Table 2 |.
emClarity processing steps | Binning | GPU card | # GPU units | Time |
---|---|---|---|---|
| ||||
ctf estimate | 1 | Tesla V100 | 1 | 25 m |
template search | 8 | Tesla V100 | 1 | 1 h |
init cycle 0–2 | 6 | Tesla V100 | 4 | 40 m |
tomoCPR-1 cycle 3–5 | 5 | Tesla V100 | 4 | 2.5 h |
tomoCPR-2 cycle 6–8 | 4 | Tesla V100 | 4 | 3 h |
Classification | 4 | Tesla V100 | 4 | 40 m |
cycle 9–10 | 4 | Tesla V100 | 4 | 1.5 h |
tomoCPR-3 cycle 11–13 | 3 | Tesla V100 | 4 | 2 h |
tomoCPR-4 cycle 14–16 | 2 | Tesla V100 | 4 | 3 h |
tomoCPR-5 cycle 17–18 | 1 | Tesla V100 | 4 | 10 h |
avg FinalAlignment | 1 | Tesla V100 | 4 | 1 h |
cisTEM reconstruct/refine | 1 | Tesla V100 (cpu) | 12 cpu cores | 1.5 h |
Note: Each tomogram is divided into multiple sub-regions (one VLP/sub-region), which are processed in parallel.