25GB in 70 seconds... with USB 3.0? Definitely...

http://thefutureofthings.com/news/5739/25gb-in-70-seconds-with-usb-3-0.html

Could prove interesting, but I gotta wonder what kinda hardware they had hooked up to be able to pump that kinda data out in the first place. Let's do some basic math...

25GB divided by 70 seconds = 357MB per second, roughly (it's actually 357142857.14285714285714285714286 by straight math but heaven forbid, so I rounded it for you)

Considering that not only do you have to be able to read it from the source that fast but also write it to the target that fast, so I'm going to guess they had some form of RAID setup that provided the content (source) as well as a RAID setup that they transferred the content to (target). SSDs, perhaps? I'm thinking so, but a few Velociraptors in each RAID should have been able to muster that bandwidth both ways.

Regardless, it's interesting to see where this will go in the future... 5Gbps... not too shabby.

Definitely looking forward to USB 3.0, but USB 2.0's "theoretical" and "actual" bandwidth left something to be desired; I usually never achieve USB 2.0's 480Mb/s claim, not to mention to CPU overhead associated with USB 2.0.

Let's hope that this new version is really that impressive in real world usage as well.

Best I've managed with USB 2.0 was about 45MB/s sustained and that was from an external hard drive, a Maxtor OneTouch 320GB that I had for a short time. Copied some DVD VOBs from the internal drive (the PC) to the external and got a solid 48MB/s sustained write over the USB 2.0 link as measured by a tool I found that measures the speeds right off the USB controllers (of course I can't find that tool nowadays, was something by one of the USB developers who created the standard itself, a non-public thing).

Then I copied the files back from the external to another location on the PC's internal drive and got roughly 44MB/s coming back in, that's the best I've ever been able to get over USB 2.0, and yes that's sustained, so figure 45 Mbps x 10 (give or take a bit or two since it's not an exact multiple of 8 anyway given the protocols) and I was getting damned near that so-called theoretical max, and I've replicated that test a few times using other external drives also.

Never had issues with it personally, also never had issues with the so-called CPU usage of USB 2.0 compared to Firewire. Testing that external drive (the OneTouch) with HDTune and HDTach showed roughly 2-4% CPU usage, so what's the big deal...

The amazing thing was that Maxtor OneTouch had Firewire 400 also, so I used that to test sustained speeds: 40MB/s peak from PC to external, 41MB/s external to PC, so right where it should be, with CPU usage at roughly 2-3%, right in line with the USB 2.0 results from the same drive.

So whoever is talking shit about USB 2.0 compared to Firewire 400 needs to STFU about it. Firewire 800 doesn't offer that much of an improvement in my testing either, and the lack of support on a wide variety of hardware these days - read that primarily Macs and some high end video hardware - means USB 2.0 has one the battle of interconnects.

Now we'll see what USB 3.0 can do...

Source is no problem, just about any raid5 array with 6+ disks on a decent controller will do 400mb/sec sequential read without any trouble at all.

I've never been able to get 40mb/s off USB 2 :\ Firewire though..it's always 40mb/s+

Strange I did a bunch of file transfer test with 2 identical hard drives between USB 2.0 and Firewire and USB 2.0 beat firewire in every case. Must of been the hard drives.

Personally, I'm not too excited. For hard drives USB 3.0 can never surpass eSATA performance-wise. The increased power capacity will certainly be welcome for bus powered devices - although 2.5" hard drives can already be powered by USB 2.0 and the lack of 12v means that desktop hard drives will still need external power.