Title: Automatic performance analysis of code changes

 

Automatic regression detection is a difficult problem because there are many moving parts, the infrastructure can have problems, benchmarks are unstable, performance changes may be very small and exclusive resources is costly.
It is also very expensive for team members to inspect manually all the benchmarks with each code change.

Simplifying PR performance analysis

Currently, if one is working on a performance sensitive change in the compiler team, the truffle team or any language team, one has to manually trigger the benchmarks of interest and manually compare each benchmark results with the baseline in bench server. This task is time consuming and error prone.

Solution we propose:

• Introduce a CI job that would intensively run all benchmarks of interest for a given team and automatically publish a report in the PR with statistical analysis of the results compared to the latest benchmark results.
• Reporting should reuse the weekly performance report tools and a new way of reporting results should be introduced.
• We will integrate ML-driven approaches to decide or propose to developers which jobs need to be scheduled again.

This job code is utilized for the majority of our temporary hires. The individual is performing hourly job duties as defined under the Fair Labor Standards Act.





As part of Oracle's employment process candidates will be required to successfully complete a pre-employment screening process. This will involve identity and employment verification, professional references, education verification and professional qualifications and memberships (if applicable).