search

edoardoCB / gpu_spmv

This repo contains the code to perform the SpMV product with the CSR, K1, AXC, and AXT formats. Using CUDA instructions and the CUSPARSE library.
GNU General Public License v3.0
2 stars 3 forks source link

Testing questions #3

Open dmikushin opened 2 years ago

dmikushin commented 2 years ago

Hi @edoardoCB , thank you for working on this project! Could you please help me to sort out the following questions:

  1. The job scheduling scripts mention two matrices - M25_Serena.csr and MDE.csr. I was unable to find these matrices in the opensource matrix market databases. Could you please provide links or references to their sources?
edoardoCB commented 10 months ago

Hi,

Source for M25_Serena.csr can be found here: https://sparse.tamu.edu/Janna/Serena

MDE.csr is a synthetic matrix with arrow-head sparsity pattern. Typically generated after applying Domain Decomposition technique for parallelization in distributed memory systems for FEM FDM codes.

Best wishes