[SOLVED] Anything else I can do to cool my 970 Pro 1TB?

Droidfreak

Reputable
May 30, 2020
27
3
4,545
www.linkedin.com
Hi guys,

I think my Samsung 970 Pro NVMe drive is running a bit too hot (note: speaking about drive temps I mean the "Drive Temperature 2" as reported by HWInfo64 which is widely assumed to be the controller temperature) and I'd like to ask for some advice.

The main components is question: ROG Strix Helios case, ROG Maximus Hero XI WiFi mobo and the aforementioned NVMe drive. I also tried to use the EKWB M.2 heatsink, but my results show that the Asus' default heatsink that was included with the mobo performs slightly better under heavy loads, otherwise there's barely any difference:

Activity
Asus heatsink
EKWB heatsink
Desktop idle
49 °C / 120 °F​
50 °C / 122 °F​
Browsing, light office work
51-55 °C / 124-131 °F​
51-55 °C / 124-131 °F​
Gaming, working with Android Studio
59-63 °C / 138-145 °F​
60-63 °C / 140-145 °F​
Bench (UserBenchmark, PassMark, CrystalDiskMark)
68-69 °C / 154-156 °F​
73-74 °C / 163-165 °F​


Now you may say "it doesn't hit temps above 70°C / 158°F , what's the problem?". Well, from what I read 70 deg is where thermal throttling starts and most folks on the Internet report temps that are about 10°C below mine, so I'm kinda uncomfortable with the fact my drive sits less than 10 degrees C below the thermal throttling. I also use Rog RAM Cache III with 4GB of RAM so my disk activity is at 0% most of the time (I use a desktop widget to keep an eye on it). In other words: I'd like to try to reduce those temps.

My guess is the problem is a result of insufficient airflow because the DIMMs kinda block the air coming from the front intake fans, here's an illustration:

q6A47C6WBlc.jpg



I also tried to move the drive into the bottom M.2 slot, however this proved to be a bad idea since in that case it sits directly underneath a fat GPU heatpipe so the temps were even worse.

I'm not aware that overclock can affect M.2 temps, but for the sake of completeness: my i9 9900KF runs at 5.1 Ghz all-core @ 1.35 V and my G. Skill F4-3600C16D-32GTRG RAM is overvolted to 1.45V (the only way I could get it stable at timings loaded from a preset: 13-14-14-350). CPU, DIMM, VRM and other temps are absolutely fine, no issues observed here.

Since the side panel is 100% tempered glass I cannot install an additional intake fan here to improve the airflow. I was thinking about some M.2 active cooler, but the conclusion from my web research was that M.2 active coolers are either useless or extremely loud with the small fans going to 7000 RPM and beyond. Any ideas what else I can do in this particular setup? Thanks.
 
Solution
I just checked my m.2 ssd temperatures using HWMonitor.
It shows 35c. at idle.
I just did a virus scan which ran for only 34 seconds. The max ssd temperature went up to 38c.
I think it takes much longer for a ssd to heat up enough to cause throttling.
Probably 2 minutes or more of continuous sequential operations.

Nothing ever measures temperatures directly.
Electrical readings are taken, then an app calculates what the temperature is.
If either the readings or the app are wrong, you will not get a good temperature.

Droidfreak

Reputable
May 30, 2020
27
3
4,545
www.linkedin.com
The temps look normal
So 60°C during gaming with 0 - 5% disk usage is normal for this drive?

If your worried either add a fan to directly blow on the M.2 or a heatsink
That's the point: the case doesn't allow to add any additional fans (see picture). The board has an included flat heatsink (look just underneath the CPU) and the EKWB heatsink that I tried looks similar to the low-profile one you attached. The Sabrent looks promising because of its higher profile and copper but I can't find it anywhere on the EU market...
 

delaro

Judicious
Ambassador
Dumb question but did you remove the label/sticker off the M.2 before applying the Thermal pads? It's a pretty common mistake. Any of those heatsinks, even the EKWB should drop the thermals around 7°C as a worst-case scenario underload.
 
Last edited:
  • Like
Reactions: Droidfreak

Droidfreak

Reputable
May 30, 2020
27
3
4,545
www.linkedin.com
Dumb question but did you remove the label/sticker off the M.2 before applying the Thermal pads? It's a pretty common mistake. Any of those heatsinks, even the EKWB should drop the thermals around 7°C as a worst-case scenario underload.

I have removed the front sticker which is plain paper, but didn't touch the one on the back because looking from the side I can see this is copper.

That is his AIO. :whistle:

Exactly, the fans belong to the Strix LC 360 RGB cooler and it wouldn't be a a good idea to unplug them :D

Blush!!
I was looking at the rear as the aio.
I was wondering why a 9900K had such a puny aio.
How about using a slot fan like this in the first slot.
https://www.newegg.com/p/1YF-00B0-00DU1?Description=pci slot fan&cm_re=pci_slot_fan--9SIAAESA6P2786--Product&quicklink=true
It will direct some cooling air over the m.2 device.

Thanks for the hint, didn't know such fans even exist, I will look into it (y)

@Droidfreak
I think it would be in your best interest to use Samsung Magician software instead of hwinfo for your Samsung drive.
The proprietary software would be more reliable in this case.

Thanks, I will try it, though I think a thermal sensor is always just.. a thermal sensor?
UPDATE it seems this software only displays the "drive temperature 1" simply as "Temperature" and I'm concerned about the other one.
 
Last edited:

Phaaze88

Titan
Ambassador
Thanks, I will try it, though I think a thermal sensor is always just.. a thermal sensor?
Yes, but there's 2 drive temperatures: the controller and the V-NAND flash.
If I recall correctly, the controller on these drives are DESIGNED to run warm - the exact opposite of the controllers on other storage drives. It's called the Phoenix Controller after all...
Sure, they can perform worse if they get too hot, but these controllers can perform worse if they get too cool as well! That part had caught me by surprise, that's why I was able to recall it from memory.
Now if I could just remember where I found that, dagnabbit...
 
I just checked my m.2 ssd temperatures using HWMonitor.
It shows 35c. at idle.
I just did a virus scan which ran for only 34 seconds. The max ssd temperature went up to 38c.
I think it takes much longer for a ssd to heat up enough to cause throttling.
Probably 2 minutes or more of continuous sequential operations.

Nothing ever measures temperatures directly.
Electrical readings are taken, then an app calculates what the temperature is.
If either the readings or the app are wrong, you will not get a good temperature.
 
Solution