[Pan-users] Re: Re: Re: 0.92 amd64

[Duncan]

Duncan posted <address@hidden>, excerpted below, 
on Fri, 14 Apr 2006 07:52:14 -0700:

> I'll bump the ulimit in tenths of a gig at a time, and see what happens,
> posting back when I get some numbers.

OK, gcc-3.4.6 will compile pan-0.92 from portage on amd64, with a 1.3 gig
ulimit -v, but not with a 1.2 gig ulimit -v.

I set top up in another konsole tab, set to update @ 0.2 second intervals,
and saw one compile thread climb to 1330 MB, so it's close to 1.3 gig.  (I
was using 1363149 KB as my 1.3 gig ulimit.  That's 1331 MB, so it's real
close to 1.3 GB.)

By contrast, with gcc 4.1.0, a virtual memory ulimit of 0.3 gig compiles
pan-0.92 just fine.  (0.2 gig fails, tho.)  That's why I didn't notice
anything unusual here and therefore had you (OP) looking all sorts of
/other/ places.  Obviously, the refactorization/rewrite that took place
with gcc-4 and is now beginning to hit stride with 4.1 /really/ makes a
difference, in this case.  3.4.6 requires 4-5X the memory of 4.1.0, to
compile pan 0.92, or at least that's the case for the Gentoo versions.

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman in
http://www.linuxdevcenter.com/pub/a/linux/2004/12/22/rms_interview.html