• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

M.2 PCIe 5.0 SSDs Set to Increase to 25 mm in Width, Might Not Fit Older Motherboards

How do you know that we can't expect 110 mm long SSDs for consumer drives in the future? Most motherboards support at least one.
Right now the only 110 mm long drives do indeed appear to be for enterprise use, but that might very well change in the future.
Because why would we need 110? We already have 8TB TLC M.2 drives in 80 factor. 110 is really only useful if you have drives with large SLC cache or MLC chips used in enterprise.
 
Wider but not shorter? I was looking forward to calling it stubby.
 
Yeah, of course they're backward compatible, but do you honestly think it's "very likely" someone will spend the greatly increased cost for a PCIe 5.0 drive just so they can see absolutely zero benefit from it? Especially when they could've bought a 3.0/4.0 drive for far less... I really do not think that's "very likely"...

You have PCIe 5.0 slots shipping on Alder Lake motherboards for the last nine months, and these AM5 mobos will have the same issues.

I think the real problem here is, how many motherboard makers are going to be thinking ahead for this not even "draft-standard?" there will b e tons of pcie5-capable mobos already shipped before they have the final standard!
 
You know, these things are backwards compatible. It is very likely people will buy them and use them with PCIE 4.0 and 3.0 platforms
I never really understood that mentality. Buy newer tech and put it in older hardware that doesnt support the spec. If you have PCIe 3.0 why buy a 5.0 drive? It will be more expensive and you cant utilize the drive as intended.
 
I never really understood that mentality. Buy newer tech and put it in older hardware that doesnt support the spec. If you have PCIe 3.0 why buy a 5.0 drive? It will be more expensive and you cant utilize the drive as intended.


When you're already paying a premium for a top 4.0, why not jump up to "future-proof" for a few dollars more (with the understanding that you'll upgrade soon, but nit TODAY)?

This makes that impossible now!

They should just stop playing footsie, and go Type-C compatibility-breakage and change the entire slot tp sodimm-syle,
 
I think the real problem here is, how many motherboard makers are going to be thinking ahead for this not even "draft-standard?"

What are you talking about? The finalized version of PCIe gen5 was released in 2019


If you have PCIe 3.0 why buy a 5.0 drive? It will be more expensive and you cant utilize the drive as intended.

In the beginning yes, they'll be a lot more expensive, but they'll soon fall to close or even price parity (just like pcie4.0 are already close or at parity with gen3) and you get better and more modern nand/controller and are still able to take advantage of higher IOPS, endurance, lower lattency, etc.
 
What are you talking about? The finalized version of PCIe gen5 was released in 2019

[
I'm talking about the new dimensions being the "not even a daft standard" - This is like running USB 3.0 over USB type-c; you knew it was designed for faster speeds, but first you needed standard 5.0 Gbps ports working

PCIe 5 has already been shipping on motherboard m.2 slots for months, but if it takes a long time for mobo makers to respond t these changers, it will mean several incompatible drives (in the future) (not to mention cutting-out all existing mobos; a real concern because older drive have a tendency of disappearing)

As long as they're breaking things with USB-C-level change , why not just change the connector to SODIMM?
 
Last edited:
Not twice, more like 5 to 10 times https://www.amazon.com/Monoprice-Cable-Locking-Latch-black/dp/B009GUX660/ (granted that one included power, but that also would be another extra thing to buy)
You can bulk buy the standard 45CM SATA cables that are usually included with motherboards for $1.80 a pack, and that usually includes a pair of cables; One straight-straight, and one straight-to-right-angle. If you want to buy 500+ packs of them, they're $1.50ea but I was just looking at orders of 1-25 units.
 
Technically we need a new PC form factor, where the M.2 drives get a dedicated place on the motherboard, but that's unlikely to happen.
This will happen when they figure out how to design a way to install video cards right side up. :p:cool:

IDK, but it seems like E-ATX, ATX might close to their saturation point of how and what fits on a motherboard. M.2 slots are already spread willy nilly on them.
 
