Re: pgcc optimizationss

Petko Manolov (petkan@varel.bg)
Fri, 18 Dec 1998 11:19:14 +0200

--------------7F34EF92AFEB469DF177984B
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Marc Lehmann wrote:

> Yeah, sure, but mentioning "it works fine with -O6" makes less-skilled
> people use it, with sometimes bad results ;)
>

You are right of cource. But everybody has a head on his shoulders to think with
;-)
I only share information that hopefully can be used for bug fixes. This list is
not for
amateurs ;-))

> I hope Linus applies bug-fixes, yes ;)

AFAIK Linus thinks egcs guys silently changed some concepts in inline asm
(clobber list) and now they are blaming the kernel. I can't say who is right but
it's obvious
gcc and pgcc produced different inline asm code. I'm pretty sure they shouldn't
;-)

> Optimizations often interact with each other in subtle ways. Most(!)
> of the bugs I find are caused by bugs in egcs that only come
> apperent in some later optimization.

I'm afraid i didn't understand this.

>
> But egcs is getting better with each second, it even does hard register
> flow analysis now ;)
>
> Lets hope egcs makes a better gcc sooner than later!

Right, I like pgcc and I wish to see it stable.

--
Petko Manolov - petkan@varel.bg
http://www.varel.bg/~petkan/






--------------7F34EF92AFEB469DF177984B
Content-Type: text/html; charset=us-ascii
Content-Transfer-Encoding: 7bit


<!doctype html public "-//w3c//dtd html 4.0 transitional//en">

Marc Lehmann wrote:

Yeah, sure, but mentioning "it works fine with -O6"
makes less-skilled

people use it, with sometimes bad results ;)

 

You are right of cource. But everybody has a head on his shoulders to think
with ;-)

I only share information that hopefully can be used for bug fixes.
This list is not for

amateurs ;-))

I hope Linus applies bug-fixes, yes ;)

AFAIK Linus thinks  egcs guys silently changed some concepts in inline
asm

(clobber list) and now they are blaming the kernel. I can't say who
is right but it's obvious

gcc and pgcc produced different inline asm code. I'm pretty sure they
shouldn't ;-)

Optimizations often interact with each other in subtle
ways. Most(!)

of the bugs I find are caused by bugs in egcs that only come

apperent in some later optimization.

I'm afraid i didn't understand this.

 

But egcs is getting better with each second, it even does hard register

flow analysis now ;)

Lets hope egcs makes a better gcc sooner than later!

Right, I like pgcc and I wish to see it stable.

-- 
Petko Manolov - petkan@varel.bg
http://www.varel.bg/~petkan/

 

--------------7F34EF92AFEB469DF177984B--


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/