Age | Commit message (Collapse) | Author |
|
Also remove it from armnnSerializer and armnnUtils.
In general code within Arm NN should only include exactly what it needs.
This will help keep compile times down, and reduces the need for a
complete rebuild when anything in the public headers changes.
ArmNN.hpp should only be provided as a convenience for user code where the
user doesn't know enough of the API to decide which specific headers
they need.
Signed-off-by: Matthew Bentham <Matthew.Bentham@arm.com>
Change-Id: I4ffcaa7662af4f88a1babf2bf17d03da6f77d665
|
|
!referencetests:218340
Change-Id: If24a604310d0363b1f09b406e4d53ebfeb106aad
Signed-off-by: Derek Lamberti <derek.lamberti@arm.com>
|
|
Change-Id: Ib00f185b431ab74fd9425d8f478bd2ddb182f74b
Signed-off-by: Derek Lamberti <derek.lamberti@arm.com>
|
|
* Create a public API for the common backend files
* Move OutputHandler to armnn internal
* Remove unused headers
Signed-off-by: Matteo Martincigh <matteo.martincigh@arm.com>
Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Change-Id: I3e86d908b021e3561befa9d45158d87d2cbb18c0
|
|
!referencetests:214319
* Reduces arm nn binary size ~15%
* Also fixed test logging black hole issues
Change-Id: Iba27db304d9a8088fa46aeb0b52225d93bb56bc8
Signed-off-by: Derek Lamberti <derek.lamberti@arm.com>
|
|
* Moved the relevant armnnUtils headers to the new location:
include/armnnUtils
* Update the header usage throughout the source code
!android-nn-driver:2387
Signed-off-by: Matteo Martincigh <matteo.martincigh@arm.com>
Change-Id: I2ba15cebcacafad2b5a1a7b9c3312ffc585e09d6
|
|
* Refactoring to enable ProfilingGuid
* Add profiling includes to Android.mk
Signed-off-by: Jan Eilers <jan.eilers@arm.com>
Change-Id: Ieb25e15e3dc302eb42817d824ad8411ac76dcfe8
|
|
This was a bug that meant invalid dot files were produced due to MemCopy
layers having a name including "->".
Change-Id: If9f5b13d433f6a7328bf0ad8c7ec89cdce2462b0
Signed-off-by: Rob Hughes <robert.hughes@arm.com>
|
|
Convolution2d workload
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I0ac08ba4864d48e6f64c4ac645dad8ea850be112
|
|
* Added ScaledInt32PerAxisDecoder implementation
* Added new case for Signed32 in MakeDecoder that returns a
ScaledInt32PerAxisDecoder if the tensor info has multiple
quantization scales
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I8b3c11091644da993044d2a0fe2aba6b06b5af56
|
|
* Add QuantizedSymm8PerAxis to armnn DataType (types.hpp) and
* Add Quantize and Dequantize template for int8 in TypeUtils to be able to compute QSymm8 of the weight
* Create PerAxisIterator for per-axis quantization
* Create QSymm8PerAxisDecoder
* Create QSymm8PerAxisEncoder
Signed-off-by: Keith Davis <keith.davis@arm.com>
Change-Id: Ibcfe0288a197b7ee50b543bdbd77b7edb8a547c2
|
|
* Refactoring of some of the public API headers
Signed-off-by: Matteo Martincigh <matteo.martincigh@arm.com>
Change-Id: I9006fe49945b10a6b83908b05aa4a6556639b491
|
|
breaking some of the nightly builds
* The TensorPrinter must be called regardless of the value of
the output tensor files, it will automatically handle an empty
file path by only printing the results to the console
* Code refactoring
Signed-off-by: Matteo Martincigh <matteo.martincigh@arm.com>
Change-Id: I548ec7cf6d51badf78643c9a6c1c56ea9200142b
|
|
* Moved QuntizeHelper.hpp to armnnUtils
* Reordered parameters for QuantizedVector and added default
values for qScale and qOffset to make life easier when
using the function for non-quantized types such as Float16
Signed-off-by: Aron Virginas-Tar <Aron.Virginas-Tar@arm.com>
Change-Id: I28c263dfa425f1316feccb4116839a84f5d568e5
|
|
* Add ArgMinMax implementation
* Add utility function to get number of elements between axis
* Add utility function to get unsigned axis
* Unit tests for ArgMinMax function
Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Change-Id: I7bc3d610dda9526190187eb87394a8ed7a4b5cdd
|
|
Signed-off-by: Narumol Prangnawarat <narumol.prangnawarat@arm.com>
Change-Id: I54d2416be305db7d3fc15fb3123694ad6740b27d
|
|
* Evaluate model accuracy using category names instead of numerical
labels.
* Add blacklist support
* Add range selection support
Signed-off-by: SiCong Li <sicong.li@arm.com>
Change-Id: I7b1d2d298cfcaa56a27a028147169404b73580bb
|
|
Use it for intermediate tensors on reference backend.
Lays the groundwork for memory management in the reference backend.
Change-Id: I7d3ee132cac31bde70ae6e1b815f4f0b03d550a6
Signed-off-by: Matthew Bentham <Matthew.Bentham@arm.com>
|
|
* Inside method AddImageResult, the number of generated output may
exceed the size of m_TopK. Limit the check to the size of m_TopK.
Signed-off-by: SiCong Li <sicong.li@arm.com>
Change-Id: Ic7513ca27a283ef6d098d5aa0a94b4f17ff04f66
|
|
* Do not use DataLayoutIndexed::GetIndex for weights and inputs,
as it causes a large regression in performance
* It turned out that the calculation of the indexes for the weights
and inputs was the way it was because of an optimization done
many months ago
* Reverted the relevant hunks and added some comments so
we won't make the same mistake again in the future
* Made the GetIndex function inline to speed up other usages
Change-Id: I343b2ef0446993086f58b9dea1f0de0ba2d92216
Signed-off-by: Matteo Martincigh <matteo.martincigh@arm.com>
|
|
for convenience
* Added GetIndex method to DataLayoutIndexed
* Refactored TensorBufferArrayView::Get to use the new method
Change-Id: Iae08b2761bddeda9e935b25e6bc4985f2d386cd3
Signed-off-by: Matteo Martincigh <matteo.martincigh@arm.com>
|
|
Change-Id: Iba91e3f3625639f01d66f81a9f3e419e0e285d66
Signed-off-by: Jim Flynn <jim.flynn@arm.com>
|
|
Change-Id: I4ac325e45f2236b8e0757d21046f117024ce3979
Signed-off-by: Éanna Ó Catháin <eanna.ocathain@arm.com>
|
|
Change-Id: I4352a645badde788b9e33b3675a9cc9c0fc3dc33
Signed-off-by: Jim Flynn <jim.flynn@arm.com>
|
|
Change-Id: I3deb2b9a37e8a8f8f2ed93c64ed0656ae911e24c
Signed-off-by: Jim Flynn <jim.flynn@arm.com>
|
|
Change-Id: I8fbf27b383a821e062f72809cc2e269fcd18851c
|
|
* Remove Equal and Greater from RefElementwiseWorkload
* Create RefComparisonWorkload and add Equal and Greater
* Update ElementwiseFunction for different input/output types
* Update TfParser to create Equal/Greater with Boolean output
* Update relevant tests to check for Boolean comparison
Change-Id: I299b7f2121769c960ac0c6139764a5f3c89c9c32
|
|
Change-Id: Ia0019134f76764cd4fe6ed9dc1423b8aba411d33
|
|
Change-Id: I7359617a307b9abb4c30b3d5f2364dc6d0f828f0
|
|
Change-Id: I3d31d6b72be1984acdb51fd9e7b5488a7aa5d832
|
|
in favor of a type-indepent implementation that takes void-pointers
* The new implementation requires the size of the type to be passed
to the function
* Updated all the usages accordingly
* Removed the old implementation no longer used
!android-nn-driver:469
Change-Id: I37f4e6d62a38fbb8ec8c39bb559a2c54c83365d4
|
|
* Unified ArmNN's weight format to [ M, I, H, W ] for the depthwise convolution
* Added conversion utilities to permute/reshape the weights as appropriate
when using CL and Neon backends
* Updated the reference implementation of the convolution
* Updated the relevant unit tests accordingly
!android-nn-driver:459
Change-Id: I07d0818efa9d1ca1e5dad82983aac1fe78eadb18
|
|
* Allow control inputs from TensorFlow graph but ignore them in ArmNN graph.
* Add utility function to test ArmNN graph structure.
* Add ArmNN graph structure tests in TensorFlow paresr to ensure that control inputs
are ignored in ArmNN graph as well as their inputs that are not used anywhere else.
Change-Id: Ib0ea0d2df85e3fc79b748fa4c9d20e0649352bc1
|
|
* Since DataLayoutIndexed is now required in the TF parser, this
changes move it to the armnnUtils library so that it'll be
accessible by the armnnTfParser
* Modified CMake files and Android.mk files accordingly
Change-Id: Ie2620359ef288aeff64cb9e9bec068a466eee0e9
|
|
* helper methods for creating TensorShape and TensorInfo objects
Change-Id: I371fc7aea08ca6bbb9c205a143ce36e8353a1c48
|
|
Change-Id: Ib11c3d36c7109198da7266955414580e8fb916b5
|
|
Change-Id: I663a0a0fccb43ee960ec070121a59df9db0bb04e
|
|
from src/armnnUtils/ParserPrototxtFixture.hpp
Change-Id: Ief836fb80c4c966f1b76c9429d4a7fb644ceb73d
|
|
in the network
Change-Id: Ide594db8c79ff67642721d8bad47624b88621fbd
|
|
Change-Id: Ie7549fd27378acfa97e68d098e338b8c9d4ea5d2
|
|
* Round to nearest instead of round to intermediate (default)
* Compute Library also round to nearest
Change-Id: I0c71e2ffd28e4e662188fd47d26ec8b538d4ba21
|
|
* Now hand in an optional vector of strings to Optimize
function in which errors/warning messages are placed.
* Optimize function changed to check outputs of each
layer. If they are Quantized 8 bit but the scale has
not been set an error message is added for each such output.
* Unit test modelled on defect report added to exercise the fix.
!android-nn-driver:152483
Change-Id: If9901f5324a516f1ab62858266b38f98dae16201
|
|
* Added RefPermuteFloat16Workload to serve as a fallback when CL
does not support the required permute configuration for FP16
* Move Half.hpp to armnnUtils as the utils library should not be
including private header files from the armnn library
Change-Id: Ibf0f698451e8406f7ed7cce470dab60b6d16361d
|
|
* Concatenation Parser function added to the TfLite Parser
Change-Id: I42a42cd765ea09a30841c66b1942b9e09a876b10
|
|
All changes are the same:
//
// Copyright © 2017 ARM Ltd. All rights reserved.
-// See LICENSE file in the project root for full license information.
+// SPDX-License-Identifier: MIT
//
Change-Id: I37eae011411133663ca9d2b059714d92f8bf8e24
|
|
|
|
|
|
|
|
Change-Id: Id3c11dc5ee94ef664374a988fcc6901e9a232fa6
|