Age | Commit message (Collapse) | Author |
|
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: Iaaff33f578c4d88f6c03c8de4af71d5347178eaa
|
|
* Adds Reference FP16 support and unit tests for layers not already supported
!referencetests:202156
Signed-off-by: Matthew Jackson <matthew.jackson@arm.com>
Change-Id: I6fc9b9ce2809e163f72e27e877025c8fb85d9fbe
|
|
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: Ida590b7d28dbee40e496794c42b682d740427cca
|
|
* Implement Float 16 Encoder and Decoder
* Add Stack Float 16 layer and create workload tests
Signed-off-by: Matthew Jackson <matthew.jackson@arm.com>
Change-Id: Ice4678226f4d22c06ebcc6db3052d42ce0c1bd67
|
|
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I0c2ead004200e053d748ea39937f2f9ed35a636b
|
|
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: Icf39434f09fd340ad664cb3b97b8bee6d9da4838
|
|
* Renamed ARMNN_COMPUTE_REF_ENABLED to ARMNN_REF_ENABLED in the
Android makefiles
Change-Id: I12af2ca00c719568274585482d428b035674737c
Signed-off-by: Matteo Martincigh <matteo.martincigh@arm.com>
|
|
* Renamed ARMCOMPUTEREF to ARMNNREF (and relative symbols) as the
previous name was misleading
!android-nn-driver:1819
Change-Id: I912e9f521750b91539e48f8a2fb4de73fb6f7425
Signed-off-by: Matteo Martincigh <matteo.martincigh@arm.com>
|
|
* Adds compatibility between Cl and Neon
* Adds related unit test
Signed-off-by: Jan Eilers <jan.eilers@arm.com>
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: Iaa454ea2a6392c89c0a83836c3156d4ddb354879
|
|
Failing VTS tests were "NeuralnetworksHidlTest.depthwise_conv2d_*"
In depthwise convolution there was a difference in weight tensor channel
order between the reference and ACL implementations. This specifically related
to NCHW. This commit:
* Adds ReorderWeightChannelsForAcl to WorkloadUtils which will correct the weight tensor channel order.
* Add unit tests to detect this problem.
Signed-off-by: Colm Donelan <Colm.Donelan@arm.com>
Change-Id: Icaeac08e14b3d5da9e222ad2f118db55ebb15d09
|
|
* Made the build of the reference backend depend on a new ARMCOMPUTEREF
macro
* Made the relevant targets dependent on the ref backend
* Moved Cl and Neon static registry initializers to separate files
* Wrapped some of the unit tests into proper ifdefs where necessary
Change-Id: I7f2c42699682630233a4c4b6aed2f005083de189
Signed-off-by: Matteo Martincigh <matteo.martincigh@arm.com>
|
|
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I9d19ae4559a98b9e740c1d4d05add0c5788093d6
|
|
* NEON support for TransposeConolution2d
* Indentation fix on SpaceToDepth LayerSupport method
Signed-off-by: Keith Davis <keith.davis@arm.com>
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Signed-off-by: Aron Virginas-Tar <aron.virginas-tar@arm.com>
Change-Id: Ie3a01a22e05f62f475f768dfe05ce561d2f3b0ad
|
|
Signed-off-by: Ferran Balaguer <ferran.balaguer@arm.com>
Change-Id: I94bd191f88e0911ad4e4727610e81cd7afa95512
|
|
* Adds ClTensorHandleFactory
* Includes some refactoring of NeonTensorHandleFactory
Signed-off-by: Jan Eilers <jan.eilers@arm.com>
Change-Id: I9f0874b1945050267c5ece970e6d9b200ed8a865
|
|
The following NeuralNetworkTests tests were failing on GpuAcc and CpuAcc:
GeneratedTests.softmax_v1_2_relaxed
GeneratedTests.softmax_v1_2_quant8
GeneratedTests.softmax_v1_2_2
GeneratedTests.softmax_v1_2_relaxed_2
GeneratedTests.softmax_v1_2_quant8_2
The default value for Softmax axis parameter in Android is -1 but is 1 in ACL.
Detect and handle this in ArmComputeUtils.ComputeSoftmaxAclAxis.
