slow gcc (was: RE: [Aranym-user] Speed up natfeat possible?)
olivier.landemarre at utbm.fr
Olivier.Landemarre at utbm.fr
Sun Feb 19 10:09:28 CET 2006
> I see now. Olivier it is important to tell the whole story before
> you state that 'gcc' is slow somehow. According to what I see you
> are actually trying to compare MagiC AES vs. XaAES (or MyAES)....
> I am glad that you found some of the reasons and I believe you
> can find further to improve the performance of your gccshell.
> I would however expect someone who has programming experience
> to first look into the issue before complaining. Do not take it
> as an insult please, this is just my mindset speaking.
No problem, it's just sometime difficult to find good reasons of problems and
because there is some software with wich there look trouble, near the same even
if it is not. For example if I write on it, it was because on Aranym Windows
there is (or was not test last fvdi version), CAB with 3 vectorial font take 5
minutes to launch, and it's so slow to draw (P4 2Ghz), while it take 15 seconds
on my Linux, now with new version it take something like 2 seconds, but I can't
explain why on Windows it's was so slow (perhaps Johan have an explanation). I
compare what I can compare, I use software sometime with bug, sometime not very
well write, sometime test only on this or this system, whole systeme, is in dev,
find good reasons is not easy. Notice someone said me that when it compile on
Hostfs it's very slow, while when it compile on ext2 it's fast and it use Make.
I have not verify it, and I will soon verify this. There is so many work.
1 month I try to do work correctly Papyrus on MyAeS, this take all my time, and
finaly I can't find the problem, no other software have this problems, and I
have not source code to find what they do, no help from developpers of this
> Feel free to report further aranym related problems. :)
> Best Regards
> PS: I try to respect one motto: "google before you ask someone
> not to waste somebody's time". And I often exchange the word
> 'google' with 'read the sources when available'.
> > -----Original Message-----
> > From: cz-bobek-lists-aranym-user-bounces at lists.bobek.cz
> > [mailto:cz-bobek-lists-aranym-user-bounces at lists.bobek.cz] On
> > Behalf Of olivier.landemarre at utbm.fr
> > Sent: Saturday, February 18, 2006 5:35 PM
> > To: aranym-user at lists.bobek.cz
> > Subject: Re: slow gcc (was: RE: [Aranym-user] Speed up
> > natfeat possible?)
> > Selon Standa Opichal <opichals at seznam.cz>:
> > > Hi!
> > >
> > > Good, now you can move this problem to MiNT list if someone will
> > > explain what is wrong. MiNT is generally slower than MagiC no matter
> > > what. You would probably need some numbers to compare. It would be
> > > great to get MagiC setup running in ARAnyM for this (it doesn't boot
> > > for me)...
> > I have take time to find problems, I have find a possible bug
> > in one of my
> > software, doing a Fread(readonly) on a file that redirect
> > stdout, if I add
> > optionnal "Deny none" I win 10%. This is still slow, so I
> > look again in my
> > source code (gccshell), and I look how I manage wait for gcc
> > stop and in same
> > time reading file. And I find I have 2 test, one using
> > evnt_multi(MU_MESSAG|MU_TIMER) with a timer of 1500ms, I wait
> > for CH_EXIT
> > message and I verify with Pwaitpid() all 1500ms if is still
> > here or not because
> > some AES not send CH_EXIT all the time, and I realise that
> > MyAeS not send
> > CH_EXIT message for non AES applications, where Magic do it,
> > so I should wait
> > all the time 1500ms before stop waiting, when there is a lot
> > of file time
> > consuming is important! I have for the moment reduce from
> > 1500 to 500ms and for
> > my test I win 33%!So first of all I have to send this problem
> > in the MyAeS list!
> > I will try to find a solution for this!
> > regards
> > olivier
> > _______________________________________________
> > cz-bobek-lists-aranym-user mailing list
> > aranym-user at lists.bobek.cz
> > https://lists.bobek.cz/mailman/listinfo/cz-bobek-lists-aranym-user
> This email has been scanned by the MessageLabs Email Security System.
> For more information please visit http://www.messagelabs.com/email
> cz-bobek-lists-aranym-user mailing list
> aranym-user at lists.bobek.cz
More information about the cz-bobek-lists-aranym-user