SPARTA vs Standard GEMM/SpGEMM

[fvella]

• [x] Discussing on block dense MKL multiplication (no batch, explicit loop) compare with MKL sparseXdense multiplication. 32 bit.

• [x] Dense matrix (rectangular, progressively larger 2,4,8,...)

• [x] verify multiplication correctness.

• [ ] Test various small graphs (small RMAT 2^8, facebook, nonsymm small MM). Not for benchmark purposes.

• [x] add timing.

• [ ] Flag per intel_mkl

• [ ] Add picture of density blocks

• [x] Code clean up e.g. removing obj file and CUDA SDK dependencies