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

[Frick]

Tons. The main reason PC cases are metal is to contain the RF noise they emit to avoid it interfering with other devices. High frequency signaling through board traces or wiring gives off noise; coils and power conversion gives off noise, your CPU and GPU operating at several GHz will give off noise, etc.

I meant enough to interfere with that kind of signaling in a meaningful way.

 

[TheLostSwede]

You can even get arm based boards with U.2 connectors, so why not simply use them? They can do SATA as well, but you obviously need a different connector on at least one end.

Mixtile Blade 3

A stackable, high-performance, Rockchip RK3588 single-board computer in a Pico-ITX form factor

www.crowdsupply.com

 

[Chrispy_]

Some people still use big SATA HDD so they "might" benefit from a interface update, but not sure.

For mechanical drives, I am 100% sure that they will not benefit from an interface update.

The best, most expensive, fastest sequential hard drives currently manufactured don't even reach 300MB/s. Modern, typical capacity consumer drives are closer to 200MB/s

SATA needs updating! That updating needs to be done with the existing connector to maintain backward compatibility.

But SATA is a shitty protocol for SSDs.

What you want is a better physical format than M.2 for NVMe, and I agree with you, but it shouldn't be SATA which is the question of your poll. Making SATA better for SSDs is just doubling-down on the mistake.

 

[Aquinus]

Hi,
Cost of U2 is more than enough to laugh at it

Because it's U.2 or because of the capacity you're getting out of it? U.2 doesn't make flash chips any cheaper and you're obviously going to be paying more if there are more flash chips in the package. That doesn't make U.2 expensive, it makes high capacity flash drives expensive. When you're looking at something like 8TB flash drives, no crap it's going to be expensive.

Edit: 1k USD for a 8TB flash drive actually doesn't sound that bad in the grand scheme of things. You'd be within a couple hundred of that if you had a bunch of smaller M.2 drives instead. Also, for anything that's 1-2TB, it makes a lot more sense just to go with M.2. You don't need the extra space that U.2 offers, so it's kind of waste of money unless you need that capacity, but if you do, it's actually not that bad. It's just a niche product that most people aren't willing to spend that kind of money on.

Edit 2: Also, if you did just use M.2 drives instead, how many PCIe lanes would you need to accomplish the same thing? That's a consideration as well because it's not like systems with a lot of PCIe lanes are cheap either.

Making SATA better for SSDs is just doubling-down on the mistake.

That sums this up nicely.

 

[lexluthermiester]

What's wrong with U.2 beyond cost again?

Backward compatibility. There's isn't any. There is nothing that is holding back the SATA connections and cabling from supporting faster speeds. It's pure shortsightedness and that needs a change.

But SATA is a shitty protocol for SSDs.

Opinion. There is nothing, anywhere, that proves the SATA protocol can not be pushed to high speeds and bandwidth. Just like USB, DisplayPort and HDMI, the existing connectors and cabling can but pushed to higher performance with the proper engineering.

 

[Chrispy_]

Edit 2: Also, if you did just use M.2 drives instead, how many PCIe lanes would you need to accomplish the same thing? That's a consideration as well because it's not like systems with a lot of PCIe lanes are cheap either.

Given that we're talking about large capacities and therefore large amounts of data transfer, even the fastest drives like the 980Pro have sustained write speeds that are unhindered by a lowly PCIe 4.0 x1 lane, (equivalent to a PCIe 3.0 x2 interface).

With the current/upcoming CPUs throwing lots of PCIe 5.0 lanes around, there's bandwidth to spare for NVMe. Gen5 x4 is 16GB/s of bandwidth and presumably that can be divided among interfaces in a reasonably intelligent way to satisfy future bulk PCIe storage.

 

[Aquinus]

Backward compatibility. There's isn't any. There is nothing that is holding back the SATA connections and cabling from supporting faster speeds. It's pure shortsightedness and that needs a change.

You won't need backwards compatibility if you continue to support SATA like we do today for the devices that don't need all of that bandwidth, for those that do, it's honestly time to move on. There is absolutely zero good reason to take a SATA device and make it usable on U.2.

Opinion. There is nothing, anywhere, that proves the SATA protocol can not be pushed to high speeds and bandwidth. Just like USB, DisplayPort and HDMI, the existing connectors and cabling can but pushed to higher performance with the proper engineering.

No, it's not. SSDs weren't even a consideration when SATA was first developed (announced in 2000.) Everything about it is based on rotational media drives (like NCQ.) It makes far more sense to stick with SATA for these use cases and move to something else (like M.2 and U.2,) for flash storage. The only reason that SSDs were made for SATA was because that was your only feasible option that would enable widespread adoption at the time. Now we have plenty of options that are far better.

Look, I see what you're saying but not every connector is like USB where there is a path forward for improvement. If there was, we would have seen it already, particularly in the server space and even there we've seen them abandon it. Just look at the direction that SAS took.

Given that we're talking about large capacities and therefore large amounts of data transfer, even the fastest drives like the 980Pro have sustained write speeds that are unhindered by a lowly PCIe 4.0 x1 lane, (equivalent to a PCIe 3.0 x2 interface).

With the current/upcoming CPUs throwing lots of PCIe 5.0 lanes around, there's bandwidth to spare for NVMe. Gen5 x4 is 16GB/s of bandwidth and presumably that can be divided among interfaces in a reasonably intelligent way to satisfy future bulk PCIe storage.

Then we can unify around a single protocol; PCIe, which has backwards compatibility with a metric crap ton of add-in cards. Still have SATA drives? Get a SATA PCIe card. Plenty exist and a lot are dirt cheap.

 

[Deleted member 24505]

I am quite happy about the 4x Gen 4 NVME on my board, i am currently only using 3. The only SATAs i am using is one for the 2TB 2.5" SATA SSD which is not a bad drive, and one for my 2TB HDD backup drive.

Unless SATA gets a major update, it will not be long till there is no SATA and boards will come with only NVME. I would be ok with that.

 

[TheLostSwede]

Backward compatibility. There's isn't any. There is nothing that is holding back the SATA connections and cabling from supporting faster speeds. It's pure shortsightedness and that needs a change.


Opinion. There is nothing, anywhere, that proves the SATA protocol can not be pushed to high speeds and bandwidth. Just like USB, DisplayPort and HDMI, the existing connectors and cabling can but pushed to higher performance with the proper engineering.

Yes, that's your opinion, but the SATA connector and cables have clear limits. However, you refuse to listen to reason and keep harping on like you know best, so why are we even discussing this? It's a waste of time, as you have decided that this is how it is and everyone else is either wrong or stupid.

 

[Aquinus]

Yes, that's your opinion, but the SATA connector and cables have clear limits. However, you refuse to listen to reason and keep harping on like you know best, so why are we even discussing this? It's a waste of time, as you have decided that this is how it is and everyone else is either wrong or stupid.

Like the USB argument. Where are these extra pins going to go in the SATA connector that is backwards compatible with the existing pin placement? That's an absolutely massive consideration for this and for those who haven't noticed, those pins in SATA connectors are pretty close together and run the length of the connector. I'm not sure how you're going to make that one work without extending the size of the connector like with SATA Express. At least USB 2.0 had room at the back of the connector for those extra pins to get to 3.0. SATA, not so much.

 

[Valantar]

And beyond 8TB? That is HDD territory for now but it's not going to be long before affordable SSD's of larger capacities start hitting the market and the NVMe form factor is not going to get the job done, nor should it be expected to. We need an update to SATA bandwidth and we need it soon.

by that time we probably have 352-layer NAND, you should easily be able to cram like 16tb of those on an m.2 if not more

your argument is invalid.

Exactly. NAND cost and per die density are both pretty much linearly tied to production processes. The only way higher capacities will become affordable is by them becoming denser, which will allow for more packed into a single m.2 drive. There is no realistic prospect of flash prices dropping that doesn't tie directly to higher density.

Hi,
Cost of U2 is more than enough to laugh at it

That's because u.2 is exclusively used in enterprise products. If there were consumer u.2 SSDs, they wouldn't be much more expensive than m.2 SSDs of the same capacity - the controller, NAND and RAM would be the same, after all.

 

[Chrispy_]

I am quite happy about the 4x Gen 4 NVME on my board, i am currently only using 3. The only SATAs i am using is one for the 2TB 2.5" SATA SSD which is not a bad drive, and one for my 2TB HDD backup drive.

Unless SATA gets a major update, it will not be long till there is no SATA and boards will come with only NVME. I would be ok with that.

SATA's not going away any time soon. It's still the best solution for mechanical drives and that is several years away from being phased out, if at all.

SATA is perfect for mechanical storage and doesn't need an interface or protocol change.
NVMe is perfect for SSDs and doesn't need a protocol change. M.2 is a limiting interface and U.2 is the answer to those limitation but it's not mainstream in the consumer space yet.

 

[Deleted member 24505]

SATA's not going away any time soon. It's still the best solution for mechanical drives and that is several years away from being phased out, if at all.

SATA is perfect for mechanical storage and doesn't need an interface or protocol change.
NVMe is perfect for SSDs and doesn't need a protocol change. M.2 is a limiting interface and U.2 is the answer to those limitation but it's not mainstream in the consumer space yet.

If it goes or if it stays, I am happy either way. It is a users choice whether to use them or not. I still have need of them, but given a bit more cash i probably would not.

 

[Valantar]

SATA's not going away any time soon. It's still the best solution for mechanical drives and that is several years away from being phased out, if at all.

SATA is perfect for mechanical storage and doesn't need an interface or protocol change.
NVMe is perfect for SSDs and doesn't need a protocol change. M.2 is a limiting interface and U.2 is the answer to those limitation but it's not mainstream in the consumer space yet.

I doubt it ever will be, as for most people more than 1-2TB of storage just isn't necessary any more. If the current (and IMO unsustainable) cloud storage paradigm collapses we can revisit this, but I don't see consumer u.2 becoming a thing when most users are using a single m.2, or maybe two.

And, of course, thanks to the flexibility of PCIe any PCIe slot or m.2 slot can be made into u.2 easily and cheaply, and any m.2 drive can be stuck in a 2.5" u.2 caddy. If you really want offboard drives, they are available for enthusiasts willing to pay a slight premium and do a small amount of DIY assembly.

 

[Deleted member 24505]

I don't use cloud storage at all, I would much rather have a big old HDD at home to store all my stuff on.

 

[Valantar]

I don't use cloud storage at all, I would much rather have a big old HDD at home to store all my stuff on.

And that's why NASes are gaining popularity

 

[lexluthermiester]

for those that do, it's honestly time to move on.

Your opinion, and based on the poll, clearly not everyone agrees and indeed most think updating the SATA spec is good idea.

There is absolutely zero good reason to take a SATA device and make it usable on U.2.

That was YOUR suggestion, not mine. I couldn't care less about U.2 in a PC.

Yes, that's your opinion

No, it isn't. Physics says otherwise. Proof you ask? I give you DisplayPort and HDMI as examples. The bandwidth and specs have more than quadrupled since their original release without any changes to the number or shape of the connectors.

SATA's not going away any time soon. It's still the best solution for mechanical drives and that is several years away from being phased out, if at all.

True, and that's as good a reason as any to update the specs/speed/bandwidth.

 

[TheLostSwede]

NVMe is perfect for SSDs and doesn't need a protocol change. M.2 is a limiting interface and U.2 is the answer to those limitation but it's not mainstream in the consumer space yet.

M.2 isn't really limiting though, it has after all gone from PCIe 2.0 to PCIe 4.0, with increased speeds accordingly and we'll soon see PCIe 5.0 support.
If you want a wider bus, then there are already options for that in the enterprise space as well, that could potentially move down to the PC space, but I doubt that'll happen, much like U.2 never taking off in the consumer space.

 

[Valantar]

That was YOUR suggestion, not mine. I couldn't care less about U.2 in a PC.

So you don't care that a solution to your proposed problem already exists? I guess that's one approach...

No, it isn't. Physics says otherwise. Proof you ask? I give you DisplayPort and HDMI as examples. The bandwidth and specs have more than quadrupled since their original release without any changes to the number or shape of the connectors.

... And SATA has quadrupled since its launch as well, from 150MB/s to 600 theoretical. SATA also operates in a much, much noisier environment, and comes with an expectation of cable costs 1/10th of those other two.

The current situation is clear: SATA3 does what it needs to do for HDDs and older/compatible SSD designs. For anything else, m.2 fulfills those needs for consumers, and u.2 for servers, data centers and extreme enthusiasts. And as m.2, PCIe and u.2 are just different physical interfaces for the same standard they are easily and cheaply converted to each other, solving most limitations that aren't traceable back to a lack of platform lanes - a problem that would also impact any faster SATA standard.

 

[TheLostSwede]

No, it isn't. Physics says otherwise. Proof you ask? I give you DisplayPort and HDMI as examples. The bandwidth and specs have more than quadrupled since their original release without any changes to the number or shape of the connectors.

You never bothered to read my previous reply to this, did you? You just want to argue your point, because you're right, no?

DisplayPort and HDMI cables are NOT inside the case where there's EMI for starters. Both standards were designed with more robust shielding, whereas SATA cables aren't shielded in general, unless we're talking eSATA. Yes, there are a lot of ground wires, but that's not the same thing. On top of that, both DisplayPort and HDMI have pretty much reached their limits over copper now, at least for anything longer than about one metre.

If the SATA cables were good enough, as I pointed out before, why did SATA, that used IDENTICAL cables, move to a different connector and cable after hitting 12 Gbps? Could it be because the cables simply weren't up to the task? Do you really think new standards are being developed for no good reason? So yes, in theory, we can jump SATA up to 12 Gbps, but the SATA-IO decided it wasn't worth it and launched SATA Express, which flopped due to the fact that there were zero drives in the market.

 

[lexluthermiester]

You never bothered to read my previous reply to this, did you?

No, not really. I've about tuned you and Valatar out as both of you seem to be either missing the point or deliberately glossing it over.

 

[TheLostSwede]

No, not really. I've about tuned you and Valatar out as both of you seem to be either missing the point or deliberately glossing it over.

No, I get your point, but you're clearly not understating the limitation that are in place, yet claim they don't exist.
It's fine to be a dream, but then reality comes calling and it all falls down.

 

[lexluthermiester]

No, I get your point, but you're clearly not understating the limitation that are in place, yet claim they don't exist.

There are ZERO physical limitations to why SATA is currently only 6Gbps. The SATA cable conductors are physically capable of up to 52Gbps per pair of data lines and SATA has two sets in a single cable. The problem is NOT the physics of the conductors. It's that it was simply never done. That's it, end of story.

 

[WhitetailAni]

There are ZERO physical limitations to why SATA is currently only 6Gbps. The SATA cable conductors are physically capable of up to 52Gbps per pair of data lines and SATA has two sets in a single cable. The problem is NOT the physics of the conductors. It's that it was simply never done. That's it, end of story.

Why did SAS switch to a new connector for 12 Gbps and up then, if the SATA connector already worked?

 

[TheLostSwede]

There are ZERO physical limitations to why SATA is currently only 6Gbps. The SATA cable conductors are physically capable of up to 52Gbps per pair of data lines and SATA has two sets in a single cable. The problem is NOT the physics of the conductors. It's that it was simply never done. That's it, end of story.

You keep saying that and I keep giving you facts that you keep ignoring, so whatever.