Age | Commit message (Collapse) | Author |
|
* Add implementation of IsUnidirectionalSequenceLstmSupported to RefLayerSupport
* Add RefUnidirectionalSequenceLstmWorkload
* Refactor Lstm to be able to use for Lstm and SequenceLstm
* Unit tests
Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Change-Id: Ibc066d213213a11b955dfefbe518de643298ba0c
|
|
* Updates the runtime options with a CustomAllocatorMap which allows to define a CustomAllocator for specific backends
* Change IBackendInternal interface to use a shared pointer to a custom allocator
* Update ClBackend.hpp/cpp to use the CustomAllocator
* Adds an example application and unit test which uses a CustomAllocator for GpuAcc
* Refactor of the interface to use MemorySource instead of the user Mapping cl_mem directly
* Modify the BackendRegistry to also hold a registry of CustomAllocators
* BackendRegistry Deregister will also deregister any allocators associated with that backend id
* set_global_allocator within the BaseMemoryManager so that it always matches the currently used allocator
Signed-off-by: Jan Eilers <jan.eilers@arm.com>
Change-Id: I156d819686021865f4375e6cb7a5c5dec8fee9e8
Signed-off-by: David Monahan <david.monahan@arm.com>
|
|
* Add switch for network details during profiling
Signed-off-by: Keith Davis <keith.davis@arm.com>
Change-Id: I8bd49fd58f0e0255598106e9ab36806ee78391d6
|
|
Signed-off-by: Keith Davis <keith.davis@arm.com>
Change-Id: I92dd410da7ad633a46d025fdc2b26093041c439b
|
|
Signed-off-by: Keith Davis <keith.davis@arm.com>
Change-Id: I7be77712a9f790928219ce91222d46cc766ab9dd
|
|
* Pass m_shapeInferenceMethod to OptimizerOptions in ExecuteNetwork
Signed-off-by: Finn Williams <Finn.Williams@arm.com>
Change-Id: I90280fb7629092d3b66e8a3968ca9e35a0df854a
|
|
* Updated FullyConnected layer member variables when cloning as
some backends still require them.
* Added SetConstant call when using deprecated AddFullyConnectedLayer
method to ensure backwards compatibility.
* Added SetConstant to SimpleSample to ensure it runs on all backends.
Signed-off-by: Matthew Sloyan <matthew.sloyan@arm.com>
Change-Id: Ie7b4e4b868f23f8fcf9c41ffd12e2ea9ea53afca
|
|
* Remove deprecated code from public examples
Signed-off-by: Francis Murtagh <francis.murtagh@arm.com>
Change-Id: Ia6c4f8d7f8c477b776d96e66085996487e8fdc75
|
|
Signed-off-by: Matthew Sloyan <matthew.sloyan@arm.com>
Change-Id: I3a3e5b017c307e374457b18712585dce210f8922
|
|
Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
Change-Id: Ifbd13e9e7a8e238c2bb1d2cb9c48fa159eac3bba
|
|
Replace these with a std::vector for portability
Change-Id: Ia6b0dc9360446ef8bba0baa88c61b4c400fcd799
Signed-off-by: Rob Hughes <robert.hughes@arm.com>
|
|
Signed-off-by: Francis Murtagh <francis.murtagh@arm.com>
Change-Id: Ia44436fc0e74ad57eb3d73cdb31956dffdc45493
|
|
Signed-off-by: Finn Williams <Finn.Williams@arm.com>
Change-Id: I2a7a9b7b21848e2b3258788f06b01159f95a0d6c
|
|
* Constant weights and biases are now stored as Constant layers.
* Updated Serializer, Deserializer and unit tests to reflect this.
* Updated TfLiteDelegate, TfLiteParser and OnnxParser.
* Updated Schema with IsConstant and ConstantTensorsAsInputs.
* Updated Ref backend to handle constant weights and
bias as inputs rather than reading from member variables.
* Added dynamic or constant input EndToEnd tests.
!android-nn-driver:5959
Signed-off-by: Matthew Sloyan <matthew.sloyan@arm.com>
Change-Id: Ibf3cf437df1100e4b322b0d303c575c6339f9696
|
|
* Fixed unit test failures happening on threads.
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: I2a6048f75ece4a9f4c2116306838ff55385aabe7
|
|
Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
Change-Id: I00a3490fd34f51f9edb3e46b90065bfc37c7946a
|
|
Change-Id: I6368925a19ee11d67b1b153cf1e7dc64eca11077
Signed-off-by: Jim Flynn <jim.flynn@arm.com>
|
|
* Created Arm NN Operator list page.
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: I50dbdb78ea9714b0551e023f6ea7f017ad4b47bb
|
|
* Add GUID as field to layer details and profiling events
* Add Optional GUID param to existing tests
* Improve Details macro to be inline function
* Fix some formatting
Signed-off-by: Keith Davis <keith.davis@arm.com>
Change-Id: I66f192a90a7642b3ee8e7dda0d3f428cce002581
|
|
* Fixed PreCompiledObject type (was const void** instead of void*)
* Fixed bug where a new shared_ptr was being created instead of allowing
std::move to convert the unique_ptr into a shared_ptr.
* Improved tests to ensure that the original pointer is maintained.
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: I3f50d51775ee0180c894c2843ed7bc990a771dea
|
|
* Add new ProfilingDetails class to construct operator details string
* Add new macro which helps append layer details to ostream
* Add ProfilingEnabled to NetworkProperties so that profiling can be
realised when loading the network
* Add further optional info to WorkloadInfo specific to convolutions
* Generalise some JsonPrinter functions into JsonUtils for reusability
* Remove explicit enabling of profiling within InferenceModel as it is
done when loading network
* Add ProfilingDetails macros to ConvolutionWorkloads for validation
Signed-off-by: Keith Davis <keith.davis@arm.com>
Change-Id: Ie84bc7dc667e72e6bcb635544f9ead7af1765690
|
|
Change-Id: I03ef0571b24ad2bddfabf3bc5b11b614e7d7c8b4
Signed-off-by: Rob Hughes <robert.hughes@arm.com>
|
|
Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
Change-Id: I2aae71349c14e2cdf43b241da4a4acf7ddef0755
|
|
* Allows to supply different input data for every execution of a
model when using the 'iterations' option in ExecuteNetwork
* Removes the option 'simultaneous-iterations' because it's functionallity
is now covered by 'iterations'
* Adds a deprecation warning message to notify users
* Little refactor of warning messages
Signed-off-by: Jan Eilers <jan.eilers@arm.com>
Change-Id: Ib3ab0d6533f6952bfee20d098a890b653c34cc12
|
|
Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Change-Id: Ic1c56a57941ebede19ab8b9032e7f9df1221be7a
|
|
Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
Change-Id: I6072afd1bb58ae931560c604fddf1efadce634f5
|
|
Signed-off-by: Francis Murtagh <francis.murtagh@arm.com>
Change-Id: I977a75b403a879603f0291a58be21a227201a273
|
|
* Avoid allocation of auxiliary memory in CpuGemmConvolution
Signed-off-by: Kevin May <kevin.may@arm.com>
Change-Id: I8c7e0a7639949140677fac24256367a90ab63fdb
|
|
* Update all documentation for removing BOOST library dependency
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: Ie21ff096a11b7a623b1b9bb6f71eb63ea6e7b615
|
|
* Added support for qasymms8 (int8) to ImageTensorGenerator
* Added qasymmu8 as alias for qasymm8 in ImageTensorGenerator
* Added support for qasymms8 (int8) to ExecuteNetwork
* Added qasymmu8 as alias for qasymm8 in ExecuteNetwork
* Set tflite to be the default model format in ImageTensorGenerator as
it's the only supported model format.
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: Ieda7b78e668ea390e3565cd65a41fe0a9c8a5b83
|
|
Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
Change-Id: I9c0fcc079553fcf5e97cc06b9e4274f2189c0852
|
|
Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Change-Id: I57bcbdec3eb0155f41af0fe7d6abf9bac2ec86eb
|
|
* Adds logic to the Runtime to activate protected mode
* Adds ProtectedContentAllocation backend capability to ClBackend
It's not fully activated yet because the CustomAllocator is
missing. Will print an error message and won't register the backend
but won't fail.
* Extends IBackendInternal with an UseCustomAllocator function.
* Adds related unit tests
Signed-off-by: Jan Eilers <jan.eilers@arm.com>
Change-Id: I64f465c5800eb104aa90db1bbf772a4148b5072f
|
|
* Added ICustomAllocator.hpp to include/armnn/backends/
* Added the ability to specify an ICustomAllocator to the CreationOptions
Signed-off-by: David Monahan <david.monahan@arm.com>
Change-Id: I820e468b691aa032c9bd2c1e1257dc9d02f981d9
|
|
* tflite shape_signatures will now be the preferred way to detect dynamic tensors
* add test utility to the parser that converts a model's tensors to dynamic
* by default tests will run a dynamic version of the model in addition to the original
* fix dynamic shape inference of unpack operator
* reactivate and fix quantize test
* add shape inference to expand dims
Signed-off-by: Finn Williams <Finn.Williams@arm.com>
Change-Id: If11ba19d813cd3590707583dff1e4eb0e6412a1d
|
|
!android-nn-driver:5966
Change-Id: Ice0b4d2872bb0e09bfc0763034a206c3a8f24af4
Signed-off-by: Rob Hughes <robert.hughes@arm.com>
|
|
Tests for TfLiteParser are not working
when the expected outputs have boolean type
Signed-off-by: Bruno Goncalves <bruno.slackware@gmail.com>
Change-Id: I16890f82e8e581f53e6e8464668c5adf3374bf2f
|
|
E.g. Equal, NotEqual, Greater, GreaterOrEqual, Less and LessOrEqual
Signed-off-by: Bruno Goncalves <bruno.slackware@gmail.com>
Change-Id: Id56ef3cc19cc5c5daa19354010c9f25766e5fd00
|
|
Refactor code around conditionally-compiled tests to avoid declaring
empty test suites, as this can cause unused function warnings to be
issued under certain combinations of compiler, warning level, and
doctest version.
