New PC with an NH-D15 and a 3600XT on open air - 85 degrees at 95 watts?

[Overtime]

I just built a new PC with an NH-D15 (both 140mm fans attached) and a 3600XT. It reaches 95 watts through AMD's "precision boost overdrive" on Ryzen Master, but temperatures reach around 81-85 degrees, even with both fans on absolutely max speed. Manual 1.33v OC at 4.5GHz hits 100 degrees, I couldn't check the watts used in time. It might also help to note that the temperature difference between fans at 100% and fans at 20% is literally nothing, the heatsink fins are barely warm, and the idle temperature is 40 degrees.

 

I am using AIDA64 Extreme Edition, Cinebench is about 8 degrees cooler for some reason. I've tried taking the NH-D15 off the CPU and screwing it back in 4 times, and I tried everything from very little to a lot of thermal paste (NT-H1). Case side panels are both off, air is relatively cold in the room.

 

The "TDP" of the NH-D15 is 220w, and I bought it due to the fact I thought it was better than a 260mm AIO, and thought I could have a ton of overclocking headroom for this 95 watt CPU. Are these temperatures and behaviors normal, or could I have done something wrong? My old 2600x at 130watts kept similar temperatures on my 212 evo, so I am very confused.

[Jurrunio]

Bad paste application or low mounting preesure if you ask me, tho Zen 2 runs much hotter than Zen+ at the same power draw due to heat density.

[Overtime]

1 minute ago, Jurrunio said:

Bad paste application or low mounting preesure if you ask me, tho Zen 2 runs much hotter than Zen+ at the same power draw due to heat density.

 

I've remounted like 4 times now, could the backplate be the issue or something? I tried screwing the mounting bracket down as much as it can possibly go, and I've repasted 4 times, each after remounting the cooler. Also, I found a 3950x running at 71 degrees on the same NH-D15, mine is on open air so it should have an advantage? 85 degrees for a much lower tier CPU must mean I am doing something wrong, right?

[porina]

Sounds similar to my 3700X system with D15. At stock (88W PPT limit) It's hitting low 80's in a case, one fan on cooler only, with RPM reduction cable fitted. Power density is a pain on Zen 2 if you like low temps.

 

3950X will have the cores spread over more than double the area compared to yours, hence easier to cool.

[Overtime]

1 minute ago, porina said:

Sounds similar to my 3700X system with D15. At stock (88W PPT limit) It's hitting low 80's in a case, one fan on cooler only, with RPM reduction cable fitted. Power density is a pain on Zen 2 if you like low temps.

 

3950X will have the cores spread over more than double the area compared to yours, hence easier to cool.

Maybe you're right, that actually sucks. No idea why people would buy these kinds of coolers if they make literally no difference for Zen 2, thanks for your help.

[-rascal-]

4 minutes ago, Overtime said:

 

I've remounted like 4 times now, could the backplate be the issue or something? I tried screwing the mounting bracket down as much as it can possibly go, and I've repasted 4 times, each after remounting the cooler. Also, I found a 3950x running at 71 degrees on the same NH-D15, mine is on open air so it should have an advantage? 85 degrees for a much lower tier CPU must mean I am doing something wrong, right?

 

You see the 3950X capping out at ~4.0 GHz.

The 3950X is not overclocked, running a stock, and auto turbo boosting to 4.0 GHz from the 3.5 GHz Base clock.

 

You are overclocking all cores to 4.5 GHz, and using 1.33V (what is your CPU LLC setting? Is it overshooting higher than 1.33V?)

 

That is a little high for Ryzen 3000-series CPUs for 2/47 use.

Supposedly, ~1.325V is the maximum safe high-current load, and beyond will degrade the CPU over time.

 

[Overtime]

2 minutes ago, -rascal- said:

 

You see the 3950X capping out at ~4.0 GHz.

The 3950X is not overclocked, running a stock, and auto turbo boosting to 4.0 GHz from the 3.5 GHz Base clock.

 

You are overclocking all cores to 4.5 GHz, and using 1.33V (what is your CPU LLC setting? Is it overshooting higher than 1.33V?)

 

That is a little high for Ryzen 3000-series CPUs for 2/47 use.

Supposedly, ~1.325V is the maximum safe high-current load, and beyond will degrade the CPU over time.

 

I was giving that overclock as a random example, I couldn't possibly use it as it's at 100 degrees and shutting down, I just thought it might be useful for someone to know it was at 100 degrees. I use 1.25v @ 4.3GHz and it's at 83 degrees now, is this a reasonable number or is it too high for my system? This OC is 91w.

[porina]

5 minutes ago, Overtime said:

Maybe you're right, that actually sucks. No idea why people would buy these kinds of coolers if they make literally no difference for Zen 2, thanks for your help.

The D15 was made long before Zen 2 existed. And they do still make a difference. The Wraith Prism that came bundled with the 3700X I find really noisy and the CPU runs even hotter. I did have slightly lower temps with a 240mm AIO than I did with the D15, but I don't know why. I wonder if the water block is more effective in spreading the heat than the heatpipes on the Noctua.

[Overtime]

Just now, porina said:

The D15 was made long before Zen 2 existed. And they do still make a difference. The Wraith Prism that came bundled with the 3700X I find really noisy and the CPU runs even hotter. I did have slightly lower temps with a 240mm AIO than I did with the D15, but I don't know why. I wonder if the water block is more effective in spreading the heat than the heatpipes on the Noctua.

That's very true, thank you for the information! I'd actually assume water would be better than heatpipes for this kind of scenario too, but the cooler was like $65 USD so an AIO would be out of budget anyways (they are all more expensive in Canada).

[porina]

1 minute ago, Overtime said:

That's very true, thank you for the information! I'd actually assume water would be better than heatpipes for this kind of scenario too, but the cooler was like $65 USD so an AIO would be out of budget anyways (they are all more expensive in Canada).

The D15 was $65? That's much lower than here. It's essentially £90 for the Chromax version, and the AIO I had (now in other system) is a Cooler Master MasterLiquid 240, at £80. (Current Amazon UK prices)

[glenalz81]

3 minutes ago, Overtime said:

I use 1.25v @ 4.3GHz and it's at 83 degrees now, is this a reasonable number or is it too high for my system?

Is this 83c in Aida?   If it is then it's kinda where i would expect it to be.

 

What's your ambient temp?

[Overtime]

11 minutes ago, porina said:

The D15 was $65? That's much lower than here. It's essentially £90 for the Chromax version, and the AIO I had (now in other system) is a Cooler Master MasterLiquid 240, at £80. (Current Amazon UK prices)

It's regularly ~$120 CAD ($90 USD), but considering I was waiting for the 3080 for like 4 months, I was able to wait for all PC component sales for the parts I wanted. I was under the assumption that a 220w gigantic air cooler would cool much better than a 260mm AIO.

 

I was watching GN's video about AIO vs air cooling, and at 150 watts, it said the NH-D15 was at 47 degrees.... I assumed if it could cool a 150w CPU to 47.5 degrees, my 95 watt OC would be significantly colder (or around the same temperature AT LEAST), but I have no clue what went wrong.

 

He also cooled a 3950x at 200w to a freezing cold 55 degrees on the same cooler (100% fan speed)!! (click for video). I defenitely didn't think I would ever need to upgrade my cooling capacity to an AIO, but I guess I was wrong. If I'm not doing anything wrong, this kind of video is a bit misleading.

10 minutes ago, glenalz81 said:

Is this 83c in Aida?   If it is then it's kinda where i would expect it to be.

 

What's your ambient temp?

Around 22 degrees in my house I'd guess. +- 1 degree.

[glenalz81]

Just now, Overtime said:

Around 22 degrees in my house I'd guess. +- 1 degree

Seems fine then.

 

2 minutes ago, Overtime said:

I was watching GN's video about AIO vs air cooling

That was a noise normalised thermal test you linked to at 40db and the 47.5c is degrees over ambient.

[Overtime]

Just now, glenalz81 said:

Seems fine then.

 

That was a noise normalised thermal test you linked to at 40db and the 47.5c is degrees over ambient.

Oh my bad, I didn't read the details on the bottom. Seemed a bit too good to be true lol. Thanks for your help!

[Cyracus]

16 hours ago, porina said:

The D15 was made long before Zen 2 existed.... I did have slightly lower temps with a 240mm AIO than I did with the D15, but I don't know why. I wonder if the water block is more effective in spreading the heat than the heatpipes on the Noctua.

The age of the d15 design doesn't affect it's cooling ability. The d15 performs in the range of 240mm AIOS, some AIOs are better, some are worse (primary factor is rad heat dissipation capacity)
 

[porina]

8 hours ago, Cyracus said:

The age of the d15 design doesn't affect it's cooling ability. The d15 performs in the range of 240mm AIOS, some AIOs are better, some are worse (primary factor is rad heat dissipation capacity)

The (unproven) speculation is that due to the more concentrated het generation in Zen 2, a heatpipe design like on most air coolers is less effective at distributing the heat for cooling. A waterblock in that sense is logically more like a single unit so the cooling is better spread.

[Cyracus]

6 hours ago, porina said:

The (unproven) speculation is that due to the more concentrated het generation in Zen 2, a heatpipe design like on most air coolers is less effective at distributing the heat for cooling.

I guess that sounds plausible

[MawzrWerx]

 

 

Hello Folks,

 

I've booked a Noctua NH-D15 Chromax which I look forward to installing and doing a couple benchmarks like Heaven and Blender 2.91.2. Sure I could run IBT or Prime95 for torture testing, even OCCT/AIDA64 but I'm not really going to push the limits as my interest is to get a bit more over than stock with my 3600XT. I have to also state that at the time when I ordered that many of the Ryzen's I had been considering weren't available at the time.

Yet I am laughing as they came available it reminded me of how a Fractal case was just released as I was just receiving my Fractal Design R5 ("non-foo foo side glass"). Now I also need to state that I'm not into sitting around and looking at blinking|pulsing LED's either as I usually to busy playing something or perhaps doing a bit of transcoding.

As we all know every processor (CPU|GPU) needs to dissipate heat either passively or actively. Generally water AIO dumps to the exterior of a case, while air units it's transferred to the case interior. So we have static pressure fans to push|pull air through fins stacks, and air flows are used to vent the air out of the case.

Of course most people are thinking, "No duh". Yes, I am aware that I am writing about what's obvious to those who've been building systems have learned. Regardless there will be people who are new to building systems. Fortunately we have content released by different reviewers around the world such as Linus, Steve, Jay and Lauren to name a few.

I'm hoping by now the OP has acquired a new case or has added/upgraded his previous fans as I read a few of his posts and they all seem to point that it's not properly ventilated overall. And one can implement a negative, neutral or positive air flow configuration.

Currently I have 4 (140mm) SilentWings fans to provide ventilation on my case. Two front intake, front door removed. One side exhaust. One rear exhaust, CPU monitor controlled.

And here are some of my results and other information; Win 10 Pro 20H2 Build 19042.844, X570 Taichi AMI P3.40 28c CCD1 74.8c, G.Skill F4-3400C16-8GSXW XMP 2.0 16-16-16-36 52 T1 31c 32Gig's, Ryzen 5 3600XT 4.625Ghz core's 3&4 FCLK|UCLK =1700, Blender 2.91.2 bmw27 3 min 59sec, Cinebench R15 = 1686 (multicore), Cinebench R20 = 3847 (multicore). Sorry as I kept this information condensed, as yet one can see that with my (old) CM 212 EVO kept my temperatures down fairly well with two Arctic 120 fans installed as I did not like the stock fans that came with cooler originally and you would be correct as I did implement a custom cooling curve. No I have not used a DDR calculator, but merely change some basic stuff in my BIOS. I had been considering a Arctic 280 AIO as well.

I cannot quote Steve, and I will concur that the Noctua cooling solution doesn't provide "magic" to cooling, nor does water solutions as even as I learn it's all the various elements that provide or resolve issue's as a whole. But for my build it should provide better cooling than my 212 which performs far better on my "old" 2600. For amusement I took my ancient Corsair H80 and installed it with the Arctic fans push|pull on the 2600 and it still worked very well after wiping off the dust from it sitting on the shelf for a few years. It's a hold over from my FX-8350 which provided much fun and learning for about 7 to 8 years. But that's a different story for another time.  lol

 

<edit> For quick content reference (Linus himself) </edit> 

 

Thank You.

Edited March 4, 2021 by MawzrWerx
added a content reference link