Aleksandr Voron

### Pull Request Readiness Checklist See details at https://github.com/opencv/opencv/wiki/How_to_contribute#making-a-good-pull-request - [ ] I agree to contribute to the project under Apache 2 License. - [ ] To the best of...

CpuGemmConv2d optimization affects performance on Apple M2/M2 Pro

4

PR https://review.mlplatform.org/c/ml/ComputeLibrary/+/10526 makes CpuGemmConv2d slower on Apple M2 / M2 Pro. The numbers below were collected on M2 Pro. On `mobilenet-v2-1.0-224` CpuGemmConv2d takes 3.18 ms before the PR and 4.12...

NEReduceMean accuracy issue for NHWC

4

Output of 'strings libarm_compute.so | grep arm_compute_version': arm_compute_version=v23.02 Build options: {'neon': '1', 'opencl': '0', 'openmp': '0', 'cppthreads': '1', 'examples': '0', 'Werror': '0', 'gemm_tuner': '0', 'reference_openmp': '0', 'validation_tests': '0', 'benchmark_tests': '0',...

Unsigned int overflow in PoolingDepthfirstGeneric

6

**Output of 'strings libarm_compute.so | grep arm_compute_version':** ``` arm_compute_version=v23.02 Build options: {'neon': '1', 'opencl': '0', 'openmp': '0', 'cppthreads': '1', 'examples': '0', 'Werror': '0', 'gemm_tuner': '0', 'reference_openmp': '0', 'validation_tests': '0', 'benchmark_tests':...

[ACL] Potentially redundant check in `acl_init_conf`

1

What is the reason of having such check? https://github.com/oneapi-src/oneDNN/blob/49a1dcd620b2e728f1cee4f44c9a827a0f720411/src/cpu/aarch64/acl_convolution_utils.cpp#L110-L111 If this condition is true and it does not return `status::unimplemented` then `arm_compute::NEDepthwiseConvolutionLayer::validate` will pass successfully later in `init_conf_depthwise` method. Does...

[CPU][ARM] Enable both f16 and f32 kernels for aarch64 and introduce runtime f16 support check

1

Inherited from https://github.com/openvinotoolkit/openvino/pull/22437

[CPU][ARM] Enable NHWC in Reduce

3

NHWC was disabled because of accuracy issue: https://github.com/ARM-software/ComputeLibrary/issues/1044 CVS-114403

Replace `acl_init_conf` check with `acl_init_conf_dw`

NEMeanStdDevNormalizationLayer returns nans for f16 tensors

3

NEMeanStdDevNormalizationLayer returns nans if srd\dst tensors are f16. The issue was reproduced on ACL 23.08 How ACL was built: `scons neon=1 opencl=0 openmp=0 cppthreads=1 arch=armv8.6-a Werror=false validation_tests=1 --jobs=8 os=macos build=native...

ACL has 3d convolution support via NEConv3D: https://arm-software.github.io/ComputeLibrary/latest/_n_e_conv3_d_8h_source.xhtml However, it's not integrated into oneDNN. Is it possible to add NEConv3D support in oneDNN? cc @milpuz01