Separate names with a comma.
Discussion in 'Computers' started by Jon_Are, Dec 10, 2004.
Simple question - all other things being equal, which is faster?
USB 2.0 = 480MB
Firewire 400 = 400MB
Pretty straight forward. I've did some tests w/ a Canon CanoScan 9900F, which has both USB 2.0 and FW400 connection, connected to a G5 Dual 2GHz. Scanning a full page color magazine cover, the USB 2.0 was a tick faster.
Given ease of use, compatibility and cost, I'd go with USB 2.0.
In theory USB 2.0 is faster. In practice that simply isn't true. They are pretty close in practice, but firewire almost always uses about half the CPU utilization of USB2.0. Also, USB2.0 shares bandwidth across all attached devices. With firewire, you can daisy-chain devices and they all get 400Mbps. Hopefully we'll see Firewire800 catch on in the next couple years. It is the first true replacement for SCSI -- speed-wise at least.
Although USB 2.0 has faster raw throughput at 480 Mbps, in practice, FireWire (1394a) is significantly faster (at 400 Mbps). In real world, FireWire can sustain 50 MB/sec. throughput whereas USB 2.0 can barely sustain 40 MB/sec. and at greater CPU utilization.
You can also get FireWire 800 interfaces now (1394b), which is significantly faster than both (at 800 Mbps). And SATA II external will leave all 3 in the dust.
Rosewill sells a FW800 ext. HD enclosure, but it's $90+. Since my PC only has USB 2.0 ports, I'll stick with USB 2.0. I don't use the G5 that much anyway.
Realize, USB is aggragate 480, but Firewire is ptp 400. Which means each device in a firewire system is afforded 400mbps of actual bandwidth; meanawhile, on USB, all devices share a common pool of 480.. so the more devices you have.. etc... etc..
That's all theoretical. I mean, have you seen people daisy chain 63 FW ext. HDs together and run a benchmark on all 63 HDs at the same time. Even if that's possible, you'll end up w/ serious bottlenecking. So what's the point?
IMO, the issue currently is that most firewire implementations use a lot of software loaded into PC memory, while USB is handled in the USB chip itself. This means USB has the potential to suffer less interruptions than firewire in data streaming operations.
This will change over time as firewire becomes even more mainstream.
Do you have an actual benchmark of someone getting 480Mbps on USB2.0? Nope, well then that's theoretical too . . . "so what's the point?" I've been in a recording studio with an I/O device, a couple HD's, and a DV camcorder all daisy-chained and they worked fine all going at the same time (using the I/O devices to record to the external drives while using the camcorder). If there are camcorders that can hookup via USB2.0 (I don't know of any, but there may be some out there) and you had the camcorder going you wouldn't be able to do much of anything else on any other USB port on the same hub.
I didn't say USB 2.0 can do "all that" and more. The point I'm trying to get across the whole time is USB 2.0 is more widely available and affordable than FW, especially FW800. I don't have to proof anything 'cause I didn't say USB 2.0 is the best. FW claims it uses isochronous transfer method, and can daisy chain up to 63 devices. Putting the two together, can the 63rd HD in the chain sustain the same transfer rate as the 1st HD in the chain? That's what I'd like to find out.
For audio signals the most desirable transfer methodology is asynchronous, not isochronous or synchronous.
USB with an asynchronous (ASIO) driver implementation is desirable.