Age | Commit message (Collapse) | Author |
|
* Add steps to build Android T in IntegratorGuide
Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
Change-Id: Id2f4c33c50c10c31b2c9d3c76875ffaf780a6bad
|
|
Signed-off-by: Colm Donelan <colm.donelan@arm.com>
Change-Id: Ie218ba025f83ef7163eb8b548a89625a271c8a4c
|
|
* AddConv with weights and bias
!armnn:8167
Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Change-Id: I9b2ad21482d503161721b7d4eee5c5d16b59a379
|
|
Signed-off-by: Jim Flynn <jim.flynn@arm.com>
Change-Id: Ia1f7a30c5d2fa3436430cbfae693b80ec240e723
|
|
Signed-off-by: Jim Flynn <jim.flynn@arm.com>
Change-Id: I908aa5c7581b7bb1eebeb8e32625c9b64a9ff7e9
|
|
* Android 13 does not differ significantly form 12 so there is no
need to add a new Android variable for 13
Signed-off-by: Kevin May <kevin.may@arm.com>
Change-Id: I4880bd93fea9bdfff6d1e18660ea637b7b6d2ace
|
|
* Adds map file to libArmnnDriver for each Hal version
* Adds map file to neuralNetwork driver for each hal version
Signed-off-by: Ryan OShea <ryan.oshea3@arm.com>
Change-Id: I644d0aefbd55edf8ad2d63d0166e9cbb9d768a42
|
|
* Removed ability to use AOSP version of flatbuffers
* The version of Flatbuffers shipped with Android P is V1.8.0 but
ACL and ArmNN need V1.12.0 or later.
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: I791855ba474c02c119c3353c5195a3ba6d48c3b5
|
|
* Fixed bug where flatbuffers was downloaded every time setup.sh was run.
* setup.sh will attempt to use the version of flatbuffers used by
android if one is available.
* setup.sh will now generate ArmnnSchema_generated.h where one does
not exist
!armnn:7625
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: I62f3be6ebfc7fbef39597295dca194d70294646c
|
|
Signed-off-by: Cathal Corbett <cathal.corbett@arm.com>
Change-Id: Ibf9bfa28099b476862ff6ac87e10ff28db8ff3ad
|
|
* Add support for Android-nn-driver
!armnn:7382
Signed-off-by: Keith Davis <keith.davis@arm.com>
Signed-off-by: Kevin May <kevin.may@arm.com>
Change-Id: I7ace53cf1c8954c1f2d5588387616d7179ef3bf7
|
|
* Fixing a segmentation fault when SanitizeQuantizationScale is
called on a LayerInputHandle whose m_OutputSlot is nullptr.
Signed-off-by: Colm Donelan <colm.donelan@arm.com>
Change-Id: I447c405086ff327a28a72938fbf82844d715a17d
|
|
!armnn:7417
Signed-off-by: Cathal Corbett <cathal.corbett@arm.com>
Change-Id: Ic37d6b8677e040c60a90358dd0a4a8eb33fb6ea0
|
|
* Corrected TensorInfo order for IsUnidirectionalSequenceLstmSupported
* outputStateOut TensorInfo is not optional.
* cellStateOut TensorInfo is not optional.
* TensorInfo Order matches other QLSTM/LSTM layers.
!armnn:7455
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: I5b0e0fa4b6e1c3da6689d9aefc9b959172c2e7d4
|
|
Signed-off-by: David Monahan <David.Monahan@arm.com>
Change-Id: If873ed691e1878e8b0c1dc52717b4418a178c72f
|
|
android-nn-driver do not execute.
* Change to ConvertionUtils.hpp fix ConcatTests_GpuAcc tests.
!armnn:7300
Signed-off-by: Cathal Corbett <cathal.corbett@arm.com>
Change-Id: I1543b57c6d1f9660916ca530b421c87e54b13f72
|
|
!armnn:7327
Change-Id: If4efc2195eb8713e98767fd4732ae9f4f2bfe09b
Signed-off-by: Jim Flynn <jim.flynn@arm.com>
|
|
* Modify all versions of ExecuteGraph to include timing information. This
can give a rough guide as to how much overhead the driver is adding.
Signed-off-by: Colm Donelan <colm.donelan@arm.com>
Change-Id: Ie168e060c22b8417efa2ede0c9bdc7ef2631b913
|
|
* Added timepoints to ArmnnDriverImpl across all versions
* Added ALOGV message to display time taken for prepareModel function
* Added ALOGV message to display time taken for prepareModelFromCache function
Signed-off-by: Ryan OShea <ryan.oshea3@arm.com>
Change-Id: I07f341099e81a0da45db0e8f575ba3c26631e024
|
|
android-nn-driver
!armnn:7222
Signed-off-by: Cathal Corbett <cathal.corbett@arm.com>
Change-Id: Ie7e3802a1a4a43abbb924a86d6608b77d19f327c
|
|
!armnn:7173
Signed-off-by: David Monahan <David.Monahan@arm.com>
Change-Id: I642734b414a785f831e1a825790a78ee5078950f
|
|
android-nn-driver""
This reverts commit 8069603dc44b7673b356f66517cd8b25af8080f0.
