Age | Commit message (Collapse) | Author |
|
* Using the new public armnnUtils API
!armnn:2386
Signed-off-by: Matteo Martincigh <matteo.martincigh@arm.com>
Change-Id: Ifa8af20ca131b1be19dbdceea6674a1274a396f3
|
|
This should all be done in the LayerSupport mechanism in Arm NN now.
Change-Id: Icbbb1be2efad2ababf8bdd5543f3e5a06958dccd
Signed-off-by: Matthew Bentham <Matthew.Bentham@arm.com>
|
|
Signed-off-by: FinnWilliamsArm <Finn.Williams@Arm.com>
Change-Id: I999d9091bc4046861433d4eb3109fe972611bd82
|
|
* Added DequantizeStatus enum to capture the status of the
dequantization process
* Added alias DequantizeResult for the return tuple
* Renamed instances of Operand, Operation and Model template
variables to HalOperand, HalOperation and HalModel for
consistency across ConversionUtils
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I2d5cf0dd9f9baba95684bb0b9fe20e63c0f377a3
|
|
* Quantization dimension other than 0 is not supported.
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: Ied3914600a754b799e98d5660ad6196c8c4fa23d
|
|
!armnn:2313
Signed-off-by: Francis Murtagh <francis.murtagh@arm.com>
Change-Id: I543136e4e2ef9aece1378d2642064cc585246645
|
|
* Fixed for failing LSTM unit tests
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: I773c5227bc8d5606924cc0472c51172476773056
|
|
* Removed TENSOR_QUANT8_SYMM from the list of generally supported
tensor data types
* Fixed tensor info in DequantizeIfRequired() for on the fly
dequantized QSymm8 weights
* Moved code for checking whether a Dequantize operator is linked
to FullyConnected or Lstm weights from ConvertDequantize() into
a separate function inside 1.2/HalPolicy.cpp
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I19ea6f89a90f553a964b87d44f8ad8a064e96f7f
|
|
permutation
Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Change-Id: Ie4a27d805d6d953fc54264ef9a0947f82c5b1ff8
|
|
* Added workaround in FC to deal with non const weights
* Added workaround in LSTM to deal with non const weights
Signed-off-by: Pablo Tello <pablo.tello@arm.com>
Signed-off-by: Matteo Martincigh <matteo.martincigh@arm.com>
Change-Id: I854eea6a74a6959606ff25b52a0ed80b3e0a18ab
|
|
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: Ife7fa63b8839465e8f9f8626f34ca8c0f4d12788
|
|
*ConvertDepthwise reports that weights are not supported with the first dimension of the tensor is not 1
*ConvertConcat was missing one case for reporting unsupported inputs.
Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Change-Id: I7759ad2fea55d272e98b9eabbf4821dc9ac826e1
|
|
Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Change-Id: Ie98688175203563ae08019c0c3c876d9c5eae24c
|
|
* Added ConvertDepthToSpace() to hal_1_0::HalPolicy and hal_1_2::HalPolicy
* Implemented ConvertDepthToSpace() template inside ConversionUtils.hpp
* Changed unsupported operation from DEPTH_TO_SPACE to HASHTABLE_LOOKUP
in GenericLayerTests/GetSupportedOperations
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I12bf73ea721e7b6d49cc4a76000b43a3f274c6f5
|
|
Signed-off-by: Kevin May <kevin.may@arm.com>
Change-Id: Icd32bf6f4d5f58e689b55c1c71b0d028155603c8
|
|
* Added IsReshapeSupported check on available backend(s) after adding Reshape layer.
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: I38a36d5a3db0d67f0e4b4be2ef672dedb75bff22
|
|
* Added check for unsupported optional parameters for L2Normalization and LocalResponseNormalization to skip the related tests rather than fail them.
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: Ia17be706b14b2e2faf8eb5a0ec335c2e3863bf16
|
|
* Templated and moved V1.0 and V1.1 Convert methods to ensure they can work with later versions of models, operations and operands.
* The V1.2 HalPolicy no longer converts V1.2 models, operations and operands to earlier versions.
* The V1.2 HalPolicy no longer passes operations to the V1.1 or V1.0 HalPolicies for conversion.
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: I5de59d43a3abb1f8ac0253dc637ad68318960c76
|
|
* PReLU layers were being skipped due to logic error.
Resubmitting due to accidental reversion.
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: I6aeed11519d19ba7ec8b4993694e414f58799028
|
|
Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Signed-off-by: Aron Virginas-Tar <aron.virginas-tar@arm.com>
Change-Id: I5c10ab18343ecf0ebeab24a436e5be2b6c2831c7
|
|
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I92700da149a800f1d8215e964d1e67056628f067
|
|
* Fixed Skipped VTS Concatenate Tests.
