wiki:u/madams/WTSRunStatistics_Stampede

Version 9 (modified by madams, 10 years ago) ( diff )

Theoretical cpu*hrs

We have 295,867 sus available for our account. Jobs submitted to normal queue. First batch on 512 cpus for 15 hours. Took a sampling of the four times by using ls -ltr (beginning, the middle, two at the end) between the selected chombos.


We have 1,440 minutes in a day. So we can calculate out the average rate:


For both of these runs we want to have 200 frames. Taking the difference from what we can see from the first rows of our jobs listed below, that means we have 177 frames left for the hydro case, and 182 for the mhd case. Using our average rates from above, that means it will take approximately 3 days to finish the hydro run, and 5 days to finish the mhd run (based on these values). There are 72 hours in 3 days and 120 hours in 5 days. One can see I'll be under the SU limit for the following cpu*hrs.

512 cores

For the hydro case:

For the hydro case:

Total:

Stampede Run Statistics

Hydro/MHD Path Date Row Editted Frames Completed/Total Filling Fracs Walltime Taken/Left, (0 days == Ready for Restart) (1-2) Info, (3) Message allocations Framerate (hrs/frame), No. cores Notes
Hydro /scratch/03517/tg828161/ProductionRuns_WireTurbulence/hydro 05-11-2015 23/200 0.038 0.622 3.5 days/0 days 34.6 gb, 164.9 mb, 56.9 mb ~0.5hrs/frame, 512 cores Chombos started May 7th at ~15:46.
/scratch/03517/tg828161/ProductionRuns_WireTurbulence/hydro/Restart 05-11-2015 23/200 TBA TBA TBA , 1024 cores Using previous run, going to do a study on number of cpus needed for sus left.
MHD /scratch/03517/tg828161/ProductionRuns_WireTurbulence/mhd 05-11-2015 18/200 0.091 0.865 20.4 days/0 days 225.3 gb, 585.0 mb, 240.9 mb ~0.75hrs/frame, 512 cores Chombos started May 8th at ~9:45.
/scratch/03517/tg828161/ProductionRuns_WireTurbulence/mhd/Restart 05-11-2015 18/200 TBA TBA TBA , 1024 cores Using previous run, going to do a study on number of cpus needed for sus left.
Note: See TracWiki for help on using the wiki.