Mailing-List: contact cygwin-help AT cygwin DOT com; run by ezmlm
List-Subscribe: <mailto:cygwin-subscribe AT cygwin DOT com>
List-Archive: <http://sourceware.org/ml/cygwin/>
List-Post: <mailto:cygwin AT cygwin DOT com>
List-Help: <mailto:cygwin-help AT cygwin DOT com>, <http://sourceware.org/ml/#faqs>
Sender: cygwin-owner AT cygwin DOT com
Mail-Followup-To: cygwin AT cygwin DOT com
Delivered-To: mailing list cygwin AT cygwin DOT com
Message-ID: <4185CEBD.2040601@x-ray.at>
Date: Mon, 01 Nov 2004 06:50:53 +0100
From: Reini Urban <rurban AT x-ray DOT at>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; de-AT; rv:1.8a3) Gecko/20040817
MIME-Version: 1.0
To: Cygwin <cygwin AT cygwin DOT com>
Subject: Re: Need an older gcc version on my cygwin due to poor performance on gcc-3.x series.
References: <000701c4bf79$fe5d1e40$316d65da AT DANNY> <4185480F DOT 3080807 AT familiehaase DOT de> <000801c4bfbe$40351a90$3a4861cb AT DANNY>
In-Reply-To: <000801c4bfbe$40351a90$3a4861cb@DANNY>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
X-IsSubscribed: yes

Danny Smith schrieb:
> Gerrit P. Haase wrote:
>>Danny Smith wrote:
>>>Gerrit wrote:
>>>>Ole Jacob Hagen wrote:
>>>>>I've compiled Octave-2.1.60 with gcc-3.3.3 successfully, but the
>>>>>performance is pretty bad with gcc-3.3.3.
>>>>>The performance should of Octave is much better, when compiling it
>>>>>with gcc-3.2.x instead.
>>>
>>>>It would make more sense to identify the problem and do s.th.
>>>>about it to resolve this issue then.
>>>
>>> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14563
>>
>>Wow, this is an interesting story about exception handling.
>>
>>What I'm asking myself now:
>>
>>Should we try again to use dwarf2 exceptions?
> 
> My personal builds of gcc-3.4.x are bullt with dwarf2 EH enabled.  No
> problems But  I don't use w32 callbacks within functions that throw.  I
> had never even thought of doing that, but I don't think much.anymore

Max,
Do we use some win32 callbacks with exceptions in setup.exe?
I do see some candidates, but I didn't follow the codepath exactly.

On the other side, our setup.exe is not that performance critical as 
octave. But postgresql and apache are likely candidates for Dwarf2 
enabled builds. I experienced severe postgresql penalties with the 
latest builds, compared to earlier versions. current beta4 only allows 
MAX_CONNECTIONS=2, vs. ~50 with earlier gcc/postgresql. But postgresql 
internals also had changed a lot lately.

>>Should we try to find the reason why SjLj exceptions are slower on
>>Cygwin than for the rest of the world?
> 
> Most of the rest of the GCC world has abandoned sjlj EH as inefficient
> Look  within function prologues.
-- 
Reini Urban
http://xarch.tu-graz.ac.at/home/rurban/

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Problem reports:       http://cygwin.com/problems.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/