Is there a practical way to hack/tune the GC to allow for a
useful live data / dynamic-space-size ratio in a setting with
many parallel threads? In particular, more than just 8%?
Background info / testcase / GC data can be found here:
We may be observing an instance of this:
> given a "bad" allocation pattern, it may be that you exhaust the
> heap due to uncollected garbage in older generations before a
> collection deep enough to collect that garbage is triggered.
> [...] In this case forcing a full collection every cycle
> (or every few cycles) should help -- watching the
> PRINT-GENERATION-STATS should tell you if this is the case, and how
> often you should force a full GC.
> (from http://permalink.gmane.org/gmane.lisp.steel-bank.devel/15630 )
In our testcase, how can we tell from the generation stats if a full GC is