Implementation of Sparse-Matrix Vector Multiplication (SpMV) in C and OpenMP for highly parallel architectures such as Intel Xeon Phi
Currently supports the following sparse storage formats:
- CRS aka CSR
- CCS aka CSC
- BCRS aka BCSR
- ELL aka ELLPack format
Desired formats to add support to (no timeline maybe never):
- COO
- HYB (COO+ELL)
- ELL_A similar to ELL except with variable length rows