From: gdemont AT my-deja DOT com
Newsgroups: comp.os.msdos.djgpp
Subject: Re: Performance 2.7.2 -> 2.8.1
Date: Thu, 19 Oct 2000 14:36:23 GMT
Organization: Deja.com - Before you buy.
Lines: 35
Message-ID: <8sn0t7$1nd$1@nnrp1.deja.com>
References: <8smgof$l4f$1 AT nnrp1 DOT deja DOT com> <971961901 DOT 585069 AT shelley DOT paradise DOT net DOT nz>
NNTP-Posting-Host: 130.125.90.100
X-Article-Creation-Date: Thu Oct 19 14:36:23 2000 GMT
X-Http-User-Agent: Mozilla/4.74 (Macintosh; U; PPC)
X-Http-Proxy: 1.0 sitelnet.unine.ch:8080 (Squid/2.1.PATCH2), 1.0 x60.deja.com:80 (Squid/1.1.22) for client 130.125.13.29, 130.125.90.100
X-MyDeja-Info: XMYDJUIDgdemont
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp
Reply-To: djgpp AT delorie DOT com


> It is unclear whether you mean the compiler is running slowly, or the
> compiled program is running slowly.

OK, I meant the generated code only - I don't care about how slow
the compiler goes. I assume that the more time it takes, the better
it tries to optimise its output.

> Where exactly have you noticed performance degradation?  (I do remember
> reading that the optimisations enabled by -O, etc. have changed between
> versions, but this should only give the compiler more work to do.)

Some of the programs I have under hand do run slower.
Target is a Pentium-S 166Mhz.
The options are
* For gcc 2.7.2 (gnat 3.10):
 -i -gnatpn -O2 -fomit-frame-pointer -funroll-loops
 -m486 -malign-loops=2 -malign-jumps=2 -malign-functions=2

* For gcc 2.8.1 (gnat 3.11-3.13):
 -i -gnatpn -O2 -fomit-frame-pointer -funroll-loops
 -mpentium (with or without the -malign... options)

The code produced by the latest GNAT, 3.13, goes faster
due to a serie of optimisations specific to Ada (using
the information given by strong typing), but does
not compensate - in my tests - the degradation brought
by 2.8.1. Hence my question. Maybe I don't switch on the
right optimisations, too.

Gautier


Sent via Deja.com http://www.deja.com/
Before you buy.