Signed-off-by: Colm Donelan <Colm.Donelan@arm.com>
Change-Id: Ibb0660e4cb0dc6bd4c804c4397fbd61f38acdd9c
|
|
* Increased MaxNumOfTensorDimensions and fixed issues related to its use
* Fixed issues caused by assuming 5d tensors are invalid
* Updated ArmComputeTensorUtils for 5d tensors
* Added 5d tensor unit tests for add, mul, stack and reshape (needed by IVGCVSW-3527)
Signed-off-by: Matthew Jackson <matthew.jackson@arm.com>
Change-Id: I5bcd64942d0d04efcc6c5acb240ad4b88e010743
|
|
Signed-off-by: James Conroy <james.conroy@arm.com>
Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Change-Id: I9f24f7d97c1e7d37ee7d58ff85c6c3f8496e52ec
|
|
* Change Getter Signatures to follow coding guidelines
Change-Id: Ic02621e834dbf79b9df63f8b4c6339f71651e944
Signed-off-by: Francis Murtagh <francis.murtagh@arm.com>
|
|
Signed-off-by: FinnWilliamsArm <Finn.Williams@arm.com>
Change-Id: Ia9bf55ca93e36ad7fa052b70e477cd46078e0a30
|
|
Change-Id: Iaec6956b5c459308d77d29f699ae4558bee66cd5
Signed-off-by: Francis Murtagh <francis.murtagh@arm.com>
|
|
* Fix issued caused by layers with 5x5 filters and depth multipliers > 1
Signed-off-by: Matthew Jackson <matthew.jackson@arm.com>
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I58435a1f0e3c7e69861dc130fad525a01e2a849d
|
|
Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Change-Id: Id990ae7445080f6b4d5a0e6f942f871879a52f75
|
|
* Add Neon Workload
* Update NeonWorkloads.hpp
* Update NeonWorkloadFactory
* Update NeonLayerSupport
* Update backends.mk and CMakeLists.txt
* Add NeonCreateWorkload test
* Enable LayerTest
!android-nn-driver:1685
Change-Id: Idd799bbf039acf0d59084d02c3b57766ce3691b5
Signed-off-by: Francis Murtagh <francis.murtagh@arm.com>
Signed-off-by: Matthew Bentham <Matthew.Bentham@arm.com>
|
|
* Rename MemoryStrategy to EdgeStrategy
* Add MemImportLayer
* Import memory rather than copy when possible
Change-Id: I1d3a9414f2cbe517dc2aae9bbd4fdd92712b38ef
Signed-off-by: Derek Lamberti <derek.lamberti@arm.com>
|
|
* Added Neon backend support for Stack
* Added unit tests for Stack on the Neon backend
Signed-off-by: Matthew Jackson <matthew.jackson@arm.com>
Change-Id: I21379966a6303285f8b65418d0d4ac7982cc9f04
|
|
Change-Id: I5cd90121e5fe1b49b394ee6f5a87e86520a30119
Signed-off-by: Pablo Tello <pablo.tello@arm.com>
|
|
* Add check in CL and Neon to ensure axis is 1 otherwise
return unsupported.
* Edit CreateWorkload test and JsonPrinter test to ensure axis of 1.
Change-Id: I499b405532e26fefc2dd1c18b6dc6005813b5604
Signed-off-by: Francis Murtagh <francis.murtagh@arm.com>
|
|
* Fix reference L2 Normalization workload to support < 4 dimensional tensors
* Add unit test for L2 Normalization with 2d tensor to Reference, Neon and CL test suites
* Fix typo in StackLayer
Signed-off-by: Matthew Jackson <matthew.jackson@arm.com>
Change-Id: I48a6a1289bcb02955b24f261bc70b467bd1abc23
|
|
Signed-off-by: Matthew Jackson <matthew.jackson@arm.com>
Change-Id: Icd89230d0698f32a5ebe36ec5012a0fcc19e1565
|
|
* Pad value for QASYMM8 is no longer stored in quantized form.
