Task #3583
Profiling of persistence framework / CR
| Status: | Closed | Start date: | 2009-06-05 | |
|---|---|---|---|---|
| Priority: | Must have | Due date: | ||
| Assigned To: | Karsten Dambekalns | % Done: | 0% |
|
| Category: | Persistence | |||
| Target version: | TYPO3 Flow Base Distribution - 1.0 alpha 2 | |||
| Sprint: | Has patch: | |||
| PHP Version: | Complexity: |
Description
We need to have an intensive profiling session for figuring out where the weaknesses currently are regarding speed of persistence.
index.php
- index.php with profiling code included
(2.4 kB)
Related issues
History
#1 Updated by Karsten Dambekalns about 6 years ago
- File index.php added
- Status changed from New to Closed
I worked on this using XHProf (http://mirror.facebook.net/facebook/xhprof/) as Xdebug is awefully slow (albeit it's traces provide more detail - which we currently don't really need). I used the blog with 100 posts in it and the attached index.php for profiling.
Make sure to turn off SpeedStep (or similar) technologies and only use CPU time for comparison, wall time may differ massively between subsequent runs. On a Mac it seems only CoolBook (http://www.coolbook.se/CoolBook.html) is available for this.
Results:- SQLite is awefully slow compared to MySQL
- Object creation takes a lot of time, see #3740 for a followup task
I could bring the CPU time down by about a quarter when replacing a few object creation calls within the persistence layer with new calls, so speeding up object creation seems to be advisable.