Re: Why does everyone hate gcc 2.95?

From: Michael Meding (Michael@Meding.net)
Date: Wed Oct 04 2000 - 04:21:57 EST


Hi there,

> I hate it because it compiles much more slowly than 2.72 and for
> my purposes,
Hm, have not checked that. Did you do benchmarks here ?

at least, the resulting code is not any faster on
> any of the following platforms: x86, SPARC, MIPS, PA-RISC, and Alpha.
Hm,

quick check of dgemm or also testgart or gears gives me a 10 to 20
percent increase in speed when compiling with -mcpu=pentium -O2 as
switches for both runs on my duron.

All this reasons are not showing that gcc-2.95.2 is somehow broken.
Please enlighten me.

With best regards

Michael

Larry McVoy schrieb:
>
> On Wed, Oct 04, 2000 at 04:28:41AM +0000, John Anthony Kazos Jr. wrote:
> > What does everyone have against gcc 2.95 on this list? I've been
> > compiling kernels successfully (read: not one single (ever) error
> > in compilation) with gcc 2.95.2 for more than a year now. What's the
> > big deal?
>
> [Fix your mail program to put in carriage returns at 72 columns, please]
>
> I hate it because it compiles much more slowly than 2.72 and for
> my purposes, at least, the resulting code is not any faster on
> any of the following platforms: x86, SPARC, MIPS, PA-RISC, and Alpha.
> I ran a bunch of tests with both on the BitKeeper source base, about
> 100K lines of code or so, and then ran the regressions as well as some
> hand picked tests. No faster. Just compiles slower. Add to that
> some distributions BRAINDEAD default of havving colorgcc be the default
> compiler (can you say fork perl to fork gcc? Can you say STUPID?), and
> you start to understand why the first thing I do is remove all that
> garbage and put back a reasonable compiler.
>
> I'm not impressed.
> --
> ---
> Larry McVoy lm at bitmover.com http://www.bitmover.com/lm
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> Please read the FAQ at http://www.tux.org/lkml/
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Sat Oct 07 2000 - 21:00:13 EST