Maksim Kutakov

>Change stride to 1 unit because of brgemm limitation What happens if the stride is different from 1?

>Change stride to 1 unit because of brgemm limitation. What exactly happens if the stride is other than 1?

PR needs internal performance validation

Internal performance validation passed successfully

> The gemm conv kernel diverges most with the upstream version, because fork enabled jit_pp kernel for jit gemm kernel but upstream still use ref_pp kernel. > > https://github.com/openvinotoolkit/oneDNN/blob/26633ae49edd4353a29b7170d9fcef6b2d79f4b3/src/cpu/gemm_convolution.hpp#L77 >...

> > > The gemm conv kernel diverges most with the upstream version, because fork enabled jit_pp kernel for jit gemm kernel but upstream still use ref_pp kernel. > >...

@dmitry-gorokhov, the PR is ready to be merged. Could you please make a final review?

@tiger100256-hu , please resolve the merge conflicts.

@EgorDuplensky , could you please review?

@nshchego , the main question is why cannot we implement `std::basic_streambuf` over the mapped memory block? If we had such an implementation, we could reuse the most of the existing...