Major bootstrap time regression on March 30

[Joe Buck]

On Fri, Apr 08, 2005 at 07:36:17PM -0400, Daniel Berlin wrote:
> [ tracking down sudden increase in bootstrap time ]
> So just so we are clear on the state:
> 
> Jim works on a <insert details here> machine:
> 
> Sees no difference in compile times from 03-29 -> 04-01
> 
> Diego works on a <insert details here> machine
> 
> Sees 50% difference in compile times from 03-29->04-01, starting on
> 03-30

One possibility: required memory size has increased, and Jim has more
memory than Diego, but the new working set size is just enough larger than
Diego's memory size to make a few compiles fall off a cliff as far as
performance is concerned, but the time difference for Jim is in the noise.