I'm talking about the new dimensions being the "not even a daft standard" - This is like running USB 3.0 over USB type-c; you knew it was designed for faster speeds, but first you needed standard 5.0 Gbps ports working

PCIe 5 has already been shipping on motherboard m.2 slots for months, but if it takes a long time for mobo makers to respond t these changers, it will mean several incompatible drives (in the future) (not to mention cutting-out all existing mobos; a real concern because older drive have a tendency of disappearing)

As long as they're breaking things with USB-C-level change , why not just change the connector to SODIMM?
Huh, this spec was agreed upon in 2020, as mentioned in the article. It just wasn't widely communicated, but then again, it normally takes some time before new standards propagate and we're only a year and a half from when the wider standard was agree on.

This will happen when they figure out how to design a way to install video cards right side up. :p:cool:
Which side is that?
IDK, but it seems like E-ATX, ATX might close to their saturation point of how and what fits on a motherboard. M.2 slots are already spread willy nilly on them.
Time for CTX? Oh, no, that was a brand of monitors...
DTX is taken.
ETX would be confusing with E-ATX.
FTX? But that's an FTP extension...
GTX? Nope.
HTX? Taken by some extension to HyperTransport.
ITX, taken.
JTX should work.
 
Huh, this spec was agreed upon in 2020, as mentioned in the article. It just wasn't widely communicated, but then again, it normally takes some time before new standards propagate and we're only a year and a half from when the wider standard was agree on.


Which side is that?

Time for CTX? Oh, no, that was a brand of monitors...
DTX is taken.
ETX would be confusing with E-ATX.
FTX? But that's an FTP extension...
GTX? Nope.
HTX? Taken by some extension to HyperTransport.
ITX, taken.
JTX should work.
MTX? nope thats audio related.
YTX? dont ask Y?

IDK maybe even go with a number convention.
 
I never really understood that mentality. Buy newer tech and put it in older hardware that doesnt support the spec. If you have PCIe 3.0 why buy a 5.0 drive? It will be more expensive and you cant utilize the drive as intended.
Here’s a scenario… So maybe 3 years from now I need a new drive for my aging but still very functional motherboard. pcie 3 drives are no longer being manufactured and cost 3 times the price of pcie 5… would be nice if I could just slap in the flavor of the day and move on…
 
What would you suggest instead?
U.2 never took off and this is much cheaper, both in terms of connectors and the drives themselves.
Also no need for any cables to connect with.

Technically we need a new PC form factor, where the M.2 drives get a dedicated place on the motherboard, but that's unlikely to happen.
The savings for M.2 versus U.2 has purely gone to the SSD makers not consumers. It's a pretty good racket for them: they get to have singular manufacturing lines and still charge whatever they want.

While I'd rather the shift to U.2 just happen at this point the that's completely impossible so a plausible solution would be adopting the new E1.S enterprise standard. The problem with that though is how they would connect to motherboards as they couldn't sit flat the way current M.2 does. I think it would be possible to do, but much like your suggestion of dedicated place (which would work well) it just isn't going to happen because there's no benefit to manufacturers.

Personally I just hope that I'll be able to use adapters to keep my SSD off the motherboard. Right now I'm using a U.2 to M.2 2.5 adapter connected to U.2 on my motherboard. There are M.2 to U.2 cables available, so I'd look into something like that for something without native U.2 motherboard support.
 
MTX? nope thats audio related.
YTX? dont ask Y?

IDK maybe even go with a number convention.
4TX?

The savings for M.2 versus U.2 has purely gone to the SSD makers not consumers. It's a pretty good racket for them: they get to have singular manufacturing lines and still charge whatever they want.

While I'd rather the shift to U.2 just happen at this point the that's completely impossible so a plausible solution would be adopting the new E1.S enterprise standard. The problem with that though is how they would connect to motherboards as they couldn't sit flat the way current M.2 does. I think it would be possible to do, but much like your suggestion of dedicated place (which would work well) it just isn't going to happen because there's no benefit to manufacturers.

Personally I just hope that I'll be able to use adapters to keep my SSD off the motherboard. Right now I'm using a U.2 to M.2 2.5 adapter connected to U.2 on my motherboard. There are M.2 to U.2 cables available, so I'd look into something like that for something without native U.2 motherboard support.
And you know this how? Also, I guess you've never been to an SMT assembly line?

