| 1 |
This directory (and its sub directories) contain a benchmark suite for |
| 2 |
DOLFIN. To run all benchmarks, simply run the script demo.py. |
| 3 |
|
| 4 |
This script runs all executables/scripts named 'bench' found in the |
| 5 |
directory tree and records the result for benchmark found in foo/bar |
| 6 |
under the name 'foo-bar' in logs/bench.log. |
| 7 |
|
| 8 |
Two pieces of information are extracted from each benchmark: |
| 9 |
|
| 10 |
1. A short description of the benchmark. This is extracted as the |
| 11 |
first line of output from the benchmark. |
| 12 |
|
| 13 |
2. One or more timings. These are extracted from output in the form |
| 14 |
|
| 15 |
BENCH <totaltime> |
| 16 |
BENCH <subtask> <subtime> |
| 17 |
|
| 18 |
If no output is given (or <totaltime> is not given), then the total |
| 19 |
running time of the program is recorded. |
| 20 |
|
| 21 |
Important notice: To run the benchmarks correctly, you need to compile |
| 22 |
DOLFIN with option --enable-optimization. Compiling DOLFIN with |
| 23 |
--enable-debug will slow down some of the benchmarks considerably. |