OT: Firewire External HD's vs. USB-intersting

FrigidNDEditing wrote on 10/21/2005, 11:19 AM
Just got an external enclosure that handles Firewire or USB and I decided to use the firewire connection instead of the usb since I wasn't cappin to it. Now I find that the Render times to the HD are faster - One minute MPEG clips that someone wasn't able to give me in .avi I can now render in like 50 seconds rather than the 1:20-1:30 over the USB - not using anything else on the USB bus. Very interesting to me. I, for one, Did not know that there would be that large of a difference in the render times - I know that USB uses CPU cycles and that Firewire doesn't need to so that may be why, but I am simply letting you guys know that even though people sometimes say that USB is just as good - it doesn't seem that that's the case

(Firewire 400 - USB 2.0) (Same enclosure used to test both connections)

Dave

Comments

John_Cline wrote on 10/21/2005, 12:08 PM
Yes, USB transfers utilize a fair amout of CPU horsepower. Firewire transfers, on the other hand, do not involve the CPU nearly as much to manage the data stream.

All things being equal, 400mb/s Firewire will always beat 480mb/s USB 2.0.

John
Spot|DSE wrote on 10/21/2005, 12:38 PM
I would tend to agree, having always experienced this myself, but there are a *lot* of posts from USB users claiming USB is running faster for them than Firewire. Personally, I've never, ever experienced this and don't trust USB2 worth a damn, but...there are too many posts in the other direction for it to be totally dismissed.
rmack350 wrote on 10/21/2005, 3:45 PM
I think the key is rendering.

Douglas, when we were first doing the render tests I noticed that you had slightly faster renders on a scsi system, presumably because of SCSI's low overhead.

It may be just as fast to capture to a USB disc but rendering while using one seems to be another issue.

Every cpu cycle counts when rendering.

On the other hand, drive speed doesn't matter as much. When using Vegas 3 on my 500MHz laptop I found that it was just as fast to render to a Zip disc on a parallel port as it was to render to a 1394 hard drive. (I'm not sure what you'd call the state of mind I was in to consider such a test, though)

Rob Mack
farss wrote on 10/21/2005, 3:54 PM
This makes perfect sense as does the claims of better USB 2 performance. Quite possibly doing just a plain file trasnfer USB 2 might pip 1394 if you did it on a fast PC, thing is, as said, if you need to do anything to the data on the way through and there's no CPU cycles left then you're screwed.
You could probably conduct some wierd science experiment that showed that rendering to/from a floppy disc was no slower than using a SCSi RAID array. All you need to do is make the process so CPU intensive that the data i/o rate becomes irrelevant to make this happen.
That's why there's no 'best' system for running Vegas, it really depends on what you do and where the bottlenecks occur in your typical process. You get more bang for your buck spending the dollars freeing up those bottlenecks in YOUR process.
Bob.