E1.S requires a compatible chassis, so I guess that's a no go, at least unless we bring back an external drive bay that could take some kind of slot for a few drives. It would obviously still require something like a U.2 cable per drive to connect them to the motherboard.

These were quite popular for a very short while. Seems like there aren't any that goes over PCIe 3.0 speeds at the moment though, so not sure if U.2 can support speeds beyond that either. U.3 seems to be for higher speed interfaces.

81+2Z8x+QRL._AC_SL1500_.jpg
 
Last edited:
And you know this how? Also, I guess you've never been to an SMT assembly line?
The expensive part of any SSD is the components not the container they're put in, but not having to maintain separate lines for a hypothetical M.2 for mobile and U.2 for desktop saves assembly money. There's no reason for them to pass this savings on to the consumer (there's no alternative formats), and one look at SATA vs NVMe shows they're maintaining higher margins on NVMe. It's entirely possible that I'm wrong here, but M.2 and U.2 share all the same expensive bits which is what I'm basing it on.

E1.S requires a compatible chassis, so I guess that's a no go, at least unless we bring back an external drive bay that could take some kind of slot for a few drives. It would obviously still require something like a U.2 cable per drive to connect them to the motherboard.

These were quite popular for a very short while. Seems like there aren't any that goes over PCIe 3.0 speeds at the moment though, so not sure if U.2 can support speeds beyond that either. U.3 seems to be for higher speed interfaces.

81+2Z8x+QRL._AC_SL1500_.jpg
Yeah E1.S would require chassis support to be compatible with every form factor which is why I'm assuming we're just stuck with M.2.

U.2 definitely supports PCIe 4.0 speeds (P5800X is the first that jumps to mind), but I don't know if any of that style adapter does as I'd imagine the market is tiny. There are PCIe 4.0 M.2 to U.2 2.5 adapters, but I haven't looked into options for connecting a motherboard M.2 to U.2 with PCIe 4.0 as I don't have an applicable setup yet. No clue what enterprise PCIe 5.0 drives will look like though since from what I understand the push is for E1.S to replace all of the current standards so who knows if U.2 will continue forward.
 
The expensive part of any SSD is the components not the container they're put in, but not having to maintain separate lines for a hypothetical M.2 for mobile and U.2 for desktop saves assembly money. There's no reason for them to pass this savings on to the consumer (there's no alternative formats), and one look at SATA vs NVMe shows they're maintaining higher margins on NVMe. It's entirely possible that I'm wrong here, but M.2 and U.2 share all the same expensive bits which is what I'm basing it on.
Well, companies are always doing this this way, but U.2 drives have larger, more expensive housings, as the heatsink simply isn't optional here, whereas most M.2 drives carry an extra cost if you want a heatsink with it. The extra cost for said heatsink is often a complete rip-off. Obviously, faster, more recent products are always going to carry a higher price, as the companies are trying to recuperate their investment, so that's not an entirely fair comparison. That said, I can pick up a 2 TB PCIe 3.0 M.2 drive for less than an equivalent SATA drive now locally, so SATA doesn't seem to have that price advantage any more.
Yeah E1.S would require chassis support to be compatible with every form factor which is why I'm assuming we're just stuck with M.2.
Stuck and stuck, but as long as there are no major changes to the PC as we know it, it seems to be what all the hardware manufacturers have agreed on for the time being, until something better comes along.
U.2 definitely supports PCIe 4.0 speeds (P5800X is the first that jumps to mind), but I don't know if any of that style adapter does as I'd imagine the market is tiny. There are PCIe 4.0 M.2 to U.2 2.5 adapters, but I haven't looked into options for connecting a motherboard M.2 to U.2 with PCIe 4.0 as I don't have an applicable setup yet. No clue what enterprise PCIe 5.0 drives will look like though since from what I understand the push is for E1.S to replace all of the current standards so who knows if U.2 will continue forward.
Yeah, not really looked into that, but U.3 seems to be one option at least. There seem to be a few different options in that market space, due to different companies promoting different standards.
 
Last edited:
Honestly they should simply change the naming to M.3 or M.25. That extra 3mm (from 22) is huge deal. All older x8/x16 expansion cards will be useless, because there is simply not enough space between M.2s; also most M.2 slots which are flat in-between PCI-Ex slots on motherboard - that x16 locking mechanism - piece of plastic will obscure the (M.3) PCB.

Kind of dumb idea to change the established standard without changing the name. Even USB consortium when it was still sane changed USB1 to USB2 just because of wiring even if plug remained the same. It'll be a mess.
 
"old motherboards"? So far, the only motherboards with PCIe 5.0 storage will be X670E/X670/B650.... I don't think it'd be too common for someone to buy a 5.0 drive in order to run it on an old mobo
You obviously haven't looked at Amazon reviews for drives of magic idiots crying about only 3300 speeds on their PCIE4 4500++ drives. People will always want to be future proof, so whilst sensible people will buy something and know it won't work at those speeds... The never-ending amount of idiots will buy it thinking it'll work either way.

A size difference would help kurb the reviews of the stupid, hell a different socket entirely would be good if it didn't mess up backwards compatibility rip.

And yes they were all Intel fanboys with their brand new system which wasn't even PCIE4 last year lul.
 
"old motherboards"? So far, the only motherboards with PCIe 5.0 storage will be X670E/X670/B650.... I don't think it'd be too common for someone to buy a 5.0 drive in order to run it on an old mobo
Like I run my 980 PRO on a PCIE 3.0 board you mean?

I gain two advantages.

Its chip wont get as hot as its spec'd to higher performance, so at same performance as 970 EVO its cooler.
I get the higher performance that still fits within 3.0 bandwidth such as the higher sustained writes and random i/o.

So M.2 is already showing its flaws if they having to increase the width. The problem when you design something to be tiny and with no headroom for scaling up the products in future. Sales of 5.0 SSD's will get crippled if this is the case as only 5.0 board/laptop owners could use them natively on boards.

Is there even a need in the consumer space for 5.0 NVME? They already really fast with limited noticeable use cases even over SATA spec.

What would you suggest instead?
U.2 never took off and this is much cheaper, both in terms of connectors and the drives themselves.
Also no need for any cables to connect with.

Technically we need a new PC form factor, where the M.2 drives get a dedicated place on the motherboard, but that's unlikely to happen.
U.2 is doing fine in the server space.

The solution is easy, but I dont think the industry likes it for whatever reason.

Either.

PCIE SSD's (gigabyte did one early on I think, so this seemed to be the original idea).
U.2
Increased width but with b/c built in in so the additional part has gaps so can still fit in a smaller width slot. I dont like this, as noticing a trend now on new boards where everything is been compromised to cram in more M.2 slots as if they think everyone wants 4+ M.2 drives. There is already a thread on here questioning why all upcoming AMD board designs only have 2/3 PCIE slots.
PCIE storage adaptors with M.2 slots on them. With burification as well this is the logical way forward combine it with my b/c width idea on the slots if they 5.0 slots. Each higher end board could come with one card included to work in second 8x/16x slot, and boards start only coming with 1 or 2 onboard M.2 slots again. This would also make lower end boards cheaper.
Consider not selling 5.0 SSDs on consumer market and as such dont compromise consumer boards for them, the need seems to be mostly epeen hype related. This would also improve the above idea so e.g. a 8 lane PCIE 5.0 slot could host either 4 PCIE 4.0 SSDs or 8 3.0 SSDs.

A suggestion would be too strong a word for my non-expert thoughts, but as to what I would have preferred seeing for desktops in an alternate universe, if not going with the 2.5” form factor for drives, then simply PCIe cards like any normal expansion. I’d presume this is ultimately the route to take anyway for some situations e.g. want a multiplicity of drives without having to settle with SATA.
This is so obvious its like staring me right in the face, yet the industry doesnt seem to want it. PCIE is i/o expansion which is what NVME is, but the consumer industry is almost treating it now as just a GPU interface.
 
Last edited:
Like I run my 980 PRO on a PCIE 3.0 board you mean?

I gain two advantages.

Its chip wont get as hot as its spec'd to higher performance, so at same performance as 970 EVO its cooler.
I get the higher performance that still fits within 3.0 bandwidth such as the higher sustained writes and random i/o.

So M.2 is already showing its flaws if they having to increase the width. The problem when you design something to be tiny and with no headroom for scaling up the products in future. Sales of 5.0 SSD's will get crippled if this is the case as only 5.0 board/laptop owners could use them natively on boards.

Is there even a need in the consumer space for 5.0 NVME? They already really fast with limited noticeable use cases even over SATA spec.


U.2 is doing fine in the server space.

The solution is easy, but I dont think the industry likes it for whatever reason.

Either.

PCIE SSD's (gigabyte did one early on I think, so this seemed to be the original idea).
U.2
Increased width but with b/c built in in so the additional part has gaps so can still fit in a smaller width slot. I dont like this, as noticing a trend now on new boards where everything is been compromised to cram in more M.2 slots as if they think everyone wants 4+ M.2 drives. There is already a thread on here questioning why all upcoming AMD board designs only have 2/3 PCIE slots.
PCIE storage adaptors with M.2 slots on them. With burification as well this is the logical way forward combine it with my b/c width idea on the slots if they 5.0 slots. Each higher end board could come with one card included to work in second 8x/16x slot, and boards start only coming with 1 or 2 onboard M.2 slots again. This would also make lower end boards cheaper.
Consider not selling 5.0 SSDs on consumer market and as such dont compromise consumer boards for them, the need seems to be mostly epeen hype related. This would also improve the above idea so e.g. a 8 lane PCIE 5.0 slot could host either 4 PCIE 4.0 SSDs or 8 3.0 SSDs.


This is so obvious its like staring me right in the face, yet the industry doesnt seem to want it. PCIE is i/o expansion which is what NVME is, but the consumer industry is almost treating it now as just a GPU interface.
Not true at all re performance etc.

Also literally no one on this planet cares about laptops, if it wasnt bad enough 3080 laptops are slower gpus than a 3070 desktop there's even more issues with different tdp on them too now making 2x 3070 or 3080 laptops performing totally different.

You don't need anything more than a pcie3 speed m.2 on a laptop so everything about sizing and laptops is a mute point.
 
Not true at all re performance etc.

Also literally no one on this planet cares about laptops, if it wasnt bad enough 3080 laptops are slower gpus than a 3070 desktop there's even more issues with different tdp on them too now making 2x 3070 or 3080 laptops performing totally different.

You don't need anything more than a pcie3 speed m.2 on a laptop so everything about sizing and laptops is a mute point.
Ironically in an earlier discussion about SSDs coming with heatsinks, a lot of people jumped in saying I was been disrespectful to the laptop market and they do matter. Sizing matters from a perspective, manufacturers are then producing products to different standards, its a contributor to cost increases and reduces compatibility, its generally considered a bad thing. Laptops or not, but even if you are willing to say the laptop market has no relevance (even though most hardware sales are via laptops) are you also saying pre 5.0 PCs are irrelevant?

Can you expand/explain your not true comment please.
 
Honestly they should simply change the naming to M.3 or M.25. That extra 3mm (from 22) is huge deal. All older x8/x16 expansion cards will be useless, because there is simply not enough space between M.2s; also most M.2 slots which are flat in-between PCI-Ex slots on motherboard - that x16 locking mechanism - piece of plastic will obscure the (M.3) PCB.

Kind of dumb idea to change the established standard without changing the name. Even USB consortium when it was still sane changed USB1 to USB2 just because of wiring even if plug remained the same. It'll be a mess.
M.3 was already taken in the server space, although I think that name changed in the end to reduce confusion. It seems to be E.1, E.2 and E.3 now instead, in L and S versions.

M.2 doesn't take its name from the card width, as there are several different widths and lengths. I was called NGFF or Next Generation Form Factor initially, but I guess that was considered too long. As you can see, there are 16.5, 22 and 30 mm wide M.2 form factors, although only the 22 mm wide were adopted for storage. Then there are solder down modules as well, mainly for WiFi and Bluetooth modules.

And in case you missed it, there are four different keys for the slots, A/E for WiFi/Bluetooth and other "low-speed" add-in cards and B/M for SSDs. On top of that, there's support for SATA AHCI drives, PCIe AHCI drives and PCIe NVMe drives. I have a feeling this slight increase in width isn't going to matter too much at the end of the day, at least not as long as 2280 drives are the standard in consumer applications.

m2-dimensions.png
 
Ok seems I misunderstood things, so its just the PCB getting wider not the actual connector?
 
Back
Top