Take survival rates of young objects into account when choosing old generation limits.

Take survival rates of young objects into account when choosing old generation limits.

Stable high survival rates of young objects both during partial and
full collection indicate that mutator is either building or modifying
a structure with a long lifetime.

In this case we aggressively raise old generation memory limits to
postpone subsequent mark-sweep collection and trade memory space
for the mutation speed.

Committed: http://code.google.com/p/v8/source/detail?r=4938

[Mads Ager (chromium), 2010-06-24 10:35:01]

I think we need to have a strategy for throttling down again before landing this
change?

http://codereview.chromium.org/2809032/diff/1/2
File src/heap.cc (right):

http://codereview.chromium.org/2809032/diff/1/2#newcode666
src/heap.cc:666: old_gen_promotion_limit_ *= 2;
We should make sure to throttle this down again when there is a decreasing
trend.

http://codereview.chromium.org/2809032/diff/1/3
File src/heap.h (right):

http://codereview.chromium.org/2809032/diff/1/3#newcode1318
src/heap.h:1318: return high_survival_rate_period_length_;
Explicitly check greater than zero?

[Vyacheslav Egorov (Chromium), 2010-06-24 11:08:33]

http://codereview.chromium.org/2809032/diff/1/2
File src/heap.cc (right):

http://codereview.chromium.org/2809032/diff/1/2#newcode666
src/heap.cc:666: old_gen_promotion_limit_ *= 2;
On 2010/06/24 10:35:02, Mads Ager wrote:
> We should make sure to throttle this down again when there is a decreasing
> trend.

I wanted just to rely on next full GC. But detecting decreasing trend might be
more reasonable. 

Added check after scavenge.

http://codereview.chromium.org/2809032/diff/1/3
File src/heap.h (right):

http://codereview.chromium.org/2809032/diff/1/3#newcode1318
src/heap.h:1318: return high_survival_rate_period_length_;
On 2010/06/24 10:35:02, Mads Ager wrote:
> Explicitly check greater than zero?

Done.

[Mads Ager (chromium), 2010-06-24 11:38:01]

LGTM