Re: Lus paa tjaerekost (or molasses in january)
Mark Hamstra (firstname.lastname@example.org)
Mon, 11 Nov 1996 11:21:41 -0500 (EST)
Linus Torvalds <torvalds@cs.Helsinki.FI> wrote:
> On Fri, 11 Nov 116, Johan Grape wrote:
> > System: Aspen durango 433MHz (PC164) 128Megs RAM
> > NCR Fast SCSI-2
> > Quantum XP.. 2gig
> > building a 2.0.18 kernel w/ basic networking and
> > SCSI driver and most features as modules.
> > time make vmlinux
> > 292.92user 28.65system 5:52.94elapsed
> That's a very reasonable time for an alpha kernel compile..
> > This is really disappointing. And the time does *not*
> > improve with second make on an idle system (most
> > files cached).
> Indeed. It should be around 90% CPU time to compile a kernel on any machine
> with reasonable amounts of memory.
> > AFAIK this compares to a 200MHz Cyrix or Pentium.
> > Is this typical, or do I have some problem with my setup?
> It's typical, and it's not a problem on you end.
> The reason it looks comparable to a 200MHz Pentium is mainly because you
> compare different things. The compiler may be similarly named, but compiling
> for x86 and compiling for alpha are pretty different things. Gcc is for some
> reason much slower on the alpha than on an x86. My personal guess is that
> there is either some instruction scheduling issue or some O(exp(n)) routine
> wrt number of registers or similar.
> And yes, I think it's sad too, as compiling is what I do most. The alpha
> does make it up in other areas, but it's still sad..
It's only slight consolation, but GCC isn't the only slow compiler on Alpha:
Microsoft Visual C++ 4.0 is pathetically slow on my Alpha when it's suffering
Bentley Systems, Inc.
To unsubscribe: send e-mail to email@example.com with
'unsubscribe' as the subject. Do not send it to firstname.lastname@example.org
Product Errata |
About Us |
Linux Info |
No Frames |
Copyright © 1995-1997 Red Hat Software. Legal notices