Skip to content

Improvement of GEMM and FFT
Compare
Choose a tag to compare
@Mellich Mellich released this 24 Aug 10:25
· 936 commits to master since this release
v0.3
0254370

Next to minor changes, bug fixes, and improvements, this release introduces the following major changes:

  • Optimize FFT for Xilinx up to a size of LOG_FFT_SIZE=9
  • Kernel replication support for GEMM and FFT
  • Remove git submodules and instead fetch dependencies using CMake. This eliminates the need to check out out the git repository to build the benchmarks.
Assets 2
Loading