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
We are looking into using ASV for tracking benchmarks for the reflex project.
Some of the planned benchmarks can take a long time to run (say minutes), and we would ideally want to shard them out. E.g. one job runs suite A, another job runs suite B, another job run suite C, etc.
Would this be a sensible or supported workflow for asv? In particular I'm looking at the results/ dir. How would one merge results/ dirs produced by different jobs? With a valid merge operation, I can fan-in all the results in a single "publish" job at the end. This gives me a single static site to browse.
I am looking at results/benchmark.json in particular.... Each run seems to produce one of these. Is it as simple as just concatenating the list within all the benchmark.json's? :)
The text was updated successfully, but these errors were encountered:
Great work team -
asv
looks nice.We are looking into using ASV for tracking benchmarks for the reflex project.
Some of the planned benchmarks can take a long time to run (say minutes), and we would ideally want to shard them out. E.g. one job runs suite A, another job runs suite B, another job run suite C, etc.
Would this be a sensible or supported workflow for
asv
? In particular I'm looking at theresults/
dir. How would one mergeresults/
dirs produced by different jobs? With a valid merge operation, I can fan-in all the results in a single "publish" job at the end. This gives me a single static site to browse.I am looking at
results/benchmark.json
in particular.... Each run seems to produce one of these. Is it as simple as just concatenating the list within all the benchmark.json's? :)The text was updated successfully, but these errors were encountered: