Review We tested 30 m.2 SSD heatsinks to find the top performer: ID-Cooling Zero M05 and M15 Review

[Admin]

Storage cooling used to be an afterthought, but PCIe 5 SSDs can run hot.

We tested 30 m.2 SSD heatsinks to find the top performer: ID-Cooling Zero M05 and M15 Review : Read more

 

[PlutoDelic]

Around a year ago, TeamGroup was mocking around with that AIO 120mm for SSD's. While niche in itself, i still think it's a wonderful idea to have that AIO there instead of as Exhaust fan.

The moment one comes around with a 140mm rad, i'm jumping ship to PCIe Gen5. But with the current stand, PCIe Gen4 is more than enough.

 

[HideOut]

We tested 30 m.2 SSD heatsinks to find the top performer: ID-Cooling Zero M05 and M15 Review​

We’ve tested 27 M.2 NVMe heatsinks to see how well they perform and what is best for your SSD’s peak performance.​

So which is it, 30 or 27?

 

[Notton]

Does the Teamgroup Z540 have an even top level between its NAND, DDR, and controller?
Some people recommend using two or more pads when the heights are uneven.

I used thick backside thermal pads on dual-sided SSDs, because there's usually a void with sparse SMD resistors under the controller and RAM. I figure the heatsink would need some pressure applied to the backside of the PCB, and not the SMDs to make proper contact with the controller.

 

[Albert.Thomas]

Around a year ago, TeamGroup was mocking around with that AIO 120mm for SSD's. While niche in itself, i still think it's a wonderful idea to have that AIO there instead of as Exhaust fan.

I should have that one for testing soon!

We tested 30 m.2 SSD heatsinks to find the top performer: ID-Cooling Zero M05 and M15 Review​

We’ve tested 27 M.2 NVMe heatsinks to see how well they perform and what is best for your SSD’s peak performance.​

So which is it, 30 or 27?

😳

 

[thestryker]

Glad to see your M.2 heatsink testing land here at Tom's and with some additions!

I wish the M05 had been out when I was getting mine as the best thing in the size was Thermalright's TR, but those don't really have fins. Fortunately they're good enough, but these would certainly have been a better choice.

The Team Group AIO seems like extreme overkill so I look forward to the ridiculousness.

 

[Amdlova]

You can place thermal pads on the back touching the motherboard droop the temps a lot

 

[Albert.Thomas]

I've realized from some of the comments on other places that I need to explain my testing methodology in a bit more detail.

1) Testing is comprised of 3x 30 minute runs. The first test is a burn in test. The 2nd test is the primary testing run, and the third test is for verification. Thus far, the results of tests 2 and 3 have always been the same. However, if that ever occurs another round of testing will happen.

2) I use a 360mm AIO during testing, as the airflow of a CPU cooler will impact results. However, there's a good argument to made that an air cooler should be used with these lower end heatsinks. In the future I plan to test a few of them with an air cooler to investigate how much it improves their performance.

 

[mac_angel]

umm, not sure if I missed it. A lot of motherboards have their own heatsinks for m.2 drives now. While they aren't going to be as great as the M15, how do they measure up to nothing at all?

 

[Albert.Thomas]

umm, not sure if I missed it. A lot of motherboards have their own heatsinks for m.2 drives now. While they aren't going to be as great as the M15, how do they measure up to nothing at all?

You bring up a good point. For the next review I submit of this type, which will cover TeamGroup's Airflow I SSD Cooler, I'll be sure to test the default heatsink my motherboard includes.

 

[thestryker]

While it'll be interesting to see what the motherboard heatsink does they vary so much depending on motherboard model. I'd imagine most motherboard heatsinks will be around the same as low-mid range third party. I think the third party ones are mostly for additional slots that don't come with heatsinks or you want PCIe 5.0 peak performance. My last motherboard came with a heatsink for the primary M.2 so at minimum I had to buy two more for the other two slots, and then I have 2 cards with 2 M.2 each so another 4.

 

[bit_user]

Thanks for this!

I stumbled into a weird gap in the market, last year. I took advantage of the SSD price crash and snapped up a Samsung PM9A3 M.2 drive, which seems to be one of the last M.2 form factor server-grade drives on the market. In fact, even Samsung seems to have discontinued the M.2 version, which is no longer listed on the product page.

PM9A3 | Data center SSD | Samsung Semiconductor Global

PM9A3 with PCIe® Gen4 features fast speeds and supports E1.S, U.2, and M.2 form factors. View the new generation of performance for powerful data centers.

semiconductor.samsung.com

Anyway, it's a double-sided 110 mm drive I wanted to mount in a motherboard with a 110 mm slot and no integrated M.2 heatsink. Worse yet, that drive runs a bit hot. This is a server board (AM4 socket), but I don't have (nor do I want) a high-airflow server case for it. So, I sought a double-sided heatsink that would provide some cooling across the entire 110 mm area. The only cooler I found was on ebay, direct shipped from China:

JEYI 22110 SSD Heatsink M.2 NVME Radiator Aluminum PCIE SSD Efficient Cooler (item #155880372217)
​

As I hoped, it does indeed contact all the chips on both sides. The bottom being rather thin, I do wish it were made of copper, but I'm sure it's better than nothing and I don't use it for anything data-intensive, anyhow. It's admittedly overkill for an OS drive, but I'm hoping it proves to be reliable.

I did see a couple other 110 mm heatsinks on the market. However, upon close examination, they didn't have full coverage of the backside (I'd guess for compatibility with 80 mm drives). If anyone is aware of other 110 mm options with full coverage on both sides, please let us know!

 

[Sleepy_Hollowed]

Sounds like the older Alienware cooling tunnels might be a good idea for many builds.

Some older Alienware desktops had a plastic tunnel for the intake air to traverse from the case fans to the video cards, and out the exhaust fans with a guided path.

Something like that could be done for these, though for home use SSDs that fast are a bit of overkill unless you're doing something special with them, in my opinion.

I'd rather have an option to throttle down the speed of these drives and get away with passive cooling instead, if possible.

 

[Albert.Thomas]

While it'll be interesting to see what the motherboard heatsink does they vary so much depending on motherboard model. I'd imagine most motherboard heatsinks will be around the same as low-mid range third party. I think the third party ones are mostly for additional slots that don't come with heatsinks or you want PCIe 5.0 peak performance. My last motherboard came with a heatsink for the primary M.2 so at minimum I had to buy two more for the other two slots, and then I have 2 cards with 2 M.2 each so another 4.

So I just finished up with this testing, and it's gonna make me rewrite parts of my testing methodology sections.... so it won't be in the next review

When I tested it, it technically passed testing in the sense that the average temperature was 78C.

However, that does not tell the whole story! The performance had horrible framepacing, so to speak, because whenever the bottom side NAND was accessed performance dropped to 3000-3200 IOPs.

So if you're using a single sided SSD, a built in heatsink should be OK if all your looking for is unthrottled performance. But if you want something that will keep your SSD properly cooled, you'll want a real heatsink.

 

[bit_user]

Sounds like the older Alienware cooling tunnels might be a good idea for many builds.

Some older Alienware desktops had a plastic tunnel for the intake air to traverse from the case fans to the video cards, and out the exhaust fans with a guided path.

Dell has been doing this sort of thing as far back as 2005, when my employer got me an Optiplex Pentium D with at least a 140 mm front intake fan being funneled into a tower heatsink atop the CPU. Since then, any Dell Precision workstation or PowerEdge server I've opened up had shrouds guiding airflow to the CPUs and memory. In the case of servers, they sell a (required) airflow kit for the GPUs, also.

I'd rather have an option to throttle down the speed of these drives and get away with passive cooling instead, if possible.

Well, they self-throttle, if they get too hot.

 

[bit_user]

... it is often a requirement to prevent throttling or even crashing in some instances.

@Albert.Thomas , you're right to cite this, but you should probably also mention that this was considered a bug and had been fixed by updated firmware from Phison (all of which is mentioned in the article you linked). People shouldn't normally be concerned about crashes from overheated SSDs (or blame the crashes they do experience on it), although I would always encourage incorporating a safety margin in your cooling setup.

every expert I’ve spoken to in the storage industry agrees that it is best practice to minimize temperature variations for the health and longevity of your SSD.

LOL, just run it at a consistent 80 C!
: D

At only $5 and $10 USD, both the M05 and M15 SSD heatsinks from ID-Cooling provide good value.

The article currently links to the M05 shipped & sold by Amazon Japan, with a price of $19.30. I didn't check how much extra shipping would cost.

I found the M15 (only) at Newegg, available for $10 (+ $10 shipping):

https://www.newegg.com/p/13C-004P-000U4?Item=9SIA4REJZ35401
​

Both are listed on AliExpress.

Jeyi FinsCold Q150

LOL, you do not want the fins to be cold, if the drive is not! And a drive temp of 78 C definitely isn't cold! That could be a sign of a bottleneck in drawing heat out of the SSD and into the fins!
: D

P.S. I wonder if you might consider testing the effect of placing a bit of some decent thermal paste between the overlapping halves of a couple 2-sided heatsinks. A lot of us have a couple old tubes of the stuff lying around (I've got some dating all the way back to when Artic Silver 5 was the leading paste), in which case it seems like it could be a virtually "free" performance upgrade worth a couple degrees.

 

[thestryker]

So I just finished up with this testing, and it's gonna make me rewrite parts of my testing methodology sections.... so it won't be in the next review

When I tested it, it technically passed testing in the sense that the average temperature was 78C.

However, that does not tell the whole story! The performance had horrible framepacing, so to speak, because whenever the bottom side NAND was accessed performance dropped to 3000-3200 IOPs.

So if you're using a single sided SSD, a built in heatsink should be OK if all your looking for is unthrottled performance. But if you want something that will keep your SSD properly cooled, you'll want a real heatsink.

That's super interesting, so if the motherboard heatsink is just a top plate there's still value in getting a third party heatsink.

 

[bit_user]

Thermalright's TR, but those don't really have fins.

Yes, I've seen these and others that seem only to add a "thermal mass", without markedly increasing surface area. It will help with short, bursty workloads, but in general seems pretty nuts.

That's super interesting, so if the motherboard heatsink is just a top plate there's still value in getting a third party heatsink.

The part I found interesting was that the NAND chips are (to some extent) mapped linearly, rather than all being interleaved. At least, that's what I assume based on the premise that it's even possible for a typical workload not to hit the chips on the bottom side.

 

[Albert.Thomas]

@Albert.Thomas , you're right to cite this, but you should probably also mention that this was considered a bug and had been fixed by updated firmware from Phison (all of which is mentioned in the article you linked). People shouldn't normally be concerned about crashes from overheated SSDs (or blame the crashes they do experience on it), although I would always encourage incorporating a safety margin in your cooling setup.

You should have seen the original draft of this article 😎

Certain articles from certain manufacturers have been removed from publication as a result of the research I did for this review

 

[thestryker]

Yes, I've seen these and others that seem only to add a "thermal mass", without markedly increasing surface area. It will help with short, bursty workloads, but in general seems pretty nuts.

The Thermalright M.2 2280 Pro, which is the one I wanted, was expensive at the time or I'd have passed on the TR due to the design. Since I didn't need it for sustained load I figured these would be fine and they did drop bare drive temps by around 10C.

 

[8086]

If you could, run these tests again using a Crucial T705 because it is two things the Team Group is not: Fastest SSD around and hottest thermals of any SSD. Then let us know if the heatsinks have any actual effect on performance.

 

[PlutoDelic]

I should have that one for testing soon!

😳

Oh nice. I've been trying to find out if they ever released but nothing's showing up.

PS, please try finding out if they have plans for a 140mm version.

PS, i agree with the post above, T705 would be a fine specimen for getting the most accurate results.

 

[Albert.Thomas]

If you could, run these tests again using a Crucial T705 because it is two things the Team Group is not: Fastest SSD around and hottest thermals of any SSD. Then let us know if the heatsinks have any actual effect on performance.

These tests are actually handicapped a bit.

When I originally starting testing these things and first experimented with different ways, a lot of the feedback I was getting was telling me that folks didn't see the point of testing - they thought it "only matters for PCI-e 5."

Because this was an extremely common reaction, all of these tests you see are actually performed on a PCI-e 4 motherboard. I tested a few of these on a PCI-e 5 system to establish the "PCI-e 5 Throttling" point I list on the graphs, but I only tested 5 of them (I observed a 5-7C increase in temperature vs PCI-e 4, depending on how strong the heatsink was)

So even if the Crucial T705 has the potential to run hotter, I'm not sure I'd be able to observe any difference using the testing setup I'm using now.

If it turns out these SSD heatsink reviews are successful and something the community enjoys, I can look at potentially migrating these tests to only a PCI-e 5 system in the future - but right now we need to see if they're even something folks are interested in.

 

[t3t4]

That thing is ridiculous! So now you would have to choose between running a GPU or an SSD cooler? I'll tell you right now they both won't simultaneously fit in my system! I use my motherboards built in heat sink on all my SSD's, and not a one of them even threaten to throttle. Now granted, I'm running gen 4 drives, but I expect my inbuilt cooling to be fine for gen 5 as well. I have:

2 Samsung 990 PRO 2TB.
1 Solidigm P44 Pro 2TB
2 WD_BLACK 1TB SN850X

All that under mobo cooling and the hottest of the bunch is the Solidigm. This is the board I'm using: MSI MPG Z790 Carbon WIFI

P.S. Oh, and there is a 4090 stuck right on top of the heat sink for most of these drives!

 

[thestryker]

I use my motherboards built in heat sink on all my SSD's, and not a one of them even threaten to throttle. Now granted, I'm running gen 4 drives, but I expect my inbuilt cooling to be fine for gen 5 as well.

That motherboard has giant M.2 heatsinks which are also double sided. If you needed additional cooling I'd be really concerned as it's a good cooling solution. Many motherboards do not have something anywhere near that good.