Table 3
Mean wall-clock time spent by the program to advance the solution by one time step with RK3 in a single cell of the computational grid, in units of µs.
| 1283 | RUSANOV | HLLC | LHLLC |
|---|---|---|---|
| LIN | 1.47 | 1.68 | 1.70 |
| LIN+VL | 1.57 | 1.80 | 1.89 |
| PAR | 1.47 | 1.74 | 1.78 |
| PPM84 | 4.12 | 4.30 | 4.31 |
| PPM08 | 3.57 | 3.75 | 3.83 |
| PSH | 2.01 | 2.36 | 2.41 |
| 2563 | RUSANOV | HLLC | LHLLC |
| LIN | 1.60 | 1.73 | 1.77 |
| LIN+VL | 1.73 | 1.87 | 1.88 |
| PAR | 1.60 | 1.81 | 1.86 |
| PPM84 | 4.23 | 4.40 | 4.47 |
| PPM08 | 3.79 | 4.03 | 4.00 |
| PSH | 2.24 | 2.47 | 2.52 |
Notes. The numbers provided in the tables are averages of five measurements, each of which is obtained by evolving the setup described in Sec. 3.2 for 100 steps. Relative errors in the averages (in the sense of 1σ) are smaller than 2% in all cases. Every simulation run for this analysis is MPI parallelized using grids with 323 cells per task. Computations are performed on 2.3 GHz Intel Xeon, Skylake-based processors.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.