SSD hitting 6Gb/s ceiling?

[Rusting In Peace]

I read the first sentence of this new Tom's article reviewing the new OCZ pci-e SSD cards:

Solid-state tech marches on, and we're already approaching SATA's 6 Gb/s ceiling.

I've seen no reviews of systems or controllers that manage 768MBps read never mind write.

Does anyone have any examples of a system actually hitting this? The April beast RAID system, albeit using SATA2 drives, couldn't saturate SATA2 fully.

I'll give you bonus points if it's not an enterprise solution!

 

[JohnnyLucky]

Here is a link to a review of the SATA Rev3.0 6Gb/s and Rev2.0 3Gb/s capabilities of nine different motherboards:

http://www.bit-tech.net/hardware/storage/2010/10/28/are-on-board-sata-6gbps-ports-fast-enough/1

A motherboard is on the opposite side of the SATA Rev3.0 6Gb/s datalink so it's capabilities should also be considered.

 

[tecmo34]

The one comment about the bit-tech review is it was based off the older C300, which was the original SATA III SSD. The new OCZ Vertex 3 Max IOPS does push the 500 MB/s plus range.

 

[sminlal]

I've seen no reviews of systems or controllers that manage 768MBps read never mind write.

SATA Version 3 has a bandwidth of 600 MByte/sec. SATA uses 10/8 encoding which requires 10 bits at the physical media level to transmit one byte of data.

 

[JohnnyLucky]

^5 +1 what sminlal said!

I knew there was something that reduced the rate but couldn't recall what it was.

 

[Rusting In Peace]

Thanks gents. All very interesting.

Judging from your post JohnnyLucky, motherboards are hitting 370MB/s at best. So there is plenty of capacity left there.

On paper the OCZ Vertex 3 MAX IOPS can manage 550MB/s read. So I guess this is approaching the capacity of SATA3.

Whilst the OCZ seems rather tasty, any PCI-e storage device not using SATA here isn't relevant I'm afraid. I mean the FusionIO ioDrive Octal can manage 6000MB/s read!

So:
■ Why the hell was SATA3 only doubled in performance over SATA2?
■ I see no SATA4 standardization coming anytime soon?
■ Are we going to see significantly more SSD PCI-e solutions because of this limitation?

 

[sminlal]

I'm not a hardware engineer, but I strongly suspect that one of the factors is that it's a lot easier to maintain the forward/backward compatibility between different versions of SATA if you stick to doubling the speed with each generation.

 

[JohnnyLucky]

There is no industry initiative to develop SATA Rev4.0.

PCI-e 3.0 was officially adopted last year. Motherboards with PCI-e 3.0 slots should start appearing in a few months.

NEWS FLASH: ASRock and Fatal1ty first to the PCI-E 3.0 party with Fatal1ty Z68 Professional

http://www.tweaktown.com/news/20044/asrock_and_fatal1ty_first_to_the_pci_e_3_0_party_with_fatal1ty_z68_professional/index.html

ASRock product page:

http://www.asrock.com/mb/overview.asp?Model=Fatal1ty Z68 Professional Gen3

 

[groberts101]

Not to mention that the HDD/SSD tech just doesn't double its speeds very often.

and yes.. PCI-E solutions are on the rise. Even raidcards are enjoying increased sales because of the need for older platforms to take advantage of these 6G SSD's.. but just like sex.. speed sell's. Also consider that firmware advancements have come quite a ways to allow long-term trimless environments for many SSD's, which of course gives much more flexibility for some controllers that could not previously be used in raided/non-trim pass-through environments.

Also of concern is the limited size of some OROM's and the inability for multiple array presence while booting from a PCI-E based SSD. Things are coming around though(2048kb bios update/larger OROM sizes and all) as the demand goes up. Good times ahead, it seems.