Sender: jur AT rz DOT hu-berlin DOT de
Message-ID: <3741BCDD.E2776759@informatik.hu-berlin.de>
Date: Tue, 18 May 1999 19:17:49 +0000
From: Jens-Uwe Rumstich <rumstich AT informatik DOT hu-berlin DOT de>
Organization: TUSCON
X-Mailer: Mozilla 4.05 [en] (X11; I; Linux 2.2.8 i586)
MIME-Version: 1.0
To: pgcc AT delorie DOT com
Subject: Re: Benchmark PGCC vs EGCS on a K6-2
References: <373F3AA2 DOT A446D611 AT informatik DOT hu-berlin DOT de> <19990517003851 DOT L10291 AT cerebro DOT laendle>
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: 8bit
Reply-To: pgcc AT delorie DOT com

Hi Marc!

> > with K6-Option, but there is an improvement with enabled K6-optimizing.
> > - both compilers create worser code with O4 than with O3
> 
> Your results are bogus. egcs does NOT make any difference between -O3 and
> -O4.  If you tests show a difference then its not compiler related.

Sh*t, it seems you are right :-(
I tried really hard to get reliable numbers, but it seems I failed :-(
I hope IŽll do better next time...

> > PS: aehm, are there any reasons, why the def_align in i386.c for the K6
> > is set to 0 instead of 5 (32 bytes cache alignment) or atleast 4 ??
> 
> pgcc uses an adaptive alignment, which is available with newer versions of
> binutils (2.9.x)

What does that mean? Do these new binutil versions care about the cache
alignment?

>       ---==---(_)__  __ ____  __       Marc Lehmann      +--

cu	
	Jens-Uwe