[about the ScanMagic 9636 S]
> It's the same scanner, just a different name and slightly modified case. I
> think the ScanMagic is noname model from Mustek, at least there is no
> "Mustek" anywhere on the scanner.
Right, the only way I knew this scanner was a Mustek was because I found
out the ScanMagic scanner has support in the Mustek backend, and later because
the sane frontends said it was a Mustek; there is not a single word "Mustek" to
be found anywhere on the box or scanner.
> Just to compare, that's the output from my Mustek ScanExpress 12000SP V 2.0.2:
*nods* I see, exactly the same.
> > I've managed to get the SCSI ISA card working with the g_NCR5380 driver
> > and ncr_addr=0x280 dtc_3181e=1 options. Scanning was very slow, so I
> > changed the USLEEP_SLEEP value to a minimum (1). I also changed
> > USLEEP_WAITLONG define to 1. Now the speed is bearable, but with color
> > scans, backtracking occurs.
> Independent from resolution settings and scan width?
Well, now that you mention it, at low resolutions and small widths,
backtracking does not seem to occur. But when I increase the scan width with
the same resolution, it will happen. If I up the resolution with the same small
width, backtracking also occurs; and scanning is very slow, but that could be
normal, i.e. even so with the Windows driver (haven't tried).
> > /usr/include/scsi/sg.h has been set to 130560 (prior to compiling SANE)
> This is not necessary for newer kernels.
Right, but it can't hurt either, can it? :) Also, I have some SCSI host adapter
emulation for CD writing which *may* profit from it, but I don't think so.
> > and as you can see in the debug output, buffersize is increased to 1024
> > kb by the mustek backend.
> I don't think this will help with ScanExpress/ScanMagic scanners. They like
> their standard 64 K blocksize.
But it won't have a negative effect either?
> Just some data for comparison (everything in color mode):
...
> I have done quite a lot of work to improve performance for the Mustek
> scanners. I hadn't much success concerning the ScanExpress Scanners in
> connection with the Mustek card, however. Meanwhile I think the problem
> isn't located in the Mustek backend but in the SCSI kernel driver. I have no
> experience with kernel programming so checking the driver has very low
> priority for me. Maybe someone else with more experience can look into the
> different "performance" of this SCSI adapter under Windows and Linux.
I can believe it's the SCSI driver; it would be good fun to see performance of
the Mustek SCSI ISA card as good as with the Windows driver, though. Somebody
with kernel programming experience should mingle with the issue. :)
> With an Adaptec 2940 or a NCR 810 clone the scanner is quite usable but not
> perfect.
Well, with this dratted backtracking, it's still usable of course, but I know
it can do better and that hurts. I might buy me another SCSI card someday to
solve the problem if the SCSI driver doesn't get better then.
Ivo
-- Source code, list archive, and docs: http://www.mostang.com/sane/ To unsubscribe: echo unsubscribe sane-devel | mail majordomo@mostang.com
This archive was generated by hypermail 2b29 : Thu Sep 21 2000 - 02:41:52 PDT