[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [MiNT] XAAES slow?
On Thu, 04 Aug 2005 14:37:34 , Odd Skancke <ozk@atari.org> wrote:
>
> tor, 04,.08.2005 kl. 12.16 +0200, skrev Peter Slegg:
> > On Thu, 4 Aug 2005 13:43:33 , Jo Even Skarstein <joska@online.no> wrote:
> > > > From: Odd Skancke [ozk@atari.org]
> > > > Sent: 2005-08-04 13:35:57 CEST
> > > > To: MiNT List [mint@fishpool.com]
> > > > Subject: Re: [MiNT] SV: XAAES slow?
> > > >
> > > > Hmm... Do you have such problems with pre 1.16.x kernels too? Or is
> > > > this a combo coming from 1.16, Milan ATI gfx driver and my drivers?
> > >
> > > My problems started sometime last fall. I have not tried earlier kernels
> > > after that, so I don't know if this is the cause of my problems. Also,
> > > sometimes everything works fine, which doesn't make troubleshooting
> > > any easier... I'm still suspecting hardware problems with my Milan.
> > >
> > > > Can either of you just for the funfactor of it, if you can, check how
> > > > the same setup works with a different gfx card?
> > >
> > > I'll try to find the old S3 card and try that.
> > >
> > > > Jo Even, you say the NE2K driver behaves like this too?
> > >
> > > Yes, it does.
> > >
> > > Jo Even
> >
> > I don't have my S3 gxf card to hand at the moment so I can't test that.
> > Jo Even, you say it works intermittently. What I found was that small
> > e-mails (< ~3500 bytes) would send properly, anything larger would stall
> > and not get sent.
> >
> > I did a lot of tests, sending myself e-mails of different sizes while
> > using 1.16.x
> >
> > With the earlier 1.15.12 all works perfectly so if you can test
> > that it may show that the problem is related to the kernel.
> >
> > Evan got me to run tcpdump and send him the output for him to
> > interpret. He posted some info about what he found.
> >
> > I am going to do some checks with wget to see if that has the same
> > problem.
>
> This leads me to think that there may be a problem with the CPU cache
> on the 060... I cannot think of other reasons why the drivers work on
> 1.15 and not 1.16.. I will investigate a bit..
>
I am not surewhy you suspect your driver if it works with 1.15, I would
have expected the kernel to be the cause.
I am not sure how to deactivate the data cache. Can it be done without
a reboot or do I need to edit the mint.cnf ?
Did I read that there has been a change in the cache sizes or block sizes
or something like that ?
Regards,
Peter