Gleydson Mazioli da Silva <[EMAIL PROTECTED]> reported:
> I would like to report the success writing to 6 Matshita CW7502 on a
> Linux system (firmware updated to 1.17) at the same time. I've created
> a shell script to do that and allocated 2MB of buffer to each cdrecord
> process and all the recording works fine.
If you are burning the same thing on all devices, I would expect the
input data to remain in memory buffers, and no problem.
> When I tried to add a Matshita CW7503 to the set I got "errors flushing
> the cache"
> (Now I'm sure that sg of Linux is very poor, about it's max transfer
> rate).
This may be related to the buffer size you chose for the processes. If
you can make this happen in -dummy mode you might try smaller buffers
and see if there is any change.
> The next thing that I will try do is got some lent recorders to do tests
> on the FreeBSD and try to reach better performance results (and probably
> with more
> recorders at the same time) :-)
Another thing which may be of interest is doing this without the
realtime priority. I assume you are not doing this on a loaded machine,
so it probably doesn't make much diference, and there may be some
oddities in behaviour.
> - The kernel 2.4 had a better sg driver, the transfer rate increased to
> 4.700/4.800
> Kb/s but isn't that I was hopping... the SCSI card was a Adaptec AIC
> 19160
> and the sg transfer rate was a lot down of the allowed by the card.
I sure hope you mean 4.8 MB and not Kb, I get ~1.5MB on a single 12x
recorder using IDE!
Most impressive, but probably not cost effective compared to the cheap
multi-copy burners being sold.
--
-bill davidsen ([EMAIL PROTECTED])
"The secret to procrastination is to put things off until the
last possible moment - but no longer" -me
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
