Skip to content

Release 24.02
Compare
Choose a tag to compare
@KevinARM KevinARM released this 22 Feb 13:27
· 1 commit to branches/armnn_24_02 since this release
v24.02
2431f76

Summary

New Features

  • ArmNN to TOSA backend:
    • LeakyRelu Activation support added
    • Quantize support added
    • Maximum support added
    • Split support added
    • Resize Nearest Neighbour support added
  • GpuFsa Backend (Dynamic Fusion)
    • RESIZE/SCALE support added
    • CAST support added
    • POOL2d support added
    • SUB support added
    • ADD support added
    • DEPTHWISE CONVOLUTION 2D support added
    • CONVOLUTION 2D support added
  • Updated to Android NDK r26b.
  • Updated to TensorFlow 2.15.
  • Added optimization to remove reshape operators where possible to CL, Neon and Ref backends.

Bug Fixes

  • Removed implicit sign conversion which could cause compile errors.
  • Fixed memory leak which only happens during profiling and reference Resize workload's align corners is true.
  • Fixed build failures on C++ 14 compilers.
  • Fixed build tool errors when building for Android target.

Other Changes

  • Delegate Unit Tests are now only built for the backends which are being built.
  • Increased end to end testing for two layer and three layer MaxPool2d.
  • In ExecuteNetwork added support to serialize to dot graph for the Arm NN Delegates.

ABI/API Changes

No API breaking front-end changes have occurred during the implementation of 24.02.

No API breaking back-end changes have occurred during the implementation of 24.02.

Note: The minimum API level supported by Arm NN AAR file is 27.

Build Dependencies

Tools Supported Version
Git 2.17.1 or later
SCons 2.4.1 (Ubuntu) 2.5.1 (Debian)
Cmake 3.19.0 (Ubuntu) and 3.19.0 (Debian)
Tensorflow 2.15.0
Onnx 1.6.0
Flatbuffer 23.5.26
Protobuf 3.12.0
Android NDK r26b
mapbox/variant 1.2.0
cxxopts 3.1.1
doctest 2.4.6
fmt 8.3.0
ghc 1.3.2
half 1.12.0
mapbox/variant 1.2.0
stb 2.16
xxd 1.10
Assets 24
Loading