Question Disappointing FX 9590 temps w/ Noctua NH-D15

Toggle sidebar Toggle sidebar
D

discourse4

Jun 6, 2019
34
2
35
If any of you wants the genuine fx9590 experience, do not use the NH-D15. I somehow cant get it over 45c even with Prime95 running for 26 hours. Granted, it was outside of a case.

I was looking to challenge myself to air-cool this, so the temp is beyond underwhelming. I was planning a 9x 120mm fan case set-up, but I will have to put this project on hold until I can get a worse cooler.

It didn't even crash once.
 
Sort by date Sort by votes
discourse4 said:
If any of you wants the genuine fx9590 experience, do not use the NH-D15. I somehow cant get it over 45c even with Prime95 running for 26 hours. Granted, it was outside of a case.

I was looking to challenge myself to air-cool this, so the temp is beyond underwhelming. I was planning a 9x 120mm fan case set-up, but I will have to put this project on hold until I can get a worse cooler.

It didn't even crash once.
Click to expand...
Suggests the cpu could be throttling which isn’t always due to cpu temperature. What motherboard are you using?
 
Upvote 0 Downvote
Since there is a genuine possible issue that it may be unintentionally underclocked (BIOS setting, heat, or perhaps a defective cpu), I will be rebuilding this and take a closer look at the clock speeds. I will post the results & pics after I get back from work.

I used a Gigabyte 990FXA-UD3, rev. 4.0.

I also have a Gigabyte 990FX-GAMING, but I honestly do not want to damage this one by putting in a 9590, so I will stick with the UD3 again.

I will also now use the low-noise adapter for the Noctua fans, to try and improve my temps.
 
Upvote 0 Downvote
what program did you measure the temp with? The fact you're telling us its 45c tells me you were using the wrong one, because if you're accurately measuring the temps you wouldn't know what the actual temp is, only the thermal margin.

I'd bet money that CPU was throttling all night long, or running right on the edge.

Next time show us some telemetry.
 
  • Like
Reactions: Barty1884 and Deleted member 217926
Upvote 0 Downvote
After ~20 minutes

Hysmkwi.png
 
Upvote 0 Downvote
COLGeek said:
What is the point here? Also, what is the ambient temperature of the room this "beast" is in?
Click to expand...

Core Clocks looks fine. There was a possible issue that it was throttling or was underclocked by the mobo.

20 minutes is probably not enough to say that this was the "stable" temp under load, but I have to leave in about an hour and I dont want to have this running while I am away. Besides, this was what I saw during the 26 hour stress test yesterday anyway.

The room temp is normal, so it is around 20-23 C if I have to guess.
 
Upvote 0 Downvote
discourse4 said:
The room temp is normal, so it is around 20-23 C if I have to guess.
Click to expand...

Yeah, no.

nS3nx4s.png


Even with an NH-D15, your idle temps would be ~10'C above ambient. So from a ~20'C ambient, your idle temps would be in around 30'C, minimum.

So, if those temps are accurate
They're not

You're <0'C..... So, you'd be outside, open to the elements, and something else would be going wrong there.


FX chips are only accurately recorded via "thermal margin" (essentially your temperature headroom, degrees to zero/thermal shutdown. So as temps increase the margin goes down).
AMD overdrive is what you're looking for: https://www.amd.com/en/technologies/amd-overdrive

discourse4 said:
20 minutes is probably not enough to say that this was the "stable" temp under load, but I have to leave in about an hour and I dont want to have this running while I am away. Besides, this was what I saw during the 26 hour stress test yesterday anyway.
Click to expand...

With an air cooler, it pretty much is. They don't need time to heat soak the way an AIO liquid cooler does.


If you want to actually have a discussion, you need an accurate starting point. Overdrive is what you need, and an accurate report of your thermal margin.

Even with that, I'm not sure what you're looking to achieve? It's a 9590.... You're challenging yourself to keep it cool on air? Is your room noticeably hotter now? If not, you've succeeded (and have a large room with a window open). Would've been easier to challenge yourself to heat a room with it.

discourse4 said:
If any of you wants the genuine fx9590 experience,
Click to expand...

Nobody wants the 9590 experience. People like to be comfortable, their motherboards to not rapidly die.... and, you know, things not to catch fire. That was the 9590's problem.
 
  • Like
Reactions: COLGeek
Upvote 1 Downvote
Using AMD overdrive. I am getting 25.12 min & 26.62 average on the Thermal Margin for all cores.

If the 0 point is 70C, then that would be in line with the current data from coretemp and hwmonitor.

Yes, the challenge was to air cool this thing. After hearing stories for years about this CPU, I was expecting a lot more than a simple 10 minute act of plopping in a Noctua cooler. I bought 10 NF-P12s in preparation to cool this thing in a LanCool One case.
 
Upvote 0 Downvote
If you go to Logging, it will have 2 decimals. Also it looks like the AMD Stability test is better than Prime95. I am getting "better" temperatures. Minimum is now at 22.12 after 5 minutes.

Anyway, I am heading off to work. I am shutting this down for now. I will probably disassemble this again as I have another use for the Noctua Cooler.

gqpPKJM.png
 
Upvote 0 Downvote
4.7 ghz, open air, and you're not running prime 95. I'm not surprised. Put it in a case and see what happens. Also for me to believe this I'd need to see a video of it working.

And in the end all this to not even game as well as a $99 Ryzen 3 and a $50 motherboard. You can keep the 9590 Experience.
 
Upvote 0 Downvote
Air cool a Ryzen 3 for my next project? I'm not crazy, I wont disrespect the thermal output of that thing.

Anyway, I put this in the worst case I have available (X Sniper 2). I removed all the case fans as well. Prime95 running on the Maximum CPU Heat setting only gets me a thermal Margin Average of 18.1 according to the Logs, but I did hit 10 for a few seconds.

I am starting to think the CPU is defective, but Time Spy & Fire Strike shows that I am in the middle of the pack of fx9590 + rx590 benchmarks.
Edit: Passmark CPU score is on the 65% percentile of other 9590 tests.

Time to move on to my next project, which is SLI in a socket 754. I'll keep this thing though, I think this will be a good testbed for Air Coolers.
 
Last edited:
Upvote 0 Downvote
The results you're getting make you the only one on the planet. I mean AMD literally sold this thing with a 120 AIO (which really is no better or even worse than the NH-D15, but still the thought it there, and the desire to move the heat exchange as far from the chip as possible to keep the temps down). I'd tend to believe the thermal sensors aren't working right, especially because you're only seeing midpack results. Probably throttling. If your thermals were that good, it should be running killer.
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom