You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It's good to have an estimate of the amount of time to run the minitest notebook, but it would also be good to have an estimate of the number of MPP hours needed as well. It is possible that users with small allocations may burn though a significant fraction in just one run through.
The text was updated successfully, but these errors were encountered:
Based on a run through the notebook yesterday, I got about 540 "raw core hours". If that translates anything at all close to 1:1 to "MPP hours", then a single run of the notebook can burn through half a standard user's allocation.
On cori the conversion factor is 2.5 so running the notebook once would exceed the default 1000 core hours. That limit was purposefully small so that people would be purposeful in understanding the amount of time they are requesting, but we could make a larger default. The default for developers known to be working at NERSC is 10k. We could document this in the notebook as making sure that you have >1500 MPP hours available before starting.
Although this does sound like a lot for a test, note that this is only 10 tiles covered by 18 exposures, but that is 50000 targets = almost a week of e/BOSS running at full efficiency.
It's good to have an estimate of the amount of time to run the minitest notebook, but it would also be good to have an estimate of the number of MPP hours needed as well. It is possible that users with small allocations may burn though a significant fraction in just one run through.
The text was updated successfully, but these errors were encountered: