When doing an analysis of application performance bottlenecks, there is a lot of guesses to make and tools which must be used (eg. valgrind, perf, oprofile, memprof or clouseau). After doing a considerable amount of work, the developer must consolidate all the results into a result which explains the loss of performance.
Clouseau is an example of how to implement remote introspection of a process to add instrumentation and get more information about EFL usage; this allows a client to list and display the attributes of all the objects used in the process.
Building on Clouseau's methods, this project proposal would be to create a framework and application for tracking, in real time, where applications spend processing and rendering time in a process. It would enable developers to easily spot potential performance bottlenecks and code which should be optimized.
- Mentors :
- Cedric Bail
- Daniel Juyung Seo
- Skills required :
- Strong C background
The difficulty of this project is assumed to be a hard one.