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
I've started to write benchmarks for Gammapy and decided that mostly high-level benchmarks running compute-heavy end-user analysis tasks are of interest and for each of those I'm interested in:
CPU time
Wall clock time
Memory use
@mdboom Would it make sense to implement a mechanism that makes it easy to do several benchmarks (namely the three I mentioned) without having to write boilerplate code? Maybe via a decorator listing which measurements I want?
The text was updated successfully, but these errors were encountered:
I have a similar need : a single benchmark for the spectroscopy RADIS code can take up to 250s ; and computing it twice for PeakMemory and CPU-wall-time takes a lot of time.
Would it be possible and still accurate to measure both the PeakMemory and the CPU-wall-time on a single run ?
I've started to write benchmarks for Gammapy and decided that mostly high-level benchmarks running compute-heavy end-user analysis tasks are of interest and for each of those I'm interested in:
@mdboom Would it make sense to implement a mechanism that makes it easy to do several benchmarks (namely the three I mentioned) without having to write boilerplate code? Maybe via a decorator listing which measurements I want?
The text was updated successfully, but these errors were encountered: