Re: More on Mustek 8000SP adventures...
Rogier Wolff (R.E.Wolff@BitWizard.nl)
Thu, 24 Jul 1997 17:47:42 +0200 (MET DST)
David Mosberger-Tang wrote:
>
> >>>>> On Thu, 24 Jul 1997 08:13:18 +0200 (MET DST), R.E.Wolff@BitWizard.nl (Rogier Wolff) said:
>
> Roger> I'm writing sane-support for the Tamarack SCSI scanner. I had
> Roger> the same problem: my NCR810 driver would report an "NCR DEAD"
> Roger> when I tried to scan a slightly non-trivial amount of data.
>
> Roger> Initially I increased the timeout to one minute. However
> Roger> nowadays, my driver simply reduces the scan-size to at most
> Roger> an inch at a time. The Tamarack takes less than 10 seconds
> Roger> to scan an inch, so there is no problem. This way I can run a
> Roger> standard kernel. (well almost).
>
> Roger> Mustec maintainers: This might also be applicable for the
> Roger> mustec backend....
>
> Hmmh, I don't think I would want to _force_ this behavior, since it
> would slow down scanning by quite a bit. However, maybe we could put
> an option into mustek.conf that lets you limit the maximum distance
> scanned in one chunk. Then we can distribute SANE with a mustek.conf
> that is safe for owners of an unpatched Linux ncr810 driver. This
> leaves the option of getting back the old behavior when it's known to
> be safe.
The Tamarack looses 2 seconds on each "read". At 32k/read and 25Mb per
scan, that comes to 2000 seconds. In this case the buffer size is the
limiting factor.
When scanning at a lower resolution (or less bit/pixel), (you
recommend 128k buffer, I use a megabyte at home... :-) you get into
the timeout problem.
By scanning at most 1 inch at a time, you are going to do at most 10
"reads" too many. On my Tamarack that gives you 20 seconds delay on a
total of 11 * 10 seconds.
Roger.
--
Source code, list archive, and docs: http://www.azstarnet.com/~axplinux/sane/
To unsubscribe: mail -s unsubscribe sane-devel-request@listserv.azstarnet.com