* Reason for revert: Try reenable import aligned host memory in android-nn-driver
* Added a check to ArmNNDriverImpl.cpp to not call ExecuteWithDummyInputs with GpuAcc
* Added new android-nn-driver driver options to enable / disable Import and Export
* Import is disabled by default for now due to conv2d issues
* Export is enabled by default
!armnn:7147
Change-Id: I91110c58ebb3931d1c458e3774944e55c1250dd8
Signed-off-by: David Monahan <David.Monahan@arm.com>
|
|
* Added a note in the Integrator guide for the new error while building android
* Removed boost from FAQs
Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
Change-Id: I1dc08bf2ff772550ac4e766bdf791301ea11cd5c
|
|
* In the destructor of ArmnnPreparedModel, DumpJsonProfilingIfRequired
was being called with a potentially null profiler.
Signed-off-by: Colm Donelan <colm.donelan@arm.com>
Change-Id: Id0d3a2403878fedc4e05dc498edb48d574232c71
|
|
Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
Change-Id: I8fbfa37b2ccd7066dc2dda6d595fe33eca79618b
|
|
Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
Change-Id: I4be1bda78c4343f8ccbb8df331538cb4f5d763a6
|
|
This reverts commit 8509422e1e7d629f88d6a5e4e4bded8a682f435b.
Reason for revert: Reverting as this might be the cause of several segfaults on CI
Change-Id: I902b4ddaec23dc46a2459f8512ec458e1aa722f3
|
|
* Use ARMNN_ETHOS_ENABLE variable to enable it
Signed-off-by: Kevin May <kevin.may@arm.com>
Change-Id: I5465b7738556604367262cf9aa8e1ce6e01639b9
|
|
!armnn:7025
Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Change-Id: Iaa6112fb52d0f2942e6c52f90d96a8dc96000518
|
|
When building Android S this is during the build ending up as
both unsigned long long and unsigned long, this makes the
build break when building the tests.
As both seem to be used during a normal build it will not help
to just change the %ul to %ull as you get an error on that instead.
This fixes so it's always printing with unsigned long.
Signed-off-by: Zingo Andersen <zingo.andersen@arm.com>
Change-Id: I1d1e4b8ab72c563a3b570f30112a7ce076f0a54e
|
|
* 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
|
|
Change-Id: I95e6a3819ba99e41883f9d64a0bd6ce8706a7a4f
Signed-off-by: Zingo Andersen <zingo.andersen@arm.com>
|
|
* When handling Tensors DumpTensor was automatically trying to turn
them into ConstTensors but then threw an exceptions when IsConstant
returned false
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: I8681bb3dd41cfe19c60fbd1cc9394c8b6cca551e
|
|
* 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
|
|
!armnn:6495
* Made fixes to 3 files ArmnnPreparedModel(_1_2/_1_3).cpp to set
TensorInfo to constant in PrepareMemoryForInputs().
* Fixes made due to cts and vts failures in
https://review.mlplatform.org/c/ml/armnn/+/6495
when a Tensor is passed to a ConstTensor without the
TensorInfo isConstant parameter being set.
Signed-off-by: Cathal Corbett <cathal.corbett@arm.com>
Change-Id: Iee179586652f365665a22f1e3409c1488c2d8bb3
|
|
* 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
|
|
* Removed unreachable code
* break after if else where both branches return
* Removed unused operations
* Result of dstPtr++ is unused
* Fixed possible overflow
* Axis dimensionSequence assignment can result in overflow where rank
is not 4
* Removed use of old-style casts
* Fixed spelling mistakes in error messages
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: If2a7ab63fc1d200cb18b494d99a67bbddb42f0f8
|
|
Signed-off-by: Zingo Andersen <zingo.andersen@arm.com>
Change-Id: I95e6a3819ba60e41885f9d64c0bd6ce8706a7b1e
|
|
Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Change-Id: Ic82d8ac6461e404f41e8888336e8d7514bef6433
|
|
Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Change-Id: Icabc39559efd9e621657eb48454fe52155e8c9ff
|
|
Signed-off-by: Kevin May <kevmay01@e126674.cambridge.arm.com>
Change-Id: I92334f3215160ccdd3eb974456a6400bced42065
|
|
* Changed asserts to check for errors and return appropriate values or
throw exceptions
* Changed unit tests to use Doctest's long macro names as the short
macro names clashed with Android's Logging macros
* Removed unused #includes
* Clarified ambiguous #includes
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: Ice92a37590df727fd581d3be5ff2716665f26a13
|
|
!armnn:6254
Signed-off-by: Jan Eilers <jan.eilers@arm.com>
Change-Id: I728048874839cddedc8421233294d51d7f3e2415
|
|
!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
|
|
Signed-off-by: Matthew Sloyan <matthew.sloyan@arm.com>
Change-Id: I2006f1d696d155742227eed940d7d97ed9385a66
|
|
* Some tests on GpuAcc and CpuAcc P & Q were skipping because
IsConstant flag wasn't set on TensorInfo passed to validation function.
Signed-off-by: Matthew Sloyan <matthew.sloyan@arm.com>
Change-Id: Idd977bf621b71bb9625d8fc6620c29915de4b431
|
|
* Constant weights and biases are now stored as Constant layers.
!armnn:5942
Signed-off-by: Matthew Sloyan <matthew.sloyan@arm.com>
Change-Id: I867a696eb684df2d0d5f7db8dd9c38933445f388
|
|
Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Change-Id: I2c3f5f2a638bf99e379f91b9d1d1c99d491a0153
|