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: I048e1d233353c0560ae03a7cc1ed5199295352bc
|
|
workloads
* Refactoring: use existing utility function for creating arm_compute::PadStrideInfo
objects in CL and NEON convolution workloads instead of duplicating code
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: Id5e5a0f264e20af99dabce8dd8c6b782dedb94e6
|
|
Change-Id: I3061fa980fccdfb358ebfaf2870b57b110ed182e
Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
|
|
* Renamed NeonResizeBilinearWorkload.* to NeonResizeWorkload.* and added support for ResizeNearestNeighbour
* Added CreateWorkload and LayerTests for Neon backend
Signed-off-by: Ellen Norris-Thompson <ellen.norris-thompson@arm.com>
Change-Id: I72f5340608a0928f8b32a41d1915ee2c35849f18
|
|
* Update NeonConvolution2dWorkload
* Enable unit tests
Signed-off-by: Jan Eilers <jan.eilers@arm.com>
Change-Id: I1709e71e4944c5cbdcc99a003479c1fce45a612a
|
|
* Includes fixes for ResizeNearestNeighbor on CL and NEON
!android-nn-driver:1523
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I959e43cab21dce10a9072320fada11fcb57d41f2
|
|
* Update neon lstm workload
* Add unit tests
* Add isLstmSupported
Change-Id: I493c159137f6544b0f2532d16d4fafd7a7e587e5
Signed-off-by: Jan Eilers <jan.eilers@arm.com>
|
|
Signed-off-by: FinnWilliamsArm <Finn.Williams@arm.com>
Change-Id: I327187b007e1aa18f3385766f4a82e113301a2f3
|
|
* Add Neon backend support for SpaceToDepth
* Enabled Neon Unit, CreateWorkload and
EndToEnd tests for SpaceToDepth.
* Added QSymm16 support to NeonTensorHandle.
Signed-off-by: Ellen Norris-Thompson <ellen.norris-thompson@arm.com>
Change-Id: Ibbb0bce5ad77dace2bf4c4c111006d2fadf844e8
|
|
Signed-off-by: FinnWilliamsArm <Finn.Williams@arm.com>
Change-Id: I685e17d2e721e4b0de50d88b04b63a5c2cfdc772
|
|
Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Change-Id: Id6ab073ae3e0ff5f60c2f4bef975798ce71c9f88
|
|
Add Neon missing Unit Test for Uint8.
Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Change-Id: I9207be8c8cf572ae9cbbf155010bb6dabe0c971c
|
|
* Updated ACL pin to latest master.
* Minor changes to Softmax Neon/CL uint8
workloads to reflect refactoring in ACL.
!android-nn-driver:1476
Change-Id: I1c5005ddbcccdb41d8cb09d3fa61cf3ce0e9ffdb
Signed-off-by: James Conroy <james.conroy@arm.com>
|
|
Change-Id: I2f92f7336d3b4188962a9d08b0e7c6372b1c648f
Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
|
|
!android-nn-driver:1451
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: Ieedbce1f6e95891137a250fdd07e2f7e4e1f4828
|
|
* Update SoftMaxUint8Workload for CL and NEON to deal with
quantization parameters as vectors.
* Change Sigmoid Activation function QAsymm8 tests to use
scale=1.f/256.f and offset=0 as quatization output parameters.
!android-nn-driver:1417
Signed-off-by: Ferran Balaguer <ferran.balaguer@arm.com>
Change-Id: Ief91f10193fbbbad0c0124ece41f0bf4e0dcd992
|
|
Signed-off-by: Ferran Balaguer <ferran.balaguer@arm.com>
Change-Id: I938d97e6a2bfab8ff5f4548efeeaa43b7f5ccda2
|
|
Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Change-Id: Ia655b1b6c5f57d2b7c0d9dd17be342cd01f36c09
|
|
* Renamed RefMeanFloat32Workload and RefMeanUint8Workload
to RefMeanWorkload, updated references to reflect this
change.
* Refactored RefFloorWorkload to use Decoders/Encoders,
to support the use of multiple data types.
* Deleted reference Unit8 Mean tests as they were
duplicates of the Float32 tests. Refactored these tests
to support multiple data types and updated references.
* Adjusted the values used in the tests' input tensors so
that they are more like floating point numbers
e.g. change 1.0f to 1.5f.
* Replace size_t with unsigned int in Mean ref workload,
for better compatibility with the Encoder/Decoder,
removed some unnecessary casts after this.
* Added ValidateTensorDataTypesMatch() function to
WorkloadData.cpp, added CreateIncorrectDimensionsErrorMsg
function to RefLayerSupport.cpp.
* Added passing and failing tests for ref IsMeanSupported.
Signed-off-by: James Conroy <james.conroy@arm.com>
Change-Id: Id3d44463d1385255c727a497d4026d21a49e7eb2
|
|
Change-Id: Ia68da09d8f0fb0a04af9cb61062d7edaa5f1b887
Signed-off-by: Derek Lamberti <derek.lamberti@arm.com>
|