Age | Commit message (Collapse) | Author |
|
* Unidirectional Sequence Lstm is supported in android-nn-driver.
* CTS and VTS pass for Unidirectional Sequence Lstm if there are
any without weights as inputs. If that isn't the case use the
tests implemented in the android driver for LSTM.
Signed-off-by: Cathal Corbett <cathal.corbett@arm.com>
Change-Id: Ie3a237b67f5101c4f9a2c7bea796e9c674bedba1
|
|
* Fixed test failures.
!armnn:6617
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Signed-off-by: Kevin May <kevin.may@arm.com>
Change-Id: I9989ece8999d67dd40dfcf69b73f4d80f71687a4
|
|
* Cached serialized ArmNN model.
!armnn:6384
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Signed-off-by: Kevin May <kevin.may@arm.com>
Change-Id: I78120a7f8ea892a28c0ff25f1b54e67a4f912574
|
|
Signed-off-by: Kevin May <kevmay01@e126674.cambridge.arm.com>
Change-Id: I92334f3215160ccdd3eb974456a6400bced42065
|
|
!armnn:6211
Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Change-Id: Ie5e5d160f6d826c30f1fff298a4a73151c1ccdbb
|
|
* ArmnnDriver queries the options and returns CPU or GPU
depending on which is the first backend listed in the options
* Resolves MLCE-401
Change-Id: If4e63e144507e817449f37926711fa325861b57d
Signed-off-by: Pablo Tello <pablo.tello@arm.com>
|
|
!armnn:6053
Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Change-Id: I678be047c0919d98dbd22d22f4c1fd1ea80ea0b7
|
|
!armnn:5801
Signed-off-by: Finn Williams <Finn.Williams@arm.com>
Change-Id: I9964d0899ce752441f380edddbd974010257b2dd
|
|
!armnn:5755
Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Change-Id: I82dc434ee9bfff452ec6c27527a5b3f91ebfe2ea
|
|
Signed-off-by: Finn Williams <Finn.Williams@arm.com>
Change-Id: I1f13d04100fdb119495b9e3054425bf3babc59f1
|
|
!armnn:5494
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: Ie99d58f7421c2634c23d2a68d1acaaac651963a4
|
|
* MEAN operator second input data type is different than Reduction operators
* Use existing ConvertMean() function for MEAN operator
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: I211cb77515d88e54ded9646eebe3e901d437827c
|
|
Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Change-Id: Ib88544dd6adc3d6a0d2f4c0b395e05b89bc8ad3e
|
|
Signed-off-by: Finn Williams <Finn.Williams@arm.com>
Change-Id: I15ca7db703065b0c70766a18b1a9a213834f39e0
|
|
Signed-off-by: Kevin May <kevin.may@arm.com>
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: I1e1db52322092c6b1b7ac6183c3adc90aabcec24
|
|
* Added --number-of-threads command line option to android-nn-driver
!armnn:5068
Signed-off-by: Matthew Sloyan <matthew.sloyan@arm.com>
Change-Id: I248326bf3c0355a7a17b70cb1aac7b7976820c12
|
|
* Implemented serialization of the network on android-nn-driver
!armnn:4850
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: I3caf07bd4d1d2a3068c58f0b344303c4cf977ca6
|
|
* Added Flatbuffers dependency to android-nn-driver
* Added CacheLoadedNetwork options to android-nn-driver
Options will be android.hardware.neuralnetworks@1.3-service-armnn -v -c GpuAcc -n arm-armnn -s -q <file_path> &
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: I2cdf0b84062cb1c6ad4346cd64b00161cfe85920
|
|
Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Change-Id: Ia3909f32a837ade88e848de7f9ee0bf003bc8c01
|
|
* Added -a command line option to enable fast_math on Neon/CL when launching the NNAPI driver
* The invocation will look something like:
/vendor/bin/hw/android.hardware.neuralnetworks@1.3-service-armnn -v -c GpuAcc -n arm-armnn -a
* Enabling fast_math may cause an increase in VTS or CTS test failures due to a change in precision
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: Ifd8931f95efd65d3ae3aca9d9d61bd381e4e24a5
|
|
Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Change-Id: I69dd78d47628355c207a450119b054b04581c729
|
|
* Add QLSTM support for Android NN Driver
* Add overrideOutputInfo parameter to SetupAndTrackLayerOutputSlot
* Add optional condition to GetInputScalar
* Refactor Quantized 16 Bit LSTM impl