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: I29e7dcdedefc0e9c54f86fa5de23aa714c469585
|
|
Signed-off-by: Finn Williams <Finn.Williams@arm.com>
Change-Id: I9d0d0a75d54c25075d8d87c6265e350486157f5c
|
|
Signed-off-by: Finn Williams <Finn.Williams@arm.com>
Change-Id: Ibd683c147b6f2fb966a8f90105c9213880100cf8
|
|
Signed-off-by: Finn Williams <Finn.Williams@arm.com>
Change-Id: I3f2928ba86a9d306a7eb400db3a420e42cf3fa7e
|
|
* A model that has Inputs with different quantized scales is not compliant with 1.1
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: Ifb8277d78f05b5ef017effa879322a08c0efc851
|
|
ConvertDepthwiseConv2d()
* Removed code for reading data layout flag (only available in HAL1.2)
* Removed code for reading dilation parameters (only available in HAL1.2)
* Added more restrictive input size checks (== instead of >=, as none
of the inputs are optional in HAL1.0/1.1)
* Removed superfluous input size validation from 1.0/HalPolicy.cpp
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I8387120b9abcc8458f7fac6df577cb6453051be3
|
|
* PReLU layers were being skipped due to logic error.
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: Ic0ec707da784b44d72958dee02901b0173b25e5c
|
|
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: I9f9d11b19de591ec8936822ca229b4090ed318fe
|
|
* Report dynamic output tensors as unsupported for all operations,
regardless of HAL level
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I2341dc96be965886666b75515e9a226d813a1591
|
|
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I55502cf2abcf28a1c1810269f1955ca78b23c39a
|
|
* Fixed for failing Conv2d, DepthwiseConv2d, and Activation tests on Hal 1.0 and 1.1 in Q
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Signed-off-by: Aron Virginas-Tar <aron.virginas-tar@arm.com>
Change-Id: I435338b90b6c501320083f2fd9372e3a4ac3c32c
|
|
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I9b1b45f7723147456d2fb9c5b94822d581cf446e
|
|
* Fixed Pad and PadV2 failures and skips.
* Templated ConvertPad to enable float16 tests to run.
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: I50ded84fe44ea5d5949e877f383f32adff88680d
|
|
* Updated ConvertToActivation, ConvertConv2d, and ConvertDepthwiseConv2d functions
to infer output shape from input if it is dynamic
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: Ie24fbfd87c6186c69c3ecba5c68a6866507fb449
|
|
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: Idd10f34babc0d2552d599872b853ba5fb5c98351
|
|
!armnn:1508
Signed-off-by: Ferran Balaguer <ferran.balaguer@arm.com>
Change-Id: Ica5fcb683f101bde9e651f0be0f5b9b4c409d1aa
|
|
* Changed calculation when calculating implicit padding to take dilation into account.
* Added CalcPadding function that takes dilation as a parameter.
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: I38981d9b2cfe8ef36cd0a14d090b541111e7bf34
|
|
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I8fc7a716455be3f51b51177f6896a73790a41fc3
|
|
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: If3da7b9015fb816bb27b85ab491102005859b3b0
|
|
VTS/NN
tests
* Updated ConvertPrelu to support input broadcasting
* Updated the BroadcastTensor utility function so that it preserves
the order of the inputs
* Updated the documentation
Signed-off-by: Matteo Martincigh <matteo.martincigh@arm.com>
Change-Id: Ibbac6f7161132740e61c85f597f8be70cd5d7325
|
|
* Added ConvertNearestNeighbor to 1.2/HalPolicy
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I9858849b70be36e7e5ced3dae667c85c03e50d4f
|
|
* Updated android-nn-driver to run VTS tests with 1.2 Driver
Change-Id: I99a73b1e32dcf5bc655d88c2dbe0e610e245ea15
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
|
|
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I162cdca922655d1bd71b18fc5d2937351f8879be
|
|
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I3968d682b67439d4b4aca670d074ea4f8993a39a
|
|
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: Idce519adcd4a2983e8c338b2e4483e040ad879d2
|
|
* Updated Android.mk to build HAL 1.2 driver
* Added 1.2 HalPolicy and ArmnnDriver
* Added 1.2 ArmnnPreparedModel
* Updated converters and utilities to accept new HAL 1.2 operands and operand types.
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: I62856deab24e106f72cccce09468db4971756fa6
|
|
Explicitly use HAL V1_0 Operand, OperandType, and IPreparedModelCallback
Change-Id: If6aa6e9dc1bd0b7673ec247ac3560e7f072ef699
Signed-off-by: Matthew Bentham <matthew.bentham@arm.com>
|
|
!armnn:962
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: Ia70c4b8397e5d6422d1947fca8c8db29a423822e
|
|
Change-Id: I4abe1f750801911570b6dc65c187b828c5929b5f
Signed-off-by: Nattapat Chaimanowong <nattapat.chaimanowong@arm.com>
|