| 1 |
This benchmark tests parallel speedup for assembly and solve. |
| 2 |
It is organized as follows: |
| 3 |
|
| 4 |
assemble-poisson: simple program for assembling Poisson matrix |
| 5 |
solve-poisson: simple program for solving Poisson's equation |
| 6 |
|
| 7 |
submit-bench: script for submitting jobs using dolfin_utils.pjobs |
| 8 |
analyse-bench: script for analysing output from submit-bench |
| 9 |
|
| 10 |
bench: script used on benchbot for simple checking of speedup |
| 11 |
|
| 12 |
Possible future improvements: |
| 13 |
|
| 14 |
1. MUMPS vs UMFPACK gives skewed results (MUMPS faster?) |
| 15 |
2. Time assemble_cells and apply separately |
| 16 |
3. Use barriers between timings |