Table 2.
Imaging runtime on the “Dawn” cluster, using both CPU sockets of each of the 15 compute nodes.
| Imaging method | Run time | Factor | Throughput |
|---|---|---|---|
| (KVis/s) | |||
| w-stacking default | 4 min | 1 | 720 |
| (o = 63, nwl = 32, p = 1.2) | |||
| w-stacking minimum | 7 min | 1.75 | 410 |
| (o = 4 K, nwl = 500, p = 1) | |||
| w projection | 27 min | 6.75 | 110 |
| Direct transform | 38 h | 570 | 1.3 |
| IDG CPU | 4 min | 1 | 720 |
| IDG GPU | 16 s | 0.07 | 11000 |
Notes. The IDG GPU imager additionally uses one of the Tesla K40 GPUs on each node. The factor is the relative time with respect to w stacking with default settings. The last column specifies the visibility throughput for a single node.
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.