This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: gcc compile-time performance
- From: dewar at gnat dot com (Robert Dewar)
- To: davem at redhat dot com, dewar at gnat dot com
- Cc: ak at suse dot de, dberlin at dberlin dot org, dhazeghi at pacbell dot net,gcc at gcc dot gnu dot org, neil at daikokuya dot demon dot co dot uk, shebs at apple dot com
- Date: Fri, 17 May 2002 21:19:09 -0400 (EDT)
- Subject: Re: gcc compile-time performance
> No, I don't mean a bootstrap of GCC. I mean a bootstrap of something
> like a proprietary CAD tool.
What does it mean to "bootstrap" a proprietary CAD tool?
Surely you mean build and not bootstrap. It is true that my example of
GNAT bootstrapping is a fairly small case (between half a million and
a million lines of actual source - not counting the equivalent of headers
read) compiled in 10 minutes on a slow machine.
I agree that this is not spectacular in speed, but in practice if a complete
system build takes a couple of hours (which is typical of compiling a few
million lines on a very fast machine) that's not terrible.
I certainly understand the desire for very fast compilation. The whole of
the Realia compiler was designed with this in mind, it compiled at 10,000
lines/minute on a 4.77 MHz PC 1, and a bootstrap took less than a minute
on a 25Mhz 386.
When I first came to using GCC, I was dismayed by the speed (but then I
was using a 25MHz 486), Now that I use a 1.35MHz PIII, I find the speed
acceptable, and I am more concerned with performance of the generated code.