Change-Id: Ie8fa98ad5ee4a62174ef91ca80f1df62b7fde937
Signed-off-by: Keith Davis <keith.davis@arm.com>
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
|
|
Signed-off-by: Kevin May <kevin.may@arm.com>
Change-Id: I13be6896dd313bfeaf2f6b08ce0e47116116c3c0
|
|
* Add new 1.3 files HalPolicy, ArmnnDriver, ArmnnDriverImpl
* Add new .rc file for 1.3 service
* Add ArmnnPreparedModel_1_3 and implement new functions
* Update Android.mk with 1.3 driver and service
* Refactor ifdef to include ARMNN_ANDROID_NN_V1_3
* Create Utils getMainModel for new 1.3 Model Main Subgraph
* Use android Utils to convertToV1_X in ArmnnPrepapredModel_1_3
* Refactor HAL 1.2 convert functions into ConversionUtils_1_2.hpp
* Replace ArmnnBurstExecutorWithCache with call to ExecutionBurstServer
Signed-off-by: Kevin May <kevin.may@arm.com>
Change-Id: I514069e9e1b16bcd1c4abfb5d563d25ac22d02e3
|
|
android-nn-driver
!armnn:2917
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: I6e38fd3d11dc3e30dfac0ddfc39b216b2a15d047
|
|
* Added TENSOR_BOOL8 support
* Added Broadcast support to comparision operators
!armnn:2903
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: I844e32b57399eff2dc60af9b2099145316c80cae
|
|
* Update ErrorStatus to V1_0::ErrorStatus
* Update Request to V1_0::Request
* Update OperandType to V1_2::OperandType
* Add namespace android::nn::hal in ArmnnDriverImpl for R only
* Add missing g_RelaxedFloat32toFloat16PerformancePowerUsageName
* Add namespace V1_0 or V1_1 where necessary
* Update Android.mk with R macro and android.hardware.neuralnetworks@1.3
* Remove androidnn.go
* include IAllocator in DriverTestHelpers
* Remove unused LOCAL_CFLAGS
Signed-off-by: Kevin May <kevin.may@arm.com>
Change-Id: I1787f1ed6784b3bbec017536d87d49197405e853
Signed-off-by: Kevin May <kevin.may@arm.com>
|
|
* Changed ConvertTransposeConv2d to use the unsigned values for strideX and strideY
not the signed values from the descriptor.
!ComputeLibrary:2599
!armnn:2612
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: Iee300041f3ed585820b7c7e2edd8ab541f339d59
|
|
!android-nn-driver:2642
Signed-off-by: Derek Lamberti <derek.lamberti@arm.com>
Change-Id: I4db6ea4ed0a192c85f124c4a9ced60b1666a3870
Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
|
|
!armnn:2643
* Add calls before addition of these layers in ConvertConcatenation
* Add outputInfo parameter wherever needed for IsReshapeSupported
Signed-off-by: Kevin May <kevin.may@arm.com>
Change-Id: Ic5d142ea046161960ff2fc137bd261ebb4e6ac0c
|
|
* .dot file will start with the same number as the associated input
and output tensor dump files
Change-Id: Ic64539854c2b8c34a7034fa18a142b2dfe67df7d
Signed-off-by: Jim Flynn <jim.flynn@arm.com>
IVGCVSW-4271 Add getSupportedOperations marker file
* Also fix segfault in previous graph dump code
Change-Id: I01915f47ae17d89ad3595430d073a17b1a1fda58
Signed-off-by: Jim Flynn <jim.flynn@arm.com>
|
|
* Account for deprecated Abs and Rsqrt layers in ArmNN
* Update HAL 1.2
* Neg HAL operation support added
!armnn:2542
Signed-off-by: josh minor <josh.minor@arm.com>
Change-Id: I043bf64ac7ed3aea65560d72acf664e079917baf
Signed-off-by: josh minor <josh.minor@arm.com>
|
|
* Add FP16 support
Signed-off-by: Keith Davis <keith.davis@arm.com>
Change-Id: I59a2862c734aa3678f19cc47f024f8cb92fb557e
|
|
Change-Id: I7977d697772349b8ea7eb300937409ce0a3a4dee
Signed-off-by: Pablo Tello <pablo.tello@arm.com>
|
|
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
|
|
!armnn:2313
Signed-off-by: Francis Murtagh <francis.murtagh@arm.com>
Change-Id: I543136e4e2ef9aece1378d2642064cc585246645
|
|
* 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
|
|
* 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
|
|
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
|
|
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I59645b339f3b176e5d0852769acb95f5657101d3
|
|
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
|
|
* 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
|