Should SATA get updated specs? Example SATA 4.0 @ 36Gbps, 48Gbps or 64Gbps.

[lexluthermiester]

The discussion came up in another thread and it seemed like it might be interesting to find out what everyone thinks about the subject.
Would you like an updated spec for the standard SATA connector?
Are you happy with the way things are?
Are you inbetween somewhere?

Place your vote and discuss below. Remember, forum rules apply. Let keep things friendly and civil!

 

[AusWolf]

I've recently swapped all the drives to nvme in my main PC. Not because of the speed, but for better cable management and easier servicing / less clutter.

As for secondary rigs and HTPCs, the SATA 6 Gbps spec is plenty enough.

 

[eidairaman1]

The discussion came up in another thread and it seemed like it might be interesting to find out what everyone thinks about the subject.
Would you like an updated spec for the standard SATA connector?
Are you happy with the way things are?
Are you inbetween somewhere?

Place your vote and discuss below. Remember, forum rules apply. Let keep things friendly and civil!

SATA is a good standard thats been around Circa 2002/2003, it should have its bandwidth increased or the standard merged into SAS.

Temperature and stealing of pcie slots by m.2 is reason enough to keep SATA.

5800 rig is zippy with 11 on a 250GB MX500 SATA.

 

[lexluthermiester]

SATA is a good standard thats been around Circa 2002/2003, it should have its bandwidth increased or the standard merged into SAS.

The idea is to keep the connector the same and backward compatibility intact.

 

[johnspack]

Well, I guess it's how many nvme drives can I plug into a motherboard? I have 4 sata ssds, used to have 5. Would like more. If they would give more pci-e lanes
on modern motherboards and give you 4-6 nvme slots, it would be a no brainer. Otherwise... sata ssds are going to be around for awhile.

 

[Tarte]

I think a update to SATA 4.0 @ 36 Gbps will be good enough.
I have 4 SATA SSD but no Nvme and it is OK for me.

 

[Valantar]

I see no reason for an updated SATA standard. The current spec is plenty for HDDs, and NVMe handles anything else. m.2 is vastly more practical, and there's u.2 if you need off-board mounting - which is easily adapted from any m.2 slot or PCIe slot. The lack of extra controllers is a boon for efficiency and makes for simpler chipsets and motherboards, and the inherent flexibility and scalability of PCIe is great. Thr only drawback is the increased need for PCIe controllers, but faster SATA would require that as well.

 

[R-T-B]

I wouldn't mind a new pcie over sata connector standard ala sata express, but sata simply has overheads and latencies that have no place in the high speed ssd market.

 

[TheLostSwede]

The discussion came up in another thread and it seemed like it might be interesting to find out what everyone thinks about the subject.
Would you like an updated spec for the standard SATA connector?
Are you happy with the way things are?
Are you inbetween somewhere?

Place your vote and discuss below. Remember, forum rules apply. Let keep things friendly and civil!

You know, it was already done with SATA Express which was the biggest failure in terms of connectivity standards ever. Pretty much every single motherboard had the connectors, cables were provided, but afaik not a single retail drive was ever made.

Then we got U.2, where is that in the consumer space? Seems popular in the server space though.

I don't see why we need yet another standard, when we've already have had two that didn't end up replacing SATA, where at least the first one was 100% backwards compatible.

The idea is to keep the connector the same and backward compatibility intact.

SAS tried that, they ended up topping out at 12 Gbps, with the next version getting a different connector.
That tells you that going beyond 12 Gbps on the same connector isn't possible.
SAS is pretty much PCIe/NVMe based now and shares a lot with U.2.

 

[lexluthermiester]

You know, it was already done with SATA Express which was the biggest failure in terms of connectivity standards ever.

Right, because it wasn't compatible with the existing standard and connector.

SAS tried that, they ended up topping out at 12 Gbps, with the next version getting a different connector.
That tells you that going beyond 12 Gbps on the same connector isn't possible.

No, it just means the research wasn't done. If 40Gbps can be done over USB through that connector and cabling, 36Gbps can be done over SATA with the same or enhanced connector that is still compatible.

 

[Valantar]

If they could make a PCIe x1 link work over something resembling a regular SATA cable that could be interesting, but anyone who has touched PCIe 4.0 risers knows transmitting those kinds of speeds over a cable is... tricky at best, requiring high quality cabling with very good shielding etc - which is expensive. So either you need expensive, semi-exotic cabling or this tops out at a single lane of PCIe 3.0 (~900MB/s theoretical, ~800 real world?), which isn't much of an improvement over regular SATA. And if you're going PCIe anyhow, why not make it NVMe while you're at it?

This is why m.2 has taken over: motherboard mounting is far simpler, requires no fussy or expensive cables, and SATA is there as a fallback for slower storage if needed. There's no need for controllers, just more CPU/chipset lanes, which faster SATA would need as well. And the vast majority of people don't need that, crucially. m.2 absolutely has its drawbacks, and certainly isn't ideal for those wanting massive capacities (consumer platforms don't have the lanes, after all), but there doesn't seem to be a market for some in-between solution.

 

[SL2]

Right, because it wasn't compatible with the existing standard and connector.

SATA Express is backwards compatible.

 

[Valantar]

Right, because it wasn't compatible with the existing standard and connector.

No, it just means the research wasn't done. If 40Gbps can be done over USB through that connector and cabling, 36Gbps can be done over SATA with the same or enhanced connector that is still compatible.

SATA has 7 pins. USB-C has 24, of which 8 are high speed data pairs, plus more for power, ground, USB 2, etc. Could you double the SATA connector pin count while staying compatible? Probably - USB 3.0 did that. But it would be expensive and complex. And, most importantly, it would require more expensive, thicker cabling, all for a standard that still wouldn't be a match for NVMe. And u.2 already exists for that in 2.5/3.5" form factors.

 

[TheLostSwede]

Right, because it wasn't compatible with the existing standard and connector.

But that's exactly what it was. You could do two times SATA or one times SATA Express.

No, it just means the research wasn't done. If 40Gbps can be done over USB through that connector and cabling, 36Gbps can be done over SATA with the same or enhanced connector that is still compatible.

Ok, sorry, but now you're just talking smack. SATA and USB 4 uses very different cables and VERY different signalling.
First of all, SATA only has one pair of data pins for either transmitting or receiving data, whereas USB 4 has four pairs.
SATA has nine ground pins, due to much higher EMI inside a chassis than outside, where USB 4 can make do with four.
However, most importantly, USB uses 128b/132b encoding and has done so since USB 3.1 (10 Gbps), where SATA is limited to 8b/10b encoding like USB 3.0.
This encoding should be possible to improve upon, but you're not going to be able to reach speeds above 12 Gbps still, since you simply don't have enough data lines in a SATA cable.
This is why SAS moved on to a different cable/connector.
So I'm pretty certain enough research has been done. Just because you're not happy with the fact that it doesn't work the way you want it to work, doesn't mean people haven't tried.

 

[elghinnarisa]

Right, because it wasn't compatible with the existing standard and connector.

It was, though.

If they could make a PCIe x1 link work over something resembling a regular SATA cable that could be interesting, but anyone who has touched PCIe 4.0 risers knows transmitting those kinds of speeds over a cable is... tricky at best, requiring high quality cabling with very good shielding etc - which is expensive. So either you need expensive, semi-exotic cabling or this tops out at a single lane of PCIe 3.0 (~900MB/s theoretical, ~800 real world?), which isn't much of an improvement over regular SATA. And if you're going PCIe anyhow, why not make it NVMe while you're at it?

You mean like SATA Express?
"SATA Express interface supports both PCI Express and SATA storage devices by exposing two PCI Express 2.0 or 3.0 lanes and two SATA 3.0 (6 Gbit/s) ports through the same host-side SATA Express connector (but not both at the same time). Exposed PCI Express lanes provide a pure PCI Express connection between the host and storage device, with no additional layers of bus abstraction."

Edit: I was too slow!

 

[TheLostSwede]

If they could make a PCIe x1 link work over something resembling a regular SATA cable that could be interesting, but anyone who has touched PCIe 4.0 risers knows transmitting those kinds of speeds over a cable is... tricky at best, requiring high quality cabling with very good shielding etc - which is expensive. So either you need expensive, semi-exotic cabling or this tops out at a single lane of PCIe 3.0 (~900MB/s theoretical, ~800 real world?), which isn't much of an improvement over regular SATA. And if you're going PCIe anyhow, why not make it NVMe while you're at it?

This is why m.2 has taken over: motherboard mounting is far simpler, requires no fussy or expensive cables, and SATA is there as a fallback for slower storage if needed. There's no need for controllers, just more CPU/chipset lanes, which faster SATA would need as well. And the vast majority of people don't need that, crucially. m.2 absolutely has its drawbacks, and certainly isn't ideal for those wanting massive capacities (consumer platforms don't have the lanes, after all), but there doesn't seem to be a market for some in-between solution.

You can already go up to 12 Gbps over the cables, as proven by SAS, but beyond that, it's not possible due to data lane limitations. This is why the SATA cable can't reach higher speeds, but was used by some companies early on for USB 3.1 (10Gbps) signals.

 

[Valantar]

You mean like SATA Express?

Nah, I meant with a single standard SATA connector, not the grafted monstrosity that was the SATAe connector.

You can already go up to 12 Gbps over the cables, as proven by SAS, but beyond that, it's not possible due to data lane limitations. This is why the SATA cable can't reach higher speeds, but was used by some companies early on for USB 3.1 (10Gbps) signals.

That's cool, I didn't know that. Guess that partly also explains why anything faster was abandoned in lieu of PCIe NVMe - why develop a new standard with new controllers and new cabling when NVMe and u.2 already exists?

 

[SL2]

You can already go up to 12 Gbps over the cables, as proven by SAS, but beyond that, it's not possible due to data lane limitations. This is why the SATA cable can't reach higher speeds, but was used by some companies early on for USB 3.1 (10Gbps) signals.

That was my first thought when I saw the poll. If 36 - 72 GB/s was that easy to do, without too much added cost to the board, cables, or limiting cable length, it would have been done a long time ago. I'm not saying it's impossible, tho.

 

[lexluthermiester]

SATA Express is backwards compatible.

Never used it. Never cared either. The connector was not the same. Even if the protocol was compatible, the connector wasn't.

It was, though.

No it isn't. The connectors are very different.


Folks,
I'm really not interested in getting into a debate over the physics of data transmission over copper wiring. Anything that can be transmitted over circuit pathways on a motherboard can also be transmitted over a properly built cable. Example? HDMI and DisplayPort. Each of those types of data transmission can sustain high bandwidth over cabled distance without signal degradation. It's a matter of proper voltage & signal attenuation. The protocol is almost irrelevant. If the proper research had been done, the speeds I'm suggesting in the poll above would be possible or very close in an updated SATA protocol spec.

The Poll question is not worded: "Do you think it's possible?". As a tech nut and scientist, I already know it's possible, and so does everyone else who uses high bandwidth data transmission cabling. We can safely put that question to rest.

So the question of the poll stands as valid: Would you want a newer SATA spec?

 

[SL2]

The connector was not the same.

A SATA Express host port has two SATA ports in it, that's the very reason for its design: backwards compatibility.

I'm really not interested in getting into a debate over the physics of data transmission over copper wiring.

There's nothing to debate here.

Intel 9-series Chipset Could Feature SATA Express Interface

Intel may be a couple of months away from launching its Core "Haswell" processor family, and the 8-series "Lynx Point" family of chipset to back it, but is already talking to partners about the platform's successor, slated for the second quarter of 2014. Core "Haswell" introduces the industry to...

www.techpowerup.com

SATA Express is being drafted by SATA-IO into what could be SATA revision 3.2. It is essentially SATA link-layer over PCIe physical layer, although it manages to retain backwards compatibility with older SATA standards (at reduced performance, of course).

 

[TheLostSwede]

Never used it. Never cared either. The connector was not the same. Even if the protocol was compatible, the connector wasn't.


No it isn't. The connectors are very different.


Folks,
I'm really not interested in getting into a debate over the physics of data transmission over copper wiring. Anything that can be transmitted over circuit pathways on a motherboard can also be transmitted over a properly built cable. Example? HDMI and DisplayPort. Each of those types of data transmission can sustain high bandwidth over cabled distance without signal degradation. It's a matter of proper voltage & signal attenuation. The protocol is almost irrelevant. If the proper research had been done, the speeds I'm suggesting in the poll above would be possible or very close in an updated SATA protocol spec.

