Wiki source code of Benchmarking
Show last authors
| author | version | line-number | content |
|---|---|---|---|
| 1 | == WIP == | ||
| 2 | |||
| 3 | == TVB-INVERSION 1.0.0 == | ||
| 4 | |||
| 5 | === Sampling priors - Remote execution === | ||
| 6 | |||
| 7 | ==== ==== | ||
| 8 | |||
| 9 | ==== 1. Execution times ==== | ||
| 10 | |||
| 11 | In this section we provide some benchmarks regarding the execution times for the sampling priors step within tvb-inversion workflow, where we need to run a large number of simulations. | ||
| 12 | |||
| 13 | These have been computed on DAINT-CSCS HPC, on a single node, with different configurations for the number of simulations and the number of workers. | ||
| 14 | |||
| 15 | |||
| 16 | |=(% scope="row" %)((( | ||
| 17 | Model | ||
| 18 | )))|=((( | ||
| 19 | Sim length (s) | ||
| 20 | )))|=Nodes|=Nr simulations|=Nr workers|=Execution time (hh:mm) | ||
| 21 | |=MontbrioPazoRoxin|30|100|((( | ||
| 22 | 30 | ||
| 23 | )))|30|00:17 | ||
| 24 | |=MontbrioPazoRoxin|30|100|200|20|01:08 | ||
| 25 | |=MontbrioPazoRoxin|30|100|300|30|01:10 | ||
| 26 | |=MontbrioPazoRoxin|30|100|400|40|01:18 | ||
| 27 | |=MontbrioPazoRoxin|30|100|500|50|01:34 | ||
| 28 | |=MontbrioPazoRoxin|30|100|500|55|01:30 | ||
| 29 | |=MontbrioPazoRoxin|30|100|600|55|01:45 | ||
| 30 | |=MontbrioPazoRoxin|30|100|600|60|OOM | ||
| 31 | |=MontbrioPazoRoxin|60|100|500|55|((( | ||
| 32 | OOM | ||
| 33 | ))) | ||
| 34 | | | | | | | | ||
| 35 | | | | | | | | ||
| 36 | | | | | | | | ||
| 37 | | | | | | | | ||
| 38 | | | | | | | | ||
| 39 | | | | | | | | ||
| 40 | | | | | | | | ||
| 41 | | | | | | | | ||
| 42 | | | | | | | | ||
| 43 | | | | | | | | ||
| 44 | | | | | | | | ||
| 45 | | | | | | | | ||
| 46 | | | | | | | | ||
| 47 | | | | | | | | ||
| 48 | | | | | | | | ||
| 49 | |||
| 50 | |||
| 51 | ==== 2. Limitations ==== | ||
| 52 | |||
| 53 | * Reaching he memory limit on the CSCS node. | ||
| 54 | ** For a simulation of 30 seconds, we can fit 55 parallel workers in the available memory. | ||
| 55 | ** For a simulation of 60 seconds, we can fit 40 parallel workers in the available memory. | ||
| 56 | * Reaching the maximum connections on CSCS could happen during a run | ||
| 57 | ** connect once to launch the job | ||
| 58 | ** connect multiple times during the monitoring step to check the status of the job | ||
| 59 | ** connect once to stage out results |