Signed-off-by: Matthew Bentham <matthew.bentham@arm.com>
Change-Id: Ib501aef80475538a725b857d9c31d1d2f96b124d
|
|
As the bias decoder is optional, pass it as a pointer to the FullyConnected
function instead of by reference.
Signed-off-by: Matthew Bentham <matthew.bentham@arm.com>
Change-Id: I0bd2e601015e7fa1ae21b50f6c3dd7eac4176554
|
|
Because the maximum value of an int32_t isn't exactly representable
by a 32-bit float, casting it to a float produces a greater value,
and then casting that back to an int32_t is undefined. (In tested
implementations the cast saturates, so accidentally the correct
result is returned).
Using double for the intermediate value instead gives a high
enough precision that this test case can be safely expressed.
Silences warning from Undefioned Behaviour Sanitiser.
Signed-off-by: Matthew Bentham <matthew.bentham@arm.com>
Change-Id: I6fe3e89cb6ac695049c92f3b75aca0c14d0e14aa
|
|
* Fix type of literals
* Add explicit casts
* #include missing headers
* Replace use of non-standard u_int8_t
* Remove name of unused variables
* Fix DLL export macro - this was broken when some of the exporting headers
were moved to a different library.
Change-Id: Ie569c8df41a077b46c608798f39526352e2aeb3a
Signed-off-by: Rob Hughes <robert.hughes@arm.com>
|
|
* Remove all references to boost from the ArmNN build files:
* Caught one remaining include in TensorTest.cpp.
Signed-off-by: Colm Donelan <Colm.Donelan@arm.com>
Signed-off-by: Sadik Armagan <sadik.armagan@arm.com>
Change-Id: Ib8c8014cf471aa0024c977ccb23ebf96b0236aca
|
|
* This change is caused by ACL patch: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5855
Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
Change-Id: I222ff6dea155be38899145a5d096093db1b5e716
|
|
Signed-off-by: Matthew Bentham <matthew.bentham@arm.com>
Change-Id: I5889a5f44c0c22da6421d9bd06494c75d74e0440
|
|
Fail test if exception is not thrown.
Remove unnecessary references to CpuAcc backend.
Remove excess trace message.
Signed-off-by: Matthew Bentham <matthew.bentham@arm.com>
Change-Id: I22d30b47a092ac2e66f5987c656804cd31b04b21
|
|
In the assertion part of PolymorphicDowncast (and PolymorphicPointerDowncast)
there was an unnecessary static_cast of the source pointer to the destination
type. This is unnecessary because the implicit conversion of the result pointer
back to the base type is sufficient to check for correctness of the downcast,
and potentially harmful because if the downcast is actually incorrect the
behaviour of static_cast is undefined (not actually known to be a problem
in any tested implementation).
Fixes warnings detected by ubsan.
Signed-off-by: Matthew Bentham <matthew.bentham@arm.com>
Change-Id: I9126288cbb06564009f94e57f6ca4688fc3b53c4
|
|
* Add ExpandDims tests in tfliteparser
* Add support for negative axis to squeeze
Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Signed-off-by: Finn Williams <Finn.Williams@arm.com>
Change-Id: I604c9b4ac6514895e9e3d4d85c2937e797d288e0
|
|
* Optimization that searches for layers with ConstantLayers as inputs.
* The layer member variables are then redirected to these ConstantLayers.
Signed-off-by: Matthew Sloyan <matthew.sloyan@arm.com>
Change-Id: I24a2bf0e8575b808343e0bbe3897b344e94796ad
|
|
* Added IOptimizedNetwork constructor that takes another
IOptimizedNetwork and a ModelOptions.
* Changed PreCompiledLayer to use shared_ptr rather than unique_ptr
to store the PreCompiledObject (no interface changes).
* Added unit tests to ensure that PreCompiledLayer::Clone() clones
the pointer to the PreCompiledObject correctly.
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Change-Id: I3ef56055e0d189ffce9e651882d34da16c70a240
|