The Poll question is not worded: "Do you think it's possible?". As a tech nut and scientist, I already know it's possible, and so does everyone else who uses high bandwidth data transmission cabling. We can safely put that question to rest.

So the question of the poll stands as valid: Would you want a newer SATA spec?

How is SATA Express not the same? Just because you don't care, doesn't make you right.
You could plug in two SATA cables per one SATA Express port, it was just the small extension on the side that enabled SATA Express.

The fact that you refuse to be bothered by things like the physical limitations of a standard you want your updated version to be backwards compatible with, clearly shows that you don't understand how these things work. I thought you were more sensible and more of a tech than this.

 

[SL2]

This is the one I was looking for.

The fact that you refuse to be bothered by things like the physical limitations of a standard you want your updated version to be backwards compatible with, clearly shows that you don't understand how these things work. I thought you were more sensible and more of a tech than this.

IDK what's going on. You posted facts, and the he started debating it. When it didn't go his way he tries to shut it down? Yeah, no, that's not going to happen here, never worked before lol.

 

[lexluthermiester]

clearly shows that you don't understand how these things work.

Do not mistake my indifference for ignorance.

There's nothing to debate here.

You're so right about that.

IDK what's going on. You posted facts, and the he started debating it. When it didn't go his way he tries to shut it down? Yeah, no, that's not going to happen here, never worked before lol.

That's not what happened.

This poll was NOT created to discuss the technical merits of the idea of an advanced SATA spec, only whether or not users would want it in a form-factor of the original connector.

Quit with the condescending comments and the off-topic shitposting. Thank You.

 

[OliverQueen]

What I would prefer going forward is m.2 sockets fitted to cases (replacing 2.5" SSD sleds) that connect to nVME channels on the motherboard (as more channels become available on CPU's) via a slim cable similar to SATA. I know that something like this is already available in the server world (u.2? although could be wrong with the terminology as not that up on enterprise gear), but would like to see this become mainstream in consumer systems.

As for new SATA specs, I think that it is coming towards the end & moving forward with nVME would be a better option IMHO. If SATA manufacturing was ceased & put into nVME manufacturing/development, the prices of the lower end nVME (which is still faster than highest end consumer SATA), we would start seeing the cost per Terabyte on nVME come down to similar levels of high end SATA storage with lower power overheads. If going forward with SATA type devices, consumer grade SAS would be a nice replacement.

 

[Valantar]

I still just don't see the need. More product in the market doing the same thing? More controller designs competing for attention? Even more pressure on PCIe lanes from additional high speed I/O controllers, and a return to the Skylake-era bingo chart of "use one port, it deactivates two others"? No thanks. We have SATA for HDDs, mass storage, and lower speed SSDs - which are still fine. We have NVMe in all its various forms for faster storage - and it's extremely flexible.

Have an m.2 SSD but no slot, or want it out from under your GPU? Get one of these

Or one of these for tons of slots

Want it entirely off the motherboard? No problem

If the problem is that you've got X lanes and can't split them into whatever your drives need, it's far easier to add bifurcation support to a platform than to engineer new host-side and device-side controllers and whatever cabling they use to interface. It's just far simpler and more efficient across the board for the storage devices to be native PCIe.

What I would prefer going forward is m.2 sockets fitted to cases (replacing 2.5" SSD sleds) that connect to nVME channels on the motherboard (as more channels become available on CPU's) via a slim cable similar to SATA. I know that something like this is already available in the server world (u.2? although could be wrong with the terminology as not that up on enterprise gear), but would like to see this become mainstream in consumer systems.

That could be kind of neat, but m.2 drives aren't really meant for mounting to a case, are small and fragile, and don't have latching connectors, so you'd need a backplane for them - which gets expensive rather quickly as it's a bespoke PCB design for each case, more or less. Add to that the cost of PCIe 4.0 or higher compatible cabling, and you're looking at a pretty expensive setup - easily a couple hundred dollars for a few drives. This is also pretty easily bypassed with adapters like the ones above. M.2 or plain PCIe are both easily adapterd to u.2, and there are u.2 caddies for m.2 drives that mount to 2.5" mounts in your case.