Is anyone else surprised that SATA 4.0 isn't on the cards yet?

[mikeymikec]

I'm just looking at a review for a Seagate 600 SSD, and that some of the sequential read speeds hitting 550MB/sec, and I was thinking that revisions to the ATA/SATA standards came out quite a bit before any drives were capable of pushing the boundaries. I'm wondering if an unusual difficulty has been encountered in squeezing a bit more out of SATA.

I just did a quick google which suggests that the theoretical bandwidth capability is 744MB/sec (so I stand a bit corrected as I thought the maximum was 600MB/sec).

Still, it is a bit surprising given how quickly SSDs are advancing. Someone on another forum was saying that the average person doesn't need that sort of bandwidth throughput, but I think that argument could be directed at 100MB/sec with equal validity. There are times when as much disk I/O as possible is useful to me but I don't think my use is quite average (nor is it extraordinary), but to the average person browsing or dealing with their photos/music?

 

[BrightCandle]

SSDs are looking to make their own interface designed for their technology and plug directly into PCI-E like the super high performance cards. SATA will likely continue for spinning disks and the current limit will more than service them well, but its not a good interface for SSDs and they have become popular enough to actually start to develop something with them in mind.

 

[mikeymikec]

SSDs are looking to make their own interface designed for their technology and plug directly into PCI-E like the super high performance cards. SATA will likely continue for spinning disks and the current limit will more than service them well, but its not a good interface for SSDs and they have become popular enough to actually start to develop something with them in mind.

Urgh. The words "won't be supported by Windows 7" spring to mind.

 

[Hellhammer]

SATA Express is the future. While SATA-IO has said that developing SATA 12Gbps wouldn't be too hard since most of the work has already been done for SAS 12Gbps, the added complexity would increase costs quite significantly due to increased die size. SATAe is a more efficient solution as it relies purely on PCIe.

 

[Cerb]

Urgh. The words "won't be supported by Windows 7" spring to mind.

If it supports AHCI, and that implementation is compatible with Intel's, it shouldn't end up being a problem. Even if drivers end up being needed, Windows 7 can get them from a USB stick, Windows 7 will be supported for another 5+ years, and depending on the next few versions, MS might have to extend support like they did for XP, due to businesses not clamoring for Windows n+1.

 

[JellyRoll]

SATA Express will only be 8Gbps for consumer use, and 12 for enterprise.

 

[Hellhammer]

SATA Express will only be 8Gbps for consumer use, and 12 for enterprise.

That depends on the PCIe spec. It's 8Gbps with PCIe 2.0 but PCIe 3.0 increases that to 16Gbps. The enterprise version (SFF-8639) is 16Gbps and 32Gbps.

 

[JellyRoll]

That depends on the PCIe spec. It's 8Gbps with PCIe 2.0 but PCIe 3.0 increases that to 16Gbps. The enterprise version (SFF-8639) is 16Gbps and 32Gbps.

PCIe 3.0 is 8Gb/s. doubling the lanes is 16Gb/s. http://www.sata-io.org/documents/SATA_Express_Article_2012.pdf

PCIe Gen 3 (8Gb/s)provides the bump up in speed that we need with a single lane.
Including a 2nd PCIe lane in SATA Express provides the ability to support up to 16Gb/s, which gives us plenty of headroom for even faster client SSDs. PCIe Gen 4, which is anticipated to come out in 3-4 years,will double the bandwidth to 16Gb/s per lane, so SATA Express has a growth path.

The increase to 16Gb/s comes from adding another lane.

 

[Hellhammer]

PCIe 3.0 is 8Gb/s. doubling the lanes is 16Gb/s. http://www.sata-io.org/documents/SATA_Express_Article_2012.pdf

The increase to 16Gb/s comes from adding another lane.

That's true, but doubling the lanes means it's still 16Gbps (the consumer spec supports up to two PCIe lanes), not 8Gbps as you said. Of course, it'll be up to OEMs to decide how many lanes they use for SATAe but the spec itself is good for up to 16Gbps should the OEMs prefer that. Also, SATAe is PCIe so there's no longer 20% overhead due to 8b/10b encoding scheme.

 

[JellyRoll]

It is either 8 OR 16. so 8 is correct as well.
And that has nothing to do with PCIe 2.0, as you said. You said the jump came from PCie 2.0 to 3.0, not the addition of lanes (as is the case).

 

[Hellhammer]

It is either 8 OR 16. so 8 is correct as well.
And that has nothing to do with PCIe 2.0, as you said. You said the jump came from PCie 2.0 to 3.0, not the addition of lanes (as is the case).

I was referring to the maximum bandwidths supported by the spec. With PCIe 2.0 the speeds are 4Gbps (single-lane) or 8Gbps (dual-lane), so 8->16Gbps is also due to the PCIe 3.0 spec.