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
My plotters ran slowly yesterday, so I started a few tests to see if there are easy gains possible... I see here that there is some optimization for the basket size etc.
Would these settings interfere with the compression settings? We seem to be currently using ZLIB level 1, with LZMA level 7 the file size goes down by about 35% (compression factor from 3.5 to more than 5). This would be a fairly aggressive setting (uncompressing costs quite some CPU), but I think the plotters are usually limited by IO anyway - I am running some more tests to verify that.
The text was updated successfully, but these errors were encountered:
Decompressing 5GB with LZMA adds about 2min of CPU time... that doesn't seem excessive to me, but given how little the plotters need (a bit over 1min in this case) this would make them slower when load on the storage is low. I will try to collect a bit more statistics from slurm and submit a PR to make it configurable, but not change the default (yet), shout if you disagree ;-) .
My plotters ran slowly yesterday, so I started a few tests to see if there are easy gains possible... I see here that there is some optimization for the basket size etc.
Would these settings interfere with the compression settings? We seem to be currently using ZLIB level 1, with LZMA level 7 the file size goes down by about 35% (compression factor from 3.5 to more than 5). This would be a fairly aggressive setting (uncompressing costs quite some CPU), but I think the plotters are usually limited by IO anyway - I am running some more tests to verify that.
The text was updated successfully, but these errors were encountered: