[SOLVED] Ryzen 5 3600 and noctua NH-D15

[samiski]

Hi!

I got a small problem.
My first pc I ever build is great and all, but the temps are bad...

I got a ryzen 5 3600 and a noctua nh-d15 on it, yes I do know it's overkill but hey - it should cool it down right?
But it doesn't.

On prime95 I get 80-83c* and I've tried a thousand things to solve it but I can't.
Until last night I found out that when I use prime95 the cpu is on 165W!!!

Isn't that like ALOT?!
Please help me, it would be goooolden if anyone could help me solve this mess :-(!
Thank you in advance

 

[samiski]

So these are the results I got with ryzen master...

Pic nr 1 (P95 SMALL TEST) https://www.bildtagg.se/file/zk22dcdnc8cvgv89ggop0b6k
Pic nr 2 (CINEBENCH R20) https://www.bildtagg.se/file/0i7ak5l943j0gyck94zihvt

They were really different than the argus monitor for sure, but still a little high for a noctua, dont you think guys?

 

[Deleted member 2720853]

Now see, that's a more accurate power draw reading.

At this point, if you are absolutely sure that you mounted the cooler properly and that it makes contact with the surface of the CPU, and that you indeed removed the plastic on the base of the cooler before placing it on the CPU...

I would recommend you to invest in two 140mm front intakes and one good 120mm exhaust.

Choose A Case Fan

pcpartpicker.com

Pro tip, non-RGB fans always work better, less TDP on them.

Try to get airflow running through that case. Your current setup isn't the greatest in that regard, should lower temps a bit.

Other than that, I can't really recommend you anything else, someone with more experience on the matter will have to help further, seeing as your power draw on the CPU is normal.

 

[Phaaze88]

Pic nr 1 (P95 SMALL TEST) https://www.bildtagg.se/file/zk22dcdnc8cvgv89ggop0b6k

Running at 3.9ghz across all cores. TDC at 94% of 60A.

Pic nr 2 (CINEBENCH R20) https://www.bildtagg.se/file/0i7ak5l943j0gyck94zihvt

Running at 4.1ghz across all cores. TDC at 83% of 60A.

 

[Crosslhs82x2]

True on a more accurate reading.

But something dosen't seem right.
Did you at any time try use any Oc software?

 

[samiski]

@justquant

Yea, I am thinking of putting some more flow in the case, for sure.
Thanks for your help man

@Phaaze88 I am not that good at experts, are those good numbers or not?

Do you have any idea about why it's still hot?

@Crosslhs82x2 No, nothing with oc.
What doesn't seem right?

Well guys to be honest, I'm super happy with my PC... Only thing is I need to lower the temps... I mean they are not crazy high, but I don't think they are suited a noctua...

 

[Crosslhs82x2]

I guess it's the ppt tdc and edc being in red and yellow.
But the watts and amps are right.

 

[samiski]

@Crosslhs82x2 So what can I even do about these?

 

[Phaaze88]

You've got a NH-D15, and because of that, you can hit 4.1ghz all core in Cinebench R20. I'm not all that concerned with the P95 result - it's a worst case scenario, and it didn't overheat during that either.
No overheating present, and you've got a higher all core frequency thanks to that cooler.

 

[samiski]

@Phaaze88 But it performs poorly compared to other coolers when it comes to CPU temp?

 

[Crosslhs82x2]

Between the 3 different cpu's I have 2600x, 3600, and 3600x on 3 different mobo's
A asus rog strix x470 f gaming which started with a 2600x now a 3600x
A msi b450 pro carbon ac with the former2600x now a 3600
And now a asus crosshair vii x470 with the former 2600x none of them have ever showed red and yellow ppt tdc edc vales at the same time on any benchmark test or stress.

 

[samiski]

I tried playing around a little in BIOS and there was some autotuning (?) kind of button which said it would make my performance like 8% higher and something else 3% higher...

I pressed that and played one game of CS GO max settings and I got like 50-55 celcius which is really good compared to the first game i played 65-75 celcius.

So I am super happy but I tried benchmark it again and this happened with P95
https://www.bildtagg.se/file/6g8aucbtn8m3ez8eg7bokfar

and this happened with R20
https://www.bildtagg.se/file/t7wuvrfcdiqp1msc4k14avc7

I mean the R20 one is really good, but what the h*ll happened with P95?

 

[Crosslhs82x2]

Did you disable the Avx's?

 

[samiski]

@Crosslhs82x2 No, these 4 pictures I posted are on the same test, with the same settings.
But it seems like this bios autotuning button pushed down the cpu core volt to 1,1 - as u see on ryzen master or am I wrong?

 

[samiski]

FYI* I found out now it's the asus bios EZ TUNING I am talking about lol

 

[Crosslhs82x2]

The Avx's in p95 is what drove up the temp.
Try the same with them disabled.

As far as auto tuning you can but most here will advise against it.
The ppt tdc edc limits are better now and can be changed when needed to accommodate you bios settings for performance without any auto tuning feature.

 

[Phaaze88]

But it performs poorly compared to other coolers when it comes to CPU temp?

, this whole thing with Ryzen 3000 and it's 'high thermal's' has really gone over the rainbow. Some users legit have thermal issues, others are making a mountain out of a molehill.

Ryzen 3000 is unique to the cpus before it - it has more in common with a graphics card - Nvidia's 10, 16, and 20 series with their Gpu Boost Algorithm: boost frequencies are determined by the available power and thermal headroom.
The more leeway available, the faster it will try to run - at the same time, those higher speeds will make it run just as warm as it did with lesser cooling, but those temps aren't harming it though.
This behavior is also part of the reason manually OC'ing on them is so poor.


To give a more direct answer to your question though: No.
Ryzen 3600 has an advertised single core boost of 4.2ghz. It runs that easily enough on light loads, but on heavy/all thread loads, it WILL TRY to run that same 4.2ghz on all active threads depending on the available power and thermal headroom.
Because of that, looking only at thermals on these cpus isn't enough; there's a need to check the boost frequencies as well.
I think you could do with a couple more chassis fans, so you can have 2 front intake and a rear and top exhaust...

If you really want both 4.2ghz all core and some 60C(I guess?) load thermals, custom cool it - but that's bad advice for different reasons...

 

[Karadjgne]

Noctua used to post TDP ratings on their coolers, but over time decided (rightly) that TDP ratings, especially on Intel cpu's, were seriously bunk, so they quit doing that and went to a graphical representation instead.

The TDP rating of the NH-D15/S is 250w+ and has not changed since its inception.

Ryzens are not Intels. They'll pretty much stick like glue to their wattage rating, even under extreme loads, because they'll throttle back on boost in order to do so. The 3600 is a 65w cpu and that's what you can expect from it, AVX or not.

PBO can change that. It's a documented fact that some motherboard manufacturers will artificially boost PBO beyond AMD stocks settings, applying a level 1 or level 2 limit instead which gives an artificial boost to cpus for testing and performance values.

https://www.gamersnexus.net/guides/3491-explaining-precision-boost-overdrive-benchmarks-auto-oc

Asus is the only manufacturer that flat out refuses to budge on the intel/amd settings, and as a result has a reputation for poor performance at stock, but reality is the 'stock' settings for other vendors are not really 'stock' at all.

So you can run PBO, but with that cooler, won't get better performance on an Asus board since one of the parameters for boost is temps. Unless you apply a level 1 or level 2 PBO, which changes the 'stock' values.

If you look at the top guages in the first runs, they are all perfect. PPT of 88w, cpu power of 67w, TDC of 60A, EDC of 90A.

It's the second runs where PBO limits are changed, allowing for much higher outputs.

 

[dorsai]

I am using a Phanteks P400-A - one 120mm phanteks fan in the front and one in the back.

Not really good on the airflow...you should be using all 3 front fan intake positions...dumping the heat from a 2060 Super into the case is likely pulling pretty hot air through the NH-15.

 

[GarrettL]

A D-15 "not cooling" a 3600 shows something else is the issue.

How much thermal paste did you use @samiski? The D-15 is seated and tightened properly? Did you install this massive heatsink while the motherboard is mounted in the case?

You only need to apply about a pea sized amount of thermal paste in the center of the cpu then mount the cooler.

 

[Karadjgne]

With Ryzen 3000 series it's better to do the credit card smear and cover the entire surface. Intels and prior mainstream Ryzens used a monolithic chip, that's 1 chip dead center under the IHS. Ryzen 3000 uses multiple chiplets that are spread out close to the edges of the IHS for better thermal control and properties. The pea method stands a good chance of not actually covering 1 or more corners of the chiplets, unless it's a rather large pea.

Use something like Ryzen Master that uses an averaging algorithm for a singular result and it'll stand a good chance of being quite high, with any cooler. With the memory controller and cores not in the same chiplet, even using HWInfo and individual core results might not be entirely reliable

 

[Jaime Ng]

Hi bro, have you tried setting manual Vcore and CPU Clock speed? Recent 3600 CPU batch is of great quality that many report can run all core 4.2, 4.3 even 4.4 at lower than 1.2 Vcore.

Mine is currently running at 4.25 GHz, 1.175 Vcore late 2019 batch), no crashing, several stress tests are already done (no issues it seems)
I use a cheap aftermarket cooler (10$) name CR1000 and get idle around 35-40, gaming never past 55 sometime even at high 4x, P95 small ftts avx max out at high 7x.

Other notable info: ambient temp 30 (really hot here xD), case is CM MB500