Re: [Dvdrtools-users] Problem burning to LG DVDRAM GSA-4167B

[Bryan J. Smith]

Samson Lee <address@hidden> wrote:
> Hmmmm.. Just to confirm, are you also using cdrecord-prodvd
> 2.01b31 with LG GSA4167? Or are you using one of the
> 'unofficial' dvd patches?

Unofficial because it works flawlessly with the GSA-408x and
GSA-416x series.

> The drive is on secondary master. The hard drive with the
> .iso is on secondary slave. All devices are running udma2.

Bam!  That's your problem.  You will run into a noticable
performance issue when you are doing DMA from one device to
another on the same channel.  Especially from the
all-important latency aspect.

And that's assuming the ATA v. ATAPI devices and their
different vendor/model quirks aren't already in-fighting and
reseting the ATA bus (which is detrimental to performance).
Trust me on this, put each device on their own channel, and
you will avoid performance issues.  You should _only_ use the
master/slave relationship for PIO (which is slow and CPU
intensive), and _never_ for DMA -- or in the worst case,
never for transfers occurring from both devices on the same
channel simultaneously.

ATA is not SCSI.  It does _not_ do device-to-device
transfers.  One ATA device must take control of the bus,
setup the registers for transfer, transfer to memory and give
up the bus -- while the other device then does the same.  The
acquisition of the bus is the main issue -- especially when
data is trying to be fed to the recorder in real-time, while
the HD is trying to transfer into memory.

> # hdparm -i /dev/hdc
> /dev/hdc:
>   Model=HL-DT-ST DVDRAM GSA-4167B, FwRev=DL10,
>     ...  
>   UDMA modes: udma0 udma1 *udma2
>   AdvancedPM=no
>   Drive conforms to: device does not report version:
>   * signifies the current active mode

And "hdparm -i /dev/hdd"?



-- 
Bryan J. Smith     Professional, Technical Annoyance                      
address@hidden      http://thebs413.blogspot.com
----------------------------------------------------
*** Speed doesn't kill, difference in speed does ***