Bye, Bye SATA III - Hello SATA 3.2

[MrSuperb]

http://www.engadget.com/2013/08/09/sata-3-2-ratified/

 

 

The 500 MByte/s-era might find it's end soon :D

 

I hope Notebook manufacturers will use a common standard for those integrated (non-2,5") SSD modules.

[CoolBeans]

Finally, now we can have built in Raid 0 in standard SSD's without it being bottlenecked.

[Keystone Nyan Cat]

It will still take years for it to get well established. For perspective, look at USB 3.0 which is still not that widespread and properly utilized...

[EChondo]

Well the article says that this will "double" the current speed, but we only have 500MB and they are talking about 16Gb(2GB).

 

Are they misinformed/typo or does SATAIII really have a 1GB limit and not 500MB?

[UselessBread]

It will still take years for it to get well established. For perspective, look at USB 3.0 which is still not that widespread and properly utilized...

But I think that that is mostly because there is no need for a faster external interface right now.

[AraSarkisyan]

By the time everyone starts using this sata 4 will be out. \

[Glenwing]

I hope Intel increases Broadwell's main bus to more than 20Gbps we have right now, or the CPU won't have enough bandwidth between it and the chipset to operate more than 1 storage device at a time...

 

Honestly I'd rather just see SATA replaced entirely with some kind of "S-PCIe" Storage PCIe which would just be some optimally-designed right-angled PCIe connectors where the current SATA connectors are, that just have power and 2x lanes of PCIe 3.0... then we could not only have 20Gbps of bandwidth for each device but power over the same cable, and it would go directly to the PCI Express controller and not through the main PCH bus.  And when PCIe 4.0 comes along then 2x lanes would give us 40Gbps :D :D

[That Norwegian Guy]

Did you all miss this part: "The new standard centers on SATA Express, which lets SATA storage ride on the PCIe bus"

 

I don't know about you, but to me that's a dealbreaker. This is not the SATA IV you were looking for... (Haswell-E will have it though)

 

Well the article says that this will "double" the current speed, but we only have 500MB and they are talking about 16Gb(2GB).

 

Are they misinformed/typo or does SATAIII really have a 1GB limit and not 500MB?

 

Because most people will have at least 1 GPU on that bandpass and such the true bandwith would be more like "double" though it's not accurate.

[Glenwing]

Did you all miss this part: "The new standard centers on SATA Express, which lets SATA storage ride on the PCIe bus"

 

I don't know about you, but to me that's a dealbreaker. This is not the SATA IV you were looking for... (Haswell-E will have it though)

 

 

Because most people will have at least 1 GPU on that bandpass and such the true bandwith would be more like "double" though it's not accurate.

 

The new standard will still likely go through the chipset's secondary PCIe lanes though from information so far, thus will still be limited by the main bus if that is accurate.

[bcredeur97]

im wondering if PCIe lanes will be affected.... we only have so many with haswell anyway.... broadwell might have more, or at least it better.... big raid 0 setups will use up all your pcie lanes if not.

[Glenwing]

im wondering if PCIe lanes will be affected.... we only have so many with haswell anyway.... broadwell might have more, or at least it better.... big raid 0 setups will use up all your pcie lanes if not.

 

SATA Express basically uses PCIe 2.0 x4 lanes, which means it will most likely go through the chipset's secondary PCIe lanes, and won't take any from the CPU.

[bcredeur97]

SATA Express basically uses PCIe 2.0 x4 lanes, which means it will most likely go through the chipset's secondary PCIe lanes, and won't take any from the CPU.

I see. awesome then.

[ChrisxIxCross]

Couldnt we just go with internal thunderbolt as the new standard?

[GoodBytes]

While this is good news. SSD manufacture should really focus on small file performance, which is what we interact with all the time, and most of the time.

And the performance of small file is not even close to SATA-2 max speed.

[Flamewalker]

Pretty cool for raid and eSATA for that matter :P

But yer what Goodbytes said, boosting seq read is meh, they need to boost smaller file sizes,still its a good thing that there will be more bandwith.

[Kuzma]

Couldnt we just go with internal thunderbolt as the new standard?

^ This. You seem to be onto something.

[Glenwing]

Couldnt we just go with internal thunderbolt as the new standard?

 

I don't want $50 hard drive cables  :wacko:

 

Thunderbolt is expensive due to the multiplexing chips and hardware in the cables it needs to support all those kinds of different things simultaneously...

[GoodBytes]

Not to mention that it creates overheads over simply having the SSD directly on PCI-E, much like PCI-E SSD cards

[eshbop]

I don't see many consumer uses that will be reading or writing large enough sequential blocks that makes increasing SATA speeds too important. It's random I/O they need to focus on. If I could buy a drive with greatly boosted Random reads/writes I'd happy sacrifice sequential speeds.

 

Though in saying all that, current SSDs are pretty damn snappy in consumer use cases already.

[ChrisxIxCross]

I don't want $50 hard drive cables :wacko:

Thunderbolt is expensive due to the multiplexing chips and hardware in the cables it needs to support all those kinds of different things simultaneously...

Good point, however Thunderbolt isnt very widespread in the market, so naturally it will be expensive.I guarantee you that if they began implementing tb into everything the costs of production would go wayy down.

[Glenwing]

Good point, however Thunderbolt isnt very widespread in the market, so naturally it will be expensive.I guarantee you that if they began implementing tb into everything the costs of production would go wayy down.

 

Only to a certain point, it's not just because of its rarity.  There is hardware, microchips and stuff in the cables.

[MrPete1985]

what happened to 3.1?

[GoodBytes]

what happened to 3.1?

 

We are already using 3.1

Here is the list of versions: http://en.wikipedia.org/wiki/Serial_ATA#SATA_revision_3.1