Are FireWire connections crave-worthy?

OK, I admit it. I want options, even though, most of the time I don't have much use for them. Take pocket-size external hard drives, for example. I want them to be compact, light, pretty, bus-powered, and especially I want them to support USB 2.0, FireWire 400, and FireWire 800. (I would take eSATA, too, though, that wouldn't make sense until it's bus-powered capable). However, the truth is, I've used mostly just the USB 2.0 connection for personal purposes. So today, I decided to find out if FireWire 400 and FireWire 800 are worth it, as far as the throughput is concerned.

In CNET Labs, like most devices, hard drives are tested in the "real world" approach. This means the test might not show the best of what the device can do, but how well it does in a real world situation. This is the reason why, if you have read the recent reviews of pocket-size external hard drives, you will see the margin in throughput between different hard drives or different connections of one hard drive are relatively small. Our 10GB test data consists of hundreds of folders and small files (resembling the content of a typical "My Documents" folder) that create a lot of overhead for the copying process.

I tweaked this test a bit for this experiment by using single 10GB file. We have only two pocket-size drives in CNET Labs that have all three types of connections: the OWC Mercury and the G-Tech G-Drive. I put them through the test and here are the scores (in Mbps):

(Credit: Dong Ngo/CNET Networks)

(Credit: Dong Ngo/CNET Networks)

So from the charts, there are three interesting revelations:

First, it doesn't matter what connection you use, with the same amount of data, it's faster to transfer when the data is in the form of a (few) large single file(s) than of multiple small files. This is another reason why you should compress your data into one single file (in ZIP or RAR, or any other types of compression format) before copying it onto another storage device, besides the apparent fact that compressing shrinks the size of the data itself. Of course, this only makes sense if you don't factor in the time needed for the compressing process.

Second, USB 2.0 is much slower than Firewire 400. Judging by the score, FireFire 400 could be up to 40 percent faster than USB 2.0. This is very interesting because on the specs; USB 2.0 is slated to have the transfer rate of up to 480Mbps, while that of FireWire 400 is only 400Mbps.

And last but not least, FireWire 800 is faster than FireWire 400 but not by a big margin, just a few percentages points, in writing. In reading, however, it's actually slower by about the same margin. This is also very interesting as FireWire 800 is slated to potentially offer twice the speed of FireWire 400.

So the conclusion is: yes FireWire connections are definitely worth it when it comes to speed, however, there's not much difference between the two. This means if your computer has a FireWire port, by all means, go ahead and get a drive that can take advantage of that. However, it' s pretty easy to decide which one to pick between FireWire 400 and FireWire 800: either one is fine.

A little disclaimer: these findings are only applicable to external pocket-size bus-powered hard drives working with a Windows XP machine. Full-size external hard drives with separate power adapters might yield different performance patterns.

In the end, personally, I still want my drive to support all these three connections, though I might just continue to use just the USB 2.0. But that's just me.

[chandave]

Firewire implementations use dedicated hardware. USB implementations usually use the host's CPU to move data. Therefore, Firewire will always be faster unless the host CPU does nothing except handle USB functionality.

Also, are you sure you are not hitting the I/O limit of the HD or disk enclosures when testing Firewire 800?