I don't believe those move nearly enough heat at this stage. i could be wrong but everything I've seen implies they are more for embedded devices at this time.
NAND enjoys being hot, but the controller does not. AFAIK, the ideal spot is somewhere around 25~50c, but it's not like the controller throttles at 72c either. The heatsink solution works, it's just that 72c is a disappointing number for a monstrously large heatsink.
You don't care for efficiency but some of us does. You're also lack of knowledge and common sense. To make this short; excess heat is wasted energy and it's released INSIDE the case, affecting the temperatures of other components. I don't want to remember not to transfer large files while gaming.
They max out at 15 min mark after that none of these PCIe SSDs maintain their PCIe gen speed all the way from 3.0, 4.0 and 5.0. They all drop to 1500MB/s after 960 Seconds, go and read Toms Hardware review on the sustained performance graphs. Only good drives perform consistently.
Samsung are worst. Firmware problems, QC problems what not. 990 is also a pass. Better buy higher endurance NAND flash over the stupid perf numbers.
Currently there are a number of challenges for this. first off, it's the wrong form factor to fit on a NVMe drive and still be compliant with most motherboards. Secondly, a single unit can only dispel around 5W of heat, which likely is not enough for these drives (the single system with one uses two on a 7W CPU).
It's too early, and we have yet to see if they can scale or take other form factors.
I see that 5W number quoted a lot, but what I wonder about is if that number goes up when air expelled by an Airjet passes over a low-profile/slim form factor heatsink which would increase surface area and make use of outlet air for added cooling. I'm pretty sure the Airjet specs only account for the cooling capability of the unit itself. 2280 SSDs are long enough to add hypothetical, supplemental heatsinks.
It’ll be interesting to see results with 2400 MT/s chips soon. Even better will be 3200 MT/s using Kioxia/WDC’s BiCS8 process. Might be some low cost but speedy 4-channel designs.
Heat seems to be more of an issue with SSD controllers. Perhaps moving to smaller process nodes would help.
From what I've seen, handled, and read, the problem with SSD controller heat has more to do with it sitting low on the SSD. On Gen4/5 NVMe SSDs, the NAND chips get placed at the outer edges, and this causes the controller to have a thicker and less efficient heat pad.
IMO, if the manufacturers bothered to slap on a custom vapor chamber, that would improve cooling with generic flat bottom heatsinks.
In fact, Sabrent and some other makers are working on even faster SSDs with around 14 GB/s read speeds, so it is well possible that we are going to see a yet another 'sub-generation' PCIe Gen5 drives that will saturate a PCIe 4.0 x4 interface.
That should be PCIe 5.0 x 4 interface at the end of this statement?
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
15 Comments
Back to Article
meacupla - Wednesday, November 15, 2023 - link
I was hoping to see an Airjet on one of these, rather than... well whatever those are.Reflex - Wednesday, November 15, 2023 - link
I don't believe those move nearly enough heat at this stage. i could be wrong but everything I've seen implies they are more for embedded devices at this time.meacupla - Wednesday, November 15, 2023 - link
They do 15W of cooling. How much power do these Gen5 drives consume?LordanSS - Wednesday, November 15, 2023 - link
Overclock3D.net did a review on these.On their temperature testing, with the fan on, 72C temperatures. And it's not even the hottest NVMe drive.
I know it's used in bursts for read/write but these things are getting way too hot.
TheinsanegamerN - Thursday, November 16, 2023 - link
But these things are maintaining boost speeds for extended periods of time with these heatsinks and people are whining about how theyre too hot.meacupla - Saturday, November 18, 2023 - link
NAND enjoys being hot, but the controller does not.AFAIK, the ideal spot is somewhere around 25~50c, but it's not like the controller throttles at 72c either. The heatsink solution works, it's just that 72c is a disappointing number for a monstrously large heatsink.
sonny73n - Sunday, November 19, 2023 - link
You don't care for efficiency but some of us does. You're also lack of knowledge and common sense. To make this short; excess heat is wasted energy and it's released INSIDE the case, affecting the temperatures of other components. I don't want to remember not to transfer large files while gaming.Makaveli - Monday, November 20, 2023 - link
Do you normally do large file transfers in the middle of your gaming session?Silver5urfer - Monday, November 20, 2023 - link
They max out at 15 min mark after that none of these PCIe SSDs maintain their PCIe gen speed all the way from 3.0, 4.0 and 5.0. They all drop to 1500MB/s after 960 Seconds, go and read Toms Hardware review on the sustained performance graphs. Only good drives perform consistently.Samsung are worst. Firmware problems, QC problems what not. 990 is also a pass. Better buy higher endurance NAND flash over the stupid perf numbers.
Reflex - Thursday, November 16, 2023 - link
Currently there are a number of challenges for this. first off, it's the wrong form factor to fit on a NVMe drive and still be compliant with most motherboards. Secondly, a single unit can only dispel around 5W of heat, which likely is not enough for these drives (the single system with one uses two on a 7W CPU).It's too early, and we have yet to see if they can scale or take other form factors.
PeachNCream - Thursday, November 16, 2023 - link
I see that 5W number quoted a lot, but what I wonder about is if that number goes up when air expelled by an Airjet passes over a low-profile/slim form factor heatsink which would increase surface area and make use of outlet air for added cooling. I'm pretty sure the Airjet specs only account for the cooling capability of the unit itself. 2280 SSDs are long enough to add hypothetical, supplemental heatsinks.Jansen - Thursday, November 16, 2023 - link
It’ll be interesting to see results with 2400 MT/s chips soon. Even better will be 3200 MT/s using Kioxia/WDC’s BiCS8 process. Might be some low cost but speedy 4-channel designs.Heat seems to be more of an issue with SSD controllers. Perhaps moving to smaller process nodes would help.
meacupla - Thursday, November 16, 2023 - link
From what I've seen, handled, and read, the problem with SSD controller heat has more to do with it sitting low on the SSD. On Gen4/5 NVMe SSDs, the NAND chips get placed at the outer edges, and this causes the controller to have a thicker and less efficient heat pad.IMO, if the manufacturers bothered to slap on a custom vapor chamber, that would improve cooling with generic flat bottom heatsinks.
Threska - Sunday, November 19, 2023 - link
Or motherboards have an integrated heatpipe SSD heatsinkMakaveli - Monday, November 20, 2023 - link
In fact, Sabrent and some other makers are working on even faster SSDs with around 14 GB/s read speeds, so it is well possible that we are going to see a yet another 'sub-generation' PCIe Gen5 drives that will saturate a PCIe 4.0 x4 interface.That should be PCIe 5.0 x 4 interface at the end of this statement?