The development of a version of aalcalc in Rust indicates that there is room for significant improvement in the performance of this component. Running aalcalc through a profiler reveals that a significant proportion of run time (around 23%) is used in indexing the input files from summarycalc. It would be worth investigating whether there is scope for optimisation.
Version / Environment information
ktools v3.9.5.
Example data / logs
Output of flat profile from gprof profiler, showing functions that contribute > 10% of run time:
Flat profile:
Each sample counts as 0.01 seconds.
% cumulative self self total
time seconds seconds calls s/call s/call name
24.10 514.45 514.45 read
23.16 1008.78 494.33 4947282156 0.00 0.00 operator<(summary_period const&, summary_period const&)
11.68 1258.08 249.29 lseek
Ordering the summary index and reading the input files appear to be the main bottlenecks.
benhayes21
commented
1 year ago
Issue Description
The development of a version of
aalcalcin Rust indicates that there is room for significant improvement in the performance of this component. Runningaalcalcthrough a profiler reveals that a significant proportion of run time (around 23%) is used in indexing the input files fromsummarycalc. It would be worth investigating whether there is scope for optimisation.Version / Environment information
ktools v3.9.5.
Example data / logs
Output of flat profile from
gprofprofiler, showing functions that contribute > 10% of run time:Ordering the summary index and reading the input files appear to be the main bottlenecks.