SATA 3.0 Released, Solid-State Drives Rejoice | Hacker News Reader

SATA 3.0 Released, Solid-State Drives Rejoice

tech.yahoo.com

17 pointsvaksel7 comments

I can't seem to find info on this, but it doesn't look like the "trim" ATA command made it into this spec, which is a bit of a shame. This would allow the OS (well, the file system or some smart block layer) to mark regions of the device to be discarded, which would help avoid constantly copying data between banks in SSDs.

Apparently, Win7 has support for it, and on the hardware side, the latest OCZ Vertex firmware apparently does too. Last I heard, Intel are waiting until there's an official spec.

This would be too inexcusable to be true...

They can just make it a point revision or a standardised extension. There is no point in supporting this command on something like a hard disk, so it would always have to be optional anyway. It's quite possible the SATA 3 standard was too far along, or they wanted to make it an extension to SATA 2.

So it's no big deal, but I'd still like to see this standardised ASAP as I think it will drive quite a bit of file system innovation.

The article states SATA 3.0 should have a maximum throughput of 6 Gbits/second. But isn't that the same expected throughput of USB 3.0? What advantage would I have in using SATA 3.0 (Revised?) over USB 3.0?

* SATA is point-to-point, so each drive gets this bandwidth. USB is a tree, bandwidth is arbitrated between devices.

* SATA 2 came into commercial use relatively shortly after introduction. I haven't seen any USB 3.0 devices yet (host adapters/motherboards or nodes) and driver support during the USB 1.1->2.0 transition was pretty bad. Apparently, Windows users will need to wait for Microsoft to release host controller drivers. I suspect they'll only be available for Windows 7. No idea when OSX will support it, presumably Linux will be first, assuming the spec is made available to the developers.

Thanks! I've been trolling the Wikipedia pages for both of them but that's easily the best summary I've come across so far. =)

Also think of stuff like latencies. Current USB drives add like ~5ms of latency. That's doable with normal drives (which have ~10ms access times), but is killing for SSD's (which have <0.1ms access times).

Reply on news.ycombinator.com