97207b27f46
Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3024
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3280: Make all ML primitives for CL use the new interface - Part 1

- Only CLKernels have been updated

Change-Id: Ife55b847c2e39e712a186eb6ca452503d5b66937
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3001
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3378: CLScale failure

Avoid creating tests for unsupported cases

Change-Id: Ida0835af2f3d83c39544e930f8f5be3d4471fa38
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3022
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Sang-Hoon Park <sang-hoon.park@arm.com>

Add uint8 quantization example for NEON

This example performs a simple matrix multiply operation.
It includes both full precision and low precision GEMMs.
Uses square matrices with known values by default (for easy vaidation).
Uses random values when user specifies matrix sizes.

Signed-off-by: Diana Bite <diana.bite@arm.com>
Change-Id: I269d1e20dbd231c48667751e71758b2e90ec72a2
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/c/VisualCompute/ComputeLibrary/+/223887
Tested-by: bsgcomp <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/2924
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Manuel Bottini <manuel.bottini@arm.com>

COMPMID-3236: Add support QSYMM16 into S32 CLPixelwiseMultiplicationKernel

Change-Id: Ifc519f53f04fcb14ddb9c17f98cc687f34285c97
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3018
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3239: Implement QSYMM16 LayerNormalizationKernel for CL
Use NE/CLSynthetizeFunction instead of NE/CLQLSTMLayerNormalizationValidationFixture

Signed-off-by: Sheri Zhang <sheri.zhang@arm.com>
Change-Id: I62ace213a5261f2d307da6953d0521492aa05292
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3019
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3293: Test Improvements for CLCol2ImKernel

Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Change-Id: I086734a0c6d66cb782efb39ebd3e14a2809b3685
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3004
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

COMPMID-3348: Fix issue in gemmlowp_matrix_b_reduction OpenCL kernel

Change-Id: I67be39a9f60d20694dda970356c94cb6d90445ea
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3016
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>