USB 3 Makes External USB RAID ''Possible''
Symwave's SOC will take advantage of USB 3.0 and provide RAID support for external drives.
Computerworld is reporting that Symwave, one of the first companies to design silicon for USB 3.0, is claiming that its new USB 3.0 SOC (system on a chip) can be used with external storage devices and provide transfer data rates up to 500 Mbit/sec. USB 3.0 is actually designed to handle transfers of up to 5 Gbit/sec, a huge increase in throughput when compared to the 480 Mbit/sec limit seen with USB 2.0. As an example, a 25 GB HD movie would take 13.9 minutes to transport over a USB 2.0 connection, just 70 seconds over a USB 3.0 connection.
"You're pretty much communicating through a straw," said Gideon Intrater, Symwave's vice president of solutions architecture, referring to the 2.0 limitations. "USB 2 was good as long as you had 100GB on your hard drive, but now it's just way too slow."
Symwave's new SOC, developed for external storage devices including HDDs and SSDs, supposedly offers performance beyond the top speed of SATA. According to Intrater, the chip will allow speeds as high as 500 Mbit/sec because it supports RAID 0 configurations. System builders can take advantage of this feature by installing two external drives that can be addressed at the same time, offering faster data reads. In another scenario, the second drive could mirror the first USB drive.
Intrator also said that USB 3.0 can carry as much as 900 milliamps, making it easier to power a portable RAID array of two drives; USB 2.0 currently only provides 500 milliamps.

5amps... are you serious? wot do you intend on powering? a blender?
Im tired of them kinda mis labling, mbit is bits not bytes, so its like not that amazingly fast. They need to make Mbytes/sec the standard (for now) some people will read this and see 500 mb...and assume the wrong thing...BUT usb 3.0 does look badass and it is about time for it.
agreed. isp's do the same thing. its like who the eff uses those standards. my 6mbit connection is just 600KB/sec so just call it that!!
At any rate Toms, do you have info on Hot Chips Conference that's taking place now?
5 Gbits / sec = 642,5 MBytes / sec
That's pretty fast imo ...
edit: took place. It's over now.
5amps... are you serious? wot do you intend on powering? a blender?
Ya umm 5 amps would fry the hell out of any IC, MUCH less 45nm tech lol, theres a reason stuff is getting 'green' the small structures on the wafers cant stand high power, itll burn the crap.
600KB/sec = 4.8MBits/sec so no, your 6MBit connection isn't 600KB/sec.
1 Byte = 8 Bits
Anyhow, I'm looking forward to this technology - both in terms of USB-based RAID solutions and greater amperage for powering devices will be nice. Simply being able to reliably plug-in a portable USB HD without needing a special 2-plug USB Cable will be nice.
At 5 volts DC, 5 amps is a considerable amount of current. Hardly enough to power a blender, but considerable still. Depending on the speed, most blender AC motors pull between 2-5 amps on a 110 volt line.
I must admit though that I am happy to see such a larger increase in available current. I'm tires of getting messages every time I connect something to my keyboard and I'm told maximum controller power has been exceeded.
Hes saying that with his 6mbit connections he can only achieve rates of about 600kb/s.. Which is about standard for DSL.. When I had 1.5mbit dsl I could get around 150kb/s, my 3mbit connection would get my about 300kb/s and my 6mbit got me about 600kb/s... Comcast is the only provider that has actually provided me with full bandwidth. My newsgroup transfers always pegged at a solid 8mbit, or 1mbyte/sec, except for the first minute or so where Comcast provides up to 24mbit/s with their TurboBoost..
Gives me a big geeky hard on!
One of the main reasons I recommend for users of cable modems not to use the USB hookup but use the NIC whenever possible.
Geez, don't neg-vote the guy for saying something like this, he may very well not understand.
fooldog, these speeds are not chosen arbitrarily, but are based upon the current limitations in the technology and some attempt at prognosticating where the technology will likely be when the standard actually becomes a standard.
Without getting into technical detail, consider the evolution of data transfer rates for hard drives. Forgetting all the MFM and RLL stuff, and jumping right into IDE. Initially, data transfer was controlled via PIO modes. The limitation here was that it required the CPU to be involved in data transfers, and when the amount of data was particularly large, it imposed a tremendous load on the CPU. Then came DMA modes, which allowed data to be written directly to memory without putting a huge burden on the CPU (those were ATA standards 1 and 2). Then came more DMA modes, then Ultra DMA modes. We went from ATA-1 at 3.3 Mb/s to ATA-7 at 133 MB/s (theoretical). Now why not just create ATA-7 well in advance of needing that much? Because it was technologically impossible at the time. The technology simply did not exist.
It's just like the CPUs, we hit 500 MHz back in 1998, 1 GHz back in early 2000, 2 GHz in late 2000, then it wasn't until 2005 that we hit 4 GHz, now a few months shy of 2010 and we're still hovering around 4 GHz... WTF is up with that? We should be pushing 10 GHz by now. Well, we hit the wall. We're up against the theoretical limit of the technology and no one has come up with the next trick yet. So instead of going faster, they started putting more cores (essentially multiple processors) on the same chip.
I'm sorry, this is longer than I wanted it to be. The bottom line is that creating a standard is no easy task. There are a multiude of factors that have to be taken into account. Anyway, I hope this helps.