Research Article
Sparse Cholesky Factorization on FPGA Using Parameterized Model
Table 4
Performance comparison with HSL_MA87, GPU version of CHOLMOD, and supernode.
| Matrix | HSL_MA87 times (s) | CHOLMOD times (s) | One-GPU times (s) | Two-GPU times (s) | Ours |
| nd3k | 2.02 | 2.92 | 1.27 | 1.12 | 1.96 (, ) | nd24k | 28.56 | 22.17 | 14.63 | 10.12 | 10.08 (, ) | Trefethen_20000b | 12.63 | 8.49 | 5.47 | 3.94 | 3.58 (, ) |
|
|