gcc 3.1 is still very slow, compared to 2.95.3

[Robert Dewar]

> If the build machine has "reasonable" amounts of memory, setting
> this too low wastes time collecting too often.  If the build machine
> has "small" amounts of memory, setting this too high wastes time
> swapping.

yes, that's clear, but I wondered if anyone had actual quantitative 
measurements. Note that setting it too high might also increase TLB
misses, and that's what's interesting. What should be the setting on
a machine with large (1 or more gigs of physical mem).