Skip to content

v0.7

Pre-release
Pre-release
Compare
Choose a tag to compare
@vpirogov vpirogov released this 25 Apr 15:01
v0.7
f0860e3

Changes:

  • Improved performance on processors with Intel(R) AVX2 instruction set support
  • Improved performance on processors with Intel(R) AVX512 instruction set support
  • Extended backward propagation optimizations for Intel(R) AVX2 and Intel AVX512 instruction sets
  • Added SGEMM-based reference convolution implementation significantly improving performance for cases not covered by JIT convolution
  • Added JIT version of SGEMM function for Intel(R) AVX2 instruction set. This change allows to build optimized Intel(R) MKL-DNN without binary component.
  • Added backward propagation examples
Assets 11
Loading