Monday, May 23, 2011

USB 2.0 vs. Firewire 800

I recently bought a couple of Seagate FreeAgent GoFlex Desk 2 TB drives. These drives come with a USB 2.0 dock - or as they call it, an "Adapter" - which is the base the drive sits on, and provides power and I/O connections.

But you can also buy a Firewire 800 Adapter for $24 which lets you connect the drive through Firewire or USB. Since I have a MacBook Pro (with a Firewire 800 port) I thought it might be worth checking out to see if it improved the performance of the drive.

Switching a drive from one adapter to another is very easy; it simply unplugs from the bottom of the drive. Just make sure you dismount and power down the drive before you switch adapters.

As an experiment, I rendered a Final Cut Pro project on the drive while connected through Firewire 800, and then switched it to USB and repeated the test. I was a little surprised by the results:
Firewire 800 : 7:03
USB 2.0: 7:05
With a seven minute render, only 2 seconds were saved; and that 2 seconds are probably illusory because I was using my iPhone to time the process.

If you want to be able to connect a drive through the Firewire port, then get this adapter, but if you think it's going to improve performance you'll be disappointed.

Oh, one final thing, if you're shopping for these adapters make sure you get the right one; don't confuse the Seagate FreeAgent GoFlex Desk Desktop Adapter - Firewire 800/USB 2.0 STAE105, which works with the desktop drives, with the Seagate FreeAgent GoFlex Upgrade Cable FireWire 800 which works with portable GoFlex drives.

Amazon: Seagate FreeAgent GoFlex Desk 2 TB USB 2.0 External Hard Drive $109.99
Amazon: Seagate FreeAgent GoFlex Desk Desktop Adapter - Firewire 800/USB 2.0 $23.49

Update 7:00 PM

A couple of readers have suggested that the render test wasn't a very good way to test the drive. While I agree it's not a good way to test the flat-out performance of the drive, I was trying to see if it would impact my day-to-day use of Final Cut Pro, rather than just seeing if one drive was faster than another.

However, to be fair, it's probably a good idea to see if there's a difference in other situations, so I did another test:
  1. Copying a 3.75GB file from the internal drive to the external drive
  2. Duplicating the same file on the drive using the Finder duplicate command
This did produce some measurable differences:
USB copy 1:45
Firewire copy: 1:10

USB Duplicate: 3:21
Firewire Duplicate: 3:07
So it does make a difference; I'm just trying to figure out if it would make a noticeable difference in my day to day work. Copying to another drive - unless it's also a Firewire drive - won't be improved. Also, Log & Transfer will probably not improve because I'll be importing through USB.

The one area where it might improve things is playing back multiple streams. The only problem is; how do I test and quantify that? If anyone has any suggestions on how to do so, let me know and I'll try some more tests!


Super9Films said...

FCP rendering to a drive is probably not the best test for drive speed/performance.

Try copying a large group of files to the drive then from the drive for both USB and Firewire. Then compare the times. I think you'll find quite a difference.

Michael Murie said...

I'll give that a try. I was trying to see whether there was any difference in performance for the things I'm doing most of the time, which was why I tried rendering.

I doubt there would be any difference in Transfer & Log from the camera, since that's using USB too.

And most of the time, if I was copying files, it would be from another USB drive, so that's going to be hard to test too....unless I connect multiple Firewire drives to the computer...

Unknown said...

Hey Michael, Do you know of a good comparison chart that shows the different DSLRs that people use for video?

Michael Murie said...


What are you looking for? A list of cameras that do it, a specific list of features, or a comparison?


Unknown said...

Features, comparison, anything. Something like that chart of HDSDI/HDMI recorders you posted a few days ago would be ideal, if you know of anything like that. I'm doing some research for work and don't have a lot of time to research things fully.

Michael Murie said...

I haven't seen one like that, but it's a good idea. I might try and put one together.

Unknown said...

I found this but it seems outdated and incomplete at best:

I'd love to see a chart if you made one.

c said...

As Super9Films alluded to, rendering is generally limited by the processor speed long before ext. hdd transfer speeds.

I'd like to know the results of further testing, if you do any!

Michael Murie said...

I did a couple more tests which show the Firewire drive is faster; if you can get the data to it faster!

I'm trying to figure out a way to test how much it improves playback of multiple streams.

Arri said...

Although unscientific I'm going to say the same old stuff about USB. It works, is universal and handy. But it isn't as reliable as the bulletproof Firewire protocol. Try running critical streams over USB and it often burps and breaks your stream. Try playing a 5 camera multiclip and USB would choke bigtime while Firewire would chug through consistently.
I'm not badmouthing USB at all. Firewire was purpose built and built well. Still viable for a while longer. Newer techs are going to be so much faster though and all this will be moot. Can you say SCSI?

Michael Murie said...

I'm with you Arri.

Firewire should help playback performance if you have multiple streams...if I didn't do such boring/vanilla editing then I'd probably encounter that problem more often with USB drives and would see more of an advantage to the Firewire drive.

And Thunderbolt is coming, which makes this discussion even more moot.