Re: gc --aggressive

[Jeff King <peff@peff.net>]

On Tue, May 01, 2012 at 01:59:08PM -0400, Nicolas Pitre wrote:

> One final quick test if you feel like it: I've never been sure that 
> the last comparison in type_size_sort() is correct.  Maybe it should be 
> the other way around.  Currently it reads:
> 
> 	return a < b ? -1 : (a > b);

I think it is right. At least it should put recent things near the
front of the array, just as we are putting bigger things there.

> >   orig | 31.4M (100%) |   54s (100%)
> >     10 | 44.0M (140%) |  169s (310%)
> >     20 | 37.7M (120%) |  232s (428%)
> >     40 | 33.6M (107%) |  331s (608%)
> >     80 | 30.9M ( 99%) |  473s (868%)
> >    160 | 29.4M ( 94%) |  696s (1279%)
> [...]
> While keeping the size comparison commented out, you could try to 
> replace this line with:
> 
> 	return b < a ? -1 : (b > a);

No, it's not better. A few of the pack sizes are better, but some of
them are worse. And the CPU times are still quite bad. Here are the
numbers:

  orig | 31.4M (100%) |   54s (100%)
    10 | 45.6M (145%) |  158s (292%)
    20 | 39.2M (125%) |  205s (377%)
    40 | 35.1M (112%) |  275s (505%)
    80 | 32.4M (103%) |  388s (713%)
   160 | 30.6M ( 98%) |  581s (1067%)

-Peff