core: use binarysearch algorithm to improve method getMaxEffort

[majaziri]

Closes #2851

[codecov]

Codecov Report

Merging #2938 (c06f24b) into dev (53b6a18) will increase coverage by 0.04%.
The diff coverage is 100.00%.

❗ Current head c06f24b differs from pull request most recent head 68a8a1d. Consider uploading reports for the commit 68a8a1d to get more accurate results

@@             Coverage Diff              @@
##                dev    #2938      +/-   ##
============================================
+ Coverage     38.29%   38.34%   +0.04%     
- Complexity     1713     1716       +3     
============================================
  Files           600      600              
  Lines         18342    18355      +13     
  Branches       2513     2516       +3     
============================================
+ Hits           7024     7038      +14     
  Misses        10912    10912              
+ Partials        406      405       -1     

Flag	Coverage Δ
core	83.93% <100.00%> (+0.05%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
...fr/sncf/osrd/envelope_sim/PhysicsRollingStock.java	100.00% <100.00%> (+10.00%)	⬆️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[flomonster]

Few things should be adapted. Can you run a profiler to validate that this change actually improve performances.

[axrolld]

The effort returned is correct, you can consider the pysics part approved.

[flomonster]

Waiting for a benchmark

[majaziri]

benchmark.ods
After 20 iteration, i managed to fill this file to test the 'act' method in StandaloneSimulationEndpoint.
the values provided are in nanoseconds, they are slightly improved compared to the original code.
The first column represents the average of all 20 iterations.

[bloussou]

Which tool did you use to run the iteration ?

[majaziri]

Which tool did you use to run the iteration ?

I calculated the difference of time between the start and the end of the method

[bloussou]

Ok, in the future to test such code you can use tools like java-micro-benchmark : https://www.baeldung.com/java-microbenchmark-harness in order to see jvm optimizations like jvm warmup (because 20 iterations is not enough to see these iterations). We can talk about it later if you have such benchmark to run in the future 😄