New SATA specs

[Bozo]

http://www.amtdrt.inlumen.com/bin/story?StoryId=CqisGqbKbmteXyJa2mdy


Bozo :mrgrn:

 

[BooST]

Finally! No more broken weenie connectors!

 

[ddrueding]

-- An internal multi-lane cable and connector assembly for
streamlining connections between multiple internal host ports
and internal devices or short backplane.

YAY! :mrgrn:

 

[Howell]

Finally! No more broken weenie connectors!

Where did you see that the connectors were changing? The closest thing I saw was: "Among the features of the enhanced technology is that no new cables and connectors are required to support the higher signaling speeds." This only indicates that a connector change is not required.

 

[Howell]

Finally! No more broken weenie connectors!

Where did you see that the connectors were changing? The closest thing I saw was: "Among the features of the enhanced technology is that no new cables and connectors are required to support the higher signaling speeds." This only indicates that a connector change is not required.

 

[sechs]

Since when did 3Gbps=300MB/s?

The only new connectors are internal backplanes, external backplanes, and external single. No new internal single.

 

[blakerwry]

when you add overhead.. remember 1.5Gbps raw speed = 150MB/sec data rate for SATA 1.0

 

[sechs]

Are you saying that there's 84MB/s of overhead? That's awfully inefficient....

 

[blakerwry]

Efficiency isnt the name of the game, it's about speed. And if you can get 150MB/sec out of 1.5gbps raw bandwidth it's better than getting ~100 out of a 133MB/sec interface.

anyway, it's only 75MB/sec lost (decimal gigas). Personally, I dont mind loosing 2 in order to send a byte of data.

 

[Fushigi]

I haven't read the spec, but maybe it uses a start bit/stop bit sequence like old modems. Or it could use about 10% for CRC checks or other forms of error detection/correction.

 

[blakerwry]

For every byte of data transferred it sends 10 bits instead of 8. You can read the specs here: http://www.serialata.org/collateral/zipdownloads/serialata10a.ZIP It should have the answer.. although it might be burried.



One thing that I thought was interesting is that a S-ATA link is always at 100% usage capacity even if no useful data is actually being transferred. This is in contrast to P-ATA which doesnt transfer information unless it has something useful to send.

 

[GIANT]

This is essentially old news rehashed.

We already knew (at least I did for several months) there was to be a new connector for external SATA, though there were already some small outfits already doing external SATA with existing "internal" cables.

A "multi-lane" cable is a similar idea to what a PA snake is (Public Address mixer <---> stage box). I don't know exactly what they've proposed, but you will likely see a bunch of familiar SATA connectors on one end of this new cable that plug into your SATA RAID controller, and on the other end of the cable, one or more new block-style connectors that may have new block connectors that handle 4 drives per connector which plugs into the back of a SCA type of SATA drive housing. In between these connectors will be a multi-element cable. I'm guessing at a 4-drives-per-connector paradigm, since it seems that 4 drives is the lowest common denominator in multi-drive RAID (4, 8, 12).

 

[mubs]

The Inquirer reports Intel's clarification of SATA speeds here.

 

[sechs]

Oh. So they're using decimal bits, rather than some useful number.