This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

Help! OpenCL implementation for matrix vector multiply

Hello everyone

I am trying to perform matrix vector multiply(both dense and sparse matrix) using Mali, I have looked at ARM ComputeLibrary and I don't really see a specialized sgemv, just wondering if I should just use sgemm for this purpose?

Is there any recommended library I can use for sparse mv?

Thanks

  • Hi,

    The Compute Library supports only vector matrix (It's a lot more efficient than matrix vector from a memory access point of view) so you'll need to transpose and there is currently only support for dense matrices.

    The correct kernel will automatically be selected inside the GEMM function.

    If you've got more questions or need help please contact us directly on Github: github.com/.../issues