Re: Low burning speed

["Thomas Schmitt" <scdbackup@gmx.net>]

Hi,

Tomasz KÅ~Baptocz wrote:
> i finally tried out cdrskin - it behaves just
> like growiso, first zone ok, second zone slowly.
> Growis with option "speed=6" works properly,
> "speed=8" like without speed option.

This strengthens the suspicion that there is
a timing problem in the drive between incomming
data and the rotation frequency of the media.


> But what's interesting growiso+K3b and speed set
> at 6x records at 3.5 - 4x, when 8x is checked -
> first zone 3.5 - 4x, second 2.5 - 3x.

Strange (once again). The problem seems to react
very sensitively on the environment of a burn run.

I assume k3b still has the option "Show Debugging Output"
(or a similar one) which after a burn run reports the
external commands and their message output.
It would be interesting to see the effective growisofs
command which was used and led to this extra low 
performance.


> tomasz@debian:~$ growisofs -Z /dev/dvd=/dev/zero
> ...
>  212172800/4700372992 ( 4.5%) @6.0x, remaining 10:13 RBU 100.0% UBU  88.5%
>  232062976/4700372992 ( 4.9%) @4.3x, remaining 10:16 RBU 100.0% UBU  96.2%
>  254279680/4700372992 ( 5.4%) @4.8x, remaining 10:29 RBU 100.0% UBU  46.2%
>  286949376/4700372992 ( 6.1%) @7.1x, remaining 9:59 RBU 100.0% UBU  19.2%
> ...
>  4683038720/4700372992 (99.6%) @7.1x, remaining 0:01 RBU 100.0% UBU  19.2%

This higher speed indicates that the timing problem
sits on the computer side of the suspect couple
(computer versus DVD drive).
In general it seems to run better in simple situations
and gets worse with any additional activity which runs
together with the burn.

My currrent theory is that the drive does not accept new
data from the computer while it is waiting for the media
to rotate to the next write position.
Possibly the reason for this waiting is that the drive
is too optimistic about the timing and expects the
computer to send the next data more quickly after the drive
gets ready for receiving again.

New experiments:

- Retry with cdrecord.
  Check whether it really is burning at 8x speed
  by measuring the overall burning time.
  Please post some of the cdrecord message output.

- If cdrecord reliably works with an overall throughput
  of (nearly) 8x, then we need to find out the difference.
  If you did not try growisofs running as superuser
  yet, then do it now.
  Maybe the superuser gets a better performance out of the
  sg driver of your Linux system.


Have a nice day :)

Thomas