I have a rather techinical question regarding CDRW IDE-data transfer.
I would be pleased if someone could help me out.

I wanted to see to what extent seektime influences the reading performance
of a CDRW-recorder. I though I would test this with a real world data
transfer test in Windows 2000 using Windows Explorer. I decided to test
two CDRW-recorders with very different seektime like Plextor PX-W2410A with an average
of about 135 ms and AOpen CRW2010 with less than 100. The transfer would
consist of two different sets of data.

Set 1
1 file 695 MB

Set 2
7815 files in 98 directories

I thought the result would be that the recorders would be approximately equally
fast on the first set and that AOpen CRW2010 would do the second set much faster
than the Plextor drive. It was, however, not that easy.

The recorders were connected to the same IDE-channel as the harddrive
(not at the same time though). The results were:

Set 1
Plextor PX-W2410A 155s (30,6x)
AOpen CRW2040 157s (30,2x)

The results were not all that surprising. The results didn't differ much
if I connected the recorder to the secondary IDE-channel.

Set 2
Plextor PX-W2410A (268s)/(10,3x) - [273s/10,1x]
AOpen CRW2040 (281s)/(9,8x) - [328s/8,4x]

With the second set I encountered strange results. The first results (in ordinary
parenthesis) is the result when the drives were connected to the same
channel as the harddrive and the second results are when the recorder is on
the secondary channel. When the drives are on the same channel the Plextor is still
faster than the drive from AOpen indicating that it's more important to be a good
reader in general than to have a look seektime. When the drives are on the secondary
channel the drive from Plextor transfers in about the same time but CRW2010 is
50 seconds slower! I've repeated this test several times to make sure this is not
a random occurance. I haven't been able to test with other recorders though.

I've done this tests with AOpen CRW2010 in both PIO4 and UDMA33 - it did not
make any difference whatsoever on the performance of the drive. Bandwidth
is clearly not a limitation, and it really shouldn't be.

My questions are:

* Is seektime less important than the drive being a good reader in general?

* What property of the drive causes the recorder from AOpen to loose a lot
of time when using the secondary channel. It sure isn't the seektime.

Is there any good way to determine the impact of seektime on real world performance.
This test indicates that there are far more important things to worry about.