USB 2.0 vs. FireWire vs. UDMA Mode 2

[thiscloud]

I am wondering what the performance difference between the external DRX-500UL and the internal DRU-500A would be.

Looking at the specs for the various technologies, on paper it looks like USB 2.0 would be fastest at 60MB/sec (480 megabits per second / 8 bits per byte = 60 megabytes), followed by FireWire at 50MB/sec (400mbps / 8 bits), and then lastly UDMA Mode 2 at 33MB/sec. Of course this is all theoretical upper limits.

However, it's also a commonly held opinion that internal drives are faster than external. So which would be faster (I know there's differences between sustained vs. burst, and all of that, but in general...)?

[Pradeep]

The external drive is simply the internal drive in an external enclosure. There should be no difference in either write or read speeds as they are both well below maxing out either interface. I have the internal in an ADS Firewire enclosure, works well.

[thiscloud]

Pradeep: Thanks for your response. I understand that it's the same drive and that the read/write speeds would be identical for the drive itself, but the interface to the computer is different and I'm trying to figure out what kind of performance difference the interface makes.

[Pradeep]

In theory USB 2.0 would be quicker, but in real life it turns out that Firewire offers better performance. This is probably due to the maturity of the IDE->Firewire bridge chipsets. Apparently USB 2.0 bridges have problems with large block transfers as would be encountered during sustained read/write ops.

[thiscloud]

Will FireWire have better performance compared to an an internal IDE drive or would it be about the same?

[Pradeep]

Until/If there are native Firewire devices, an internal IDE drive will always have the better performance, as there is always overhead during the Firewire->IDE conversion and back again. Of course this difference may be so small as to be un-noticeable. If you are choosing between external and internal, it depends on if you have more than one computer. If so then an external drive or casing is always handy to have.

[Kennyshin]

From the CDRlabs news page

USB 2.0 vs. Firewire

http://thetechzone.com/display.php?i=227&p=1

[thiscloud]

As is sometimes the case, the discussion forums provide more information than the article itself:

http://ttzforums.com/showthread.php?s=&threadid=26608

The post from Jeff Ross (looks like he copied and pasted it from somewhere) is nicely detailed, and the links to the TechTV.com article (among others) seem to support what I've heard, that although USB 2.0's theoretically faster than FireWire, FireWire seems more stable in streaming data and therefore makes more sense for use with optical drives.

[bac]

I think there are a couple dirty secrets about IDE/Firewire bridges. (I
suspect the same is true of the IDE/USB2 bridges but I have no experience).

First, configurability. People have problems with firewire enclosures due to
things such as which UDMA mode is negotiated, and how many seconds the
controller waits before probing the device. Oxford's firmware utility lets
one configure such parameters, but the utility has not been made widely
available. I see no enclosure retailer who supports configuring the firewire
bridge through such a utility, even though one may very well need to in order
to get their device to work. These enclosures are marketed as if they
are plug and play, but they aren't necessarily.

Second, error detection and correction. When I first put my dru500a in
a firewire enclosure, I got a bit error every 2MB or so during DVD writes.
My writes would be corrupt without any error being reported. Tracked
the problem down to the IDE cable being too cheap, producing cross-talk.
An internal IDE controller would detect such errors.
Later, I attached an IDE disk into a firewire enclosure. The disk contained
data errors. With the firewire enclosure, the OS read past the errors
with no error reporting (data was merely corrupted on read). Plugged
in internally, the IDE controller reported CRC read errors for this same
disk...

Does USB2 have these same problems? If not, it is clearly the winner.

[Pradeep]

I think there are a couple dirty secrets about IDE/Firewire bridges. (I
suspect the same is true of the IDE/USB2 bridges but I have no experience).

First, configurability. People have problems with firewire enclosures due to
things such as which UDMA mode is negotiated, and how many seconds the
controller waits before probing the device. Oxford's firmware utility lets
one configure such parameters, but the utility has not been made widely
available. I see no enclosure retailer who supports configuring the firewire
bridge through such a utility, even though one may very well need to in order
to get their device to work. These enclosures are marketed as if they
are plug and play, but they aren't necessarily.

Second, error detection and correction. When I first put my dru500a in
a firewire enclosure, I got a bit error every 2MB or so during DVD writes.
My writes would be corrupt without any error being reported. Tracked
the problem down to the IDE cable being too cheap, producing cross-talk.
An internal IDE controller would detect such errors.
Later, I attached an IDE disk into a firewire enclosure. The disk contained
data errors. With the firewire enclosure, the OS read past the errors
with no error reporting (data was merely corrupted on read). Plugged
in internally, the IDE controller reported CRC read errors for this same
disk...

Does USB2 have these same problems? If not, it is clearly the winner.

My ADS Pyro firewire kit came with a utility to change the mode from optical to hard drive mode. Personally it never did anything for me and I left it on auto-detect.

[bac]

My ADS Pyro firewire kit came with a utility to change the mode from optical to hard drive mode. Personally it never did anything for me and I left it on auto-detect.

That's a start. Can you configure UDMA mode or startup
delay or anything else?