20,741
edits
User talk:Philosopher/Nasal introspection: Difference between revisions
Jump to navigation
Jump to search
m
User talk:Philosopher/Nasal introspection (view source)
Revision as of 18:03, 16 August 2013
, 16 August 2013no edit summary
m (→Race Condition ?: clarification ?) |
mNo edit summary |
||
| Line 1: | Line 1: | ||
== Logging overhead observations == | |||
FYI: I set max_threads to 0 and disabled all logging output, and ended up with a pretty steady frame latency of 125ms here, which suggests that there's also a fair overhead due to console I/O (which is a known issue in FG, due to FG's sequential main loop). It would sometimes spike to ~500ms, which is in sync with the GC stats (which end up not being purged due to the lack of processing threads running in that mode). | |||
Like you mentioned previously, for some weird reason, performance stats are massively affected once the aircraft is moving, even when just running the "zero eye candy" profile. | |||
I did get nice GC stats when exiting (after profiling just the test func): | |||
<pre> | |||
**** Nasal garbage collection statistics: objects: 837843, references: 1915555 | |||
Exit | |||
Waiting | |||
**** Nasal garbage collect: { objects: 289897, references: 658582, n: 1 } | |||
**** Nasal garbage collect: { objects: 498513, references: 1129243, n: 1 } | |||
Done | |||
_dump | |||
</pre> | |||
== Profiling individual threads via call() == | == Profiling individual threads via call() == | ||
<pre> | <pre> | ||