3 ms·
The current CPU implementation is single core, non-vectorized code. That said, ArrayFire can link with any BLAS / LAPACK library to accelerate the relevant algo
by pavanky 12y ago
The current CPU implementation is single core, non-vectorized code. That said, ArrayFire can link with any BLAS / LAPACK library to accelerate the relevant algorithms.
EDIT: The CUDA and OpenCL backends will obviously be faster than MKL. We'll be adding SSE / AVX support at some point which'll make the CPU backend faster as well.
- weitzj 12y agodid you look at http://www.yeppp.info http://www.yeppp.info for sse/avx support?
- pavanky 12y agoHa! We interviewed the author of that library!
- 14113 12y agoIn a similar vein, how does this compare with (say) CuBLAS or ClMath?
- pavanky 12y agoWe depend on those libraries. We make the API easier to understand while keeping the performance close to the upstream libraries.