Age | Commit message (Collapse) | Author |
|
* Updated FAQ to include instructions on resolving version incompatibilities between android-nn-driver, armnn and clframework.
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: I5f358d7241b9120c6896b8b8b38c89246e6573b1
|
|
* Removed the inclusion of the Half.hpp header from the
Android NN Driver, as it's a private header not part of the
now public armnnUtils API
* Refactored the code not to use that header
Signed-off-by: Matteo Martincigh <matteo.martincigh@arm.com>
Change-Id: I0feeb86ccb31e016395e623029974f599a174149
|
|
Signed-off-by: Derek Lamberti <derek.lamberti@arm.com>
Change-Id: I89ae4f3b607ccb83f1bf16e085c805d5fc5cf89c
|
|
* 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: Teresa Charlin <teresa.charlinreyes@arm.com>
Change-Id: I56959a6fd296781d4604c55179471299c321153e
|
|
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
|
|
Change-Id: I118cfe12e24b689539c634625910fb007cc436b6
|
|
* 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 numerous CTS/VTS failures related to Quantization
!armnn:2316
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: I50b7edf42e416abd230783a5081d1a6511bc42d4
|
|
* Fixed for failing LSTM unit tests
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: I773c5227bc8d5606924cc0472c51172476773056
|
|
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I3e8069d6cb4a2149396329fee4643d0fc5a12181
|
|
* 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
|
|
Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Change-Id: I344dcf039e68297bc927dcb2b3f7f0f59398be5e
|
|
* Add TensorQuant8SymmPerChannel to ArmnnDriverImpl::getCapabilities_1_2
* Add TensorQuant8Symm and TensorQuant8SymmPerChannel to Capabilities
supported operands tests
Signed-off-by: Kevin May <kevin.may@arm.com>
Change-Id: Ifbc742c5562f243172d2f4e6a2f6a8f3b6a405ea
|
|
!armnn:2211
Signed-off-by: James Conroy <james.conroy@arm.com>
Change-Id: I248bfb6a45c90f053a849f38607a4ba6c14f39d1
|
|
permutation
Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Change-Id: Ie4a27d805d6d953fc54264ef9a0947f82c5b1ff8
|
|
!armnn:2232
Signed-off-by: janeil01 <jan.eilers@arm.com>
Change-Id: Ib24989dba72a9f9acd1c8017c1f39c3ee74030ea
|
|
* Use an android strong pointer instead of a unique_ptr to
avoid undefined behaviour of releasing unique_ptr to strong pointer
Signed-off-by: Kevin May <kevin.may@arm.com>
Change-Id: Icac255ba2407796888f20ffaba99a7988cb6f63c
|
|
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: Ice7c4d3273db31130ec64edc1b76d1c9d5197961
|
|
GROUPED_CONV2D
* Report per-axis quantization as unsupported in ConvertGroupedConv2d()
until support will have been added
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: Ibadb848b60956f8c90fa8b538d50dea8d9a57d92
|
|
* 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
|
|
* Added support for the following HAL1.2 operations: EQUAL,
GREATER, GREATER_EQUAL, LESS, LESS_EQUAL and NOT_EQUAL
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I71b68db70232da4aaad28caa7b0b5f9a1d7778d0
|
|
* Added resize method (Bilinear or NearestNeighbor) to log
* Removed superfluous call to SetTensorInfo() for output
slot
!armnn:2179
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I4d1ad0d4c1df67c62c71f79125612b0dffa85455
|
|
* The version macro is set in the Version.hpp just like before,
so parsing it from file is no longer necessary
!armnn:2090
Signed-off-by: Matteo Martincigh <matteo.martincigh@arm.com>
Change-Id: I3f8d10bd7b8947f9361e50edf033ea4e93451faf
|
|
* Move the ArmNN version to a separate ArmnnVersion.txt file
* Updated makefiles accordingly
!referencetests:206978
!armnn:2089
Signed-off-by: Matteo Martincigh <matteo.martincigh@arm.com>
Change-Id: I02b838bc71611a866b3f96335e44a33b0167134f
|
|
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I59645b339f3b176e5d0852769acb95f5657101d3
|
|
INSTANCE_NORMALIZATION
* Since CpuAcc does not support FLOAT16 for INSTANCE_NORMALIZATION, it should
be removed from the list of (fully) supported types
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: Ice557a7a08eca6faa8283c399fa2b03e64e3ddc6
|
|
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: Ia4dfbda7aceb4cdfb8f83d49e4df21dedd415b7b
|
|
Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Change-Id: Ie98688175203563ae08019c0c3c876d9c5eae24c
|
|
* Add CLInstanceNormalizationLayer.cpp and CLInstanceNormalizationLayerKernel.cpp
to Android.bp
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I1f86c961c52cda95c6d92aa6108b3eb6bb942f30
|
|
* Added runtime/IWeightsManager.cpp to list of required ACL source files
!armnn:2034
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I0366e936be4a48fe140cd2571b83c8d0c08443d0
|
|
* 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
|
|
* Update ACL pin to include change of ArgMinMax NEON/CL output type to Signed32
!armnn:2010
Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Change-Id: I27c470d02253cbfd57c75c915e1e5ab9707e48dc
|
|
Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Change-Id: Ibf6c53822f728c0e15a9ca1cd2c2ad3593edbd82
|
|
* Update ACL pin to include fix for Activation.Sqrt issue on Neon
* Changes to Android.bp to build the driver with the latest ACL version
!armnn:1925
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: I59c4a7bc9c4fe7399f7d8320d2eb32bf57d6431f
|
|
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I1e0909fa7fce64e9085889d645b093b51d63d8b2
|
|
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I1baf88f15c9bdcffd38e910070c97909b7f913b3
|
|
Signed-off-by: Kevin May <kevin.may@arm.com>
Change-Id: Icd32bf6f4d5f58e689b55c1c71b0d028155603c8
|
|
* Implemented Convert() function for SQRT operator
* Fixed some typos in HAL 1.2 Policy
Change-Id: I1e0bc7a43622c52f053927248ee1a71cb797145e
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
|
|
* Changes to Android.bp to build the driver with the latest
ACL version
!armnn:1855
Change-Id: I3360c41b7b4f30c5adb46977a1ea7bd8946f433e
Signed-off-by: Matteo Martincigh <matteo.martincigh@arm.com>
|
|
* Removed unsupported operation sections
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: I79216c5c015efeb5d2e19cd85f186c2f73bd1ca8
|
|
* Renamed the cl/neon backend macros to their original values, as
they were renamed by mistake
* Reverted misleading comment
* Updated the readme file
!armnn:1837
Change-Id: I49faabae2c8ad196f51eca5f8dd049ecf74b36e1
Signed-off-by: Matteo Martincigh <matteo.martincigh@arm.com>
|
|
Change-Id: I5982d8dc733bd8bf1eed51c467df5a16a6aabcdb
Signed-off-by: Matteo Martincigh <matteo.martincigh@arm.com>
|
|
* Renamed ARMCOMPUTEREF to ARMNNREF (and relative symbols) as the
previous name was misleading
Change-Id: Ia0400a7d63a76c80fb5bdd26410553db1c15cba0
Signed-off-by: Matteo Martincigh <matteo.martincigh@arm.com>
|
|
Signed-off-by: Ellen Norris-Thompson <ellen.norris-thompson@arm.com>
Change-Id: Ie3c535789186571df9000d75ad2a62ac8c94dcb4
|