my brain is still trying to understand the implications of Shap's comment
about inferring the non-applicability of gc for a region..

in the meantime, i had a brief conversation with someone who was
responsible for reducing pause times in mono used for scripting for a major
shipping game engine from a major game vendor....

http://meta.stackoverflow.com/questions/6660/podcast-did-the-sims-3-use-mono-for-scripting-or-for-everything

basically he...  improved the Boheme GC dirty page handling, so it could
run incrementally and still finish/make progress; added a software write
barrier to dirty pages (instead of using the mmu), and made amortized the
running of finalizers over multiple cycles to avoid finalizer pauses.. (the
thread which runs finalizers is unspecified and normally separate, but they
were using mono in a single threaded fashion)

with these improvements they were able to get gc pauses, for **just script
objects**, below their 5ms budget. not sure what typical heap sizes were.





-- 
"*In science if you know what you are doing you should not be doing it. *
*In engineering if you do not know what you are doing you should not be
doing it. *
*Of course, you seldom, if ever, see either pure state.*" - Richard Hamming
_______________________________________________
bitc-dev mailing list
[email protected]
http://www.coyotos.org/mailman/listinfo/bitc-dev

Reply via email to