Recent SSD's and heat problems

[CakeMonster]

Why is there suddenly an SSD heat problem now, when we are constantly shrinking the process size the chips are manufactured on, and when we've had SSD for several years without it being a problem earlier? And is it likely to be fixed anytime soon?

 

[redzo]

http://www.tweaktown.com/reviews/6372/samsung-xp941-512gb-m-2-pcie-ssd-review/index10.html
The controller seems to be the main heat contributor with these things. I guess that there is a lot of room for improvement regarding this.

They(manufacturers)'ve probably abandoned the idea of fitting them with a stock heat sink since one of their main advantage is space constrained application use.

 

[larryccf]

http://www.tweaktown.com/reviews/6372/samsung-xp941-512gb-m-2-pcie-ssd-review/index10.html
The controller seems to be the main heat contributor with these things. I guess that there is a lot of room for improvement regarding this.

They(manufacturers)'ve probably abandoned the idea of fitting them with a stock heat sink since one of their main advantage is space constrained application use.

yeah, but why did samsung cover the damn controller with a plastic label which is an insulator? insert sound of me thunking head on desk

 

[.vodka]

Well, when you get the SSD inside an enclosure like a classic SATA one, you can use it as a heatsink. Many reviews have cracked open some of these and found thermal pads between the controller and the chassis. Still, SATA3 is a huge limit to SSD speeds and it's not difficult to see that the controller probably wasn't working as hard as it is under what M.2 enables. (for example, ~500MB/s seq R/W vs ~2GB/s seq R/W, and then you have AHCI vs NVMe where there's far less bottlenecking going on with the new standard)

Suddenly, moving to M.2 or the like removes the possibility of using something to cool the controller down, and we get problems like these. They could solve the problem using heatsinks, maybe a small one that won't interfere with cards in the PCIe slots could get rid of the throttling.

See what I mean? What's that, a few mm of clearance for a heatsink, or just remove any possibility of using a long card for that PCIe slot? It's difficult. Of course an M.2 SSD mounted on a PCIe adapter card won't have that problem and could use a huge heatsink.

 

[KingFatty]

The problem with heat is that is reduces the longevity of the NAND?

 

[bradly1101]

The problem with heat is that is reduces the longevity of the NAND?

No, the NAND degrades with writes, the controller is where all the heat is. There are M.2 interface drives in the standard 2.5" form factor with heat protection. That's the only thing I'd put in an M.2 slot if I had one.

 

[Asphodelus]

See what I mean? What's that, a few mm of clearance for a heatsink, or just remove any possibility of using a long card for that PCIe slot? It's difficult. Of course an M.2 SSD mounted on a PCIe adapter card won't have that problem and could use a huge heatsink.

I don't get why desktop M.2 slots have the card positioned flat against the mobo instead of perpendicular to the board like how PCIe slots are. Not only would it save a ton of board space, but because a standard PCIe card is ~90mm wide anyway anything up to M.2 2280 would still fit inside just about any ATX or mATX case.

As for PCIe mounted SSDs, forget a huge heatsink, it'd be better to slap on a fan like those on lower-end single slot GPUs.

 

[VirtualLarry]

No, the NAND degrades with writes

No, NAND degrades with heat too. I've had that experience, with both a 240GB OCZ Vertex Plus R2 refurb, as well as a brand-new 30GB OCZ Agility drive, in a Foxconn NanoPC with a C-70 APU, which is passively cooled, and gets very hot inside the case, where the SSD is mounted.

Not writing the NAND, but the previously static stored data in NAND degrades with heat.

So I would indeed be concerned if my SSD was heating up excessively.

 

[Elixer]

Yeah, temps still play a role in the NAND's lifespan.

I always wondered why they don't stick RAM type heatsinks on those...

 

[Zodiark1593]

They could have a desktop model with some sort of heatsink of even a slimline active cooling setup.

 

[Seba]

I don't get why desktop M.2 slots have the card positioned flat against the mobo instead of perpendicular to the board like how PCIe slots are.

Maybe because it is too fragile and in that position it will be exposed and easily broken by accident, maybe even damaging the slot/motherboard (when you install other cards). PCI and PCI-E cards have that metal bracket, which is secured to the case and also the slot/card is wider/longer.

 

[aviator79]

The M.2 form factor doesn't allow any extension, e.g. a heatsink. An PCIe adapter card is one of the few situations where you have room for a heatsink.
If you try to put this M.2 in a modern ultrabook, you will instantly see, why there is no room for any extra stuff.

 

[zir_blazer]

The M.2 factor sucks for heavy usage precisely because cooling is nonexistant. It is only good for Mobile since with the typical bursty usage scenarios you're not supposed to get get to throttling, but I wouldn't trust them for intensive sustained use, nor long term reliability.

 

[ShintaiDK]

Why is there suddenly an SSD heat problem now, when we are constantly shrinking the process size the chips are manufactured on, and when we've had SSD for several years without it being a problem earlier? And is it likely to be fixed anytime soon?

By heat problem I assume you mean Samsung and refer to the AHCI/NVME x4 with 2150MB/sec.

Samsung just went a bit too far a bit too fast. And perhaps the controller could have used a node shrink. Who knows what its manufactured with.

NAND doesn't have a heat issue.