[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: [general] Netwinder crashes during large data transfers

In message <[🔎] 20030424115901.A1745@kei.netwinder.org>
          Ralph Siemsen <ralphs@netwinder.org> wrote:

> Hmm, based on what you are describing I would tend to blame DMA between
> the drives and the tulip.  There should be an oops, hopefully, that you
> could capture via serial console...

Good idea...

> Note that having >1 drive is uncommon, therefore not well tested.  I
> would be concerned about heat generated, since it is directly above the
> processor, footbridge, and the '553 which contains the IDE controller.

Well, I modified the setup a little bit. There are two 3,5" drives 
connected to my 'winder, but obviously not inside the OS case. They live in 
an open rack outside to avoid extra cooling (and noise...).

I have to admit that I also switched off the case fan, but had never problems
with this. The machine is rock solid and stable beside the stupid copy problem
with big files...

> Can you produce a failure when using the 10-base-t interaface (not the
> tulip)?

Never tried this. But it wouldn't help much, because I'm only interested
in 100BaseT.   ;-)

> I regularly move ISO images to and from my 'winder, albeit I am running
> the DM image and a 2.4.19 kernel.  

Are there still performance issues with 2.4.x kernels on ARM? Some times
ago I read a comparison which showed that 2.4.x kernels are much slower than

> I do have a debian stable install so
> I can try reproducing this.
> Your hdparm settings match mine (except for the spindown).
> -Ralph

<snip snip>


Stefan Wuerthner                             web  http://wuerthner.dyndns.org

Reply to: