just how concerned about GPU temp of 80-82 would you be?

Summary - I have two Dell Alienware 17 R5 i9-8950HK gtx1080. Original one didn't overclock the CPU at all, even after liquid metal. Second one does overclock well. Did liquid metal on second, but ran out - stretched what I had on heat sinks and CPU/GPU, and CPU is doing ok, but now I am wondering if GPU really did need that last drop, lol.

Basically, I noticed after 5 hours of gaming that GPU on the new one was at 82. That's odd, I thought, I don't think my old one ever broke 72. Anyway, I ran Aida64 GPU test, sure enough, a few minutes in, old one is barely at 70, new one is at 78.

Here's the thing - maybe it's just that the cooler on the new one doesn't fit quite the same. Or maybe the new GPU just isn't as good, while the CPU is better. I don't really feel like doing the 2-3 hour teardown just to add that one drop of liquid metal.

On a desktop, I am more comfortable than most running my GPU at 80 24/7 for doing video and 3D renders. The only reason I am concerned (I have warranty, not that worried about burn out frankly) is just because it will also make CPU hotter than needed while in-game. I do some editing on laptop, but not much for now.

So if the games work, etc., am I ok not feeling too guilty if I just don't worry about it? In a game, I don't care if I downclock the CPU. (I have it fairly stable at 4.6, but I already run at 4.3 in-game just to ensure it won't be making too much heat itself.)

If this was the only unit I'd tested, I'd be fine, I think. I'd chalk the 80degree temp to it being a laptop, etc. But since I can see the old unit runs so much cooler, it is sort of annoying. Given that I can't really be sure that lack of liquid metal is the issue... which, come on, the surfaces were pretty well coated - just not as thoroughly as when I LM'ed the first one, had a really hard time scraping it around to make it work, and when I realized I didn't have enough, I used some on the CPU to be sure that was to my liking first... I think I might have to leave it. (Or go back to the old one but then have a CPU that isn't as efficient.)

Can anyone give some words of encouragement one way or the other?

 

PK3 or ICDIamind is probably more effecitve on both the GPU and CPU. As far as I know, the heatsink fit is mediocre on AW notebooks.

 

Well I used GC Extreme the first time, the Phobya liquid metal was a huge advantage, I'm definitely not changing it. My point was, I am not sure if I didn't use quite enough, or maybe this GPU is just not as good as the other one (or sink fit is not as good). Is it really worth ripping it all apart again just to try to get to the same eight degrees less the other machine has?

Hmm, now I am noticing I am getting power limit in Afterburner - but frame rate doesn't drop in-game. Is that normal? Nothing we can do anyway, we can't undervolt these, only over-volt via that workaround I've seen, does that sound right?

EDIT: hmm, looks like my latest fix did have an effect - I underclocked the core -50 and memory -100, looks like no power limit triggered in past hour. That sucks, +150/+300 was running well (I did run stock at 0/0 also, but got power limits there too). Temp is down though too to 75 max. I am wondering if this GPU just isn't as goof as the one in the previous unit was? I don't think it's a lack of liquid metal then, just different efficiency on this GPU.

But, actually, if I can overvolt via that workaround, hmm. Then no power limit, but temp will be back to 82ish. I guess I will just have to live with it if I want to overclock. (which I do as I am upgrading to 4K screen again soon)

 

The answer to your question is not at all. I wouldn't roll over in bed for anything less than 89C. The range of 80-82C is literally within safe operating temperatures.

 

@Kevin Thank you for that. I just needed some re-assurance. I run a render farm at 80 solid for months on end sometimes, just kind of freaked out when I realized my old unit was getting better temperatures. I re-tested my old one, and the difference wasn't even as big as I thought... the old one does get up to 75 eventually, so not the 10 degree difference i thought (more like 6 degrees). I'm not going to lose any sleep over it.

 

You say both surfaces were pretty well coated with liquid metal, that sounds fine, you don't want to put lots of liquid metal on otherwise it is in more danger of running all around your laptop when you move it, I think it's likely to be manufacturing differences between the two laptops in terms of heatsink flatness and fitment & perhaps differing amounts of voltage that your 2 GPUs use.

 

Thanks - yeah, both sides were covered, just not quite as well coated as I did the first one (but the first one I had too much and had to remove a good bit, hence I was short on the second machine using the same 1g syringe). I literally just wanted 1 or 2 more drops to spread around to feel like it was perfect. I think it will work though and probably it's as good as it will get, as you say. Just a bit of difference in the shape/angle of the heatsink would definitely cause a 5-8 degree difference, for sure.

Thanks again to you both for stopping in, it always feels better when someone re-assures you about something like this.

 

80-82C? No worries. Even 90C is within spec. Game on.

 

I agree that 82 degrees C is fine. For reference, Nvidia's desktop Pascal Founder's Edition cards consistently ran in the 80 degree C range while gaming.

There could be other variables to explain the higher GPU temps on your new notebook. The GPU might be running at a higher voltage (e.g. not as good of a chip), and/or your CPU is producing more heat because it's boosting higher and/or is also running at a higher voltage than in your previous notebook. The cooling between the CPU and GPU is not entirely independent.

Charles

 

Thank you... yes, I think it's probably just not as quite a good a chip as the last unit (but the CPU is way, way better, so this unit is a keeper). Funny thing, I did happen to try some underclocking last night just to see if it helped one game that crashes a lot. Still not even sure if it helps the crashing, but, I noticed that by the time I got to -200 core, -400 clock, I was getting the same low 70s temps as the old unit was at an OC of +150/+300. Well, maybe that last drop of liquid metal that I wish I had did affect it a little bit, but I am leaving it until such time that the metal needs redoing (hopefully not for a long, long time ).

 

Exactly this, reference Nvidia baords run at these temps.

 

Around 80c is actually kind off good. All the gaming laptops I have had easily hit 80c unless I re-paste and have the fans on MAX.

 

Yeah, I guess the first unit I had was simply exceptional (either heat sink fit or GPU chip or both) - I was surprised at first that it only hit 73 even after hours and hours of 4k gameplay (and with overclock), then I kind of got used to it, so the new one was a bit shocking compared to what I had been getting, but yeah, now that I've been using it, seems normal again to be seeing low 80s.

 

Wow, 73C with OC at load? That is indeed exceptional. Unless you live in a country where your room is freezing during winter, I would say that you indeed got a little spoiled haha.
It's exactly like the others have said though. 80-82 is completely fine for the GPU.
Eventually though, if you use your 17 r5 through the years that the temps start to climb towards 90C, that's when you'll notice the real beginning of throttling and where you have to take countermeasures.

 

My gtx870m reaching 92C when gaming. Cpu temp also 90ish. So i think it's ok?

 

Wow that is getting up there! But also an older generation GPU, so I guess that was normal? Yeah, I have calmed down about my temps. I just ran a multi day stress test in an 80 degree F room and no hiccups, so definitely no stability issues with this one running at 83-84; was just lucky to have my last one running in the 70s like it was.

 

After repasting and clearing all excessive dust, mine runs at around 74C (GTX 980M)

 

80*C sounds fine to me.
Could be better probably... but you also have to keep in mind that most OEM's don't really optimise the cooling assembly, thermal paste application or voltages for individual units (which is something they should do - these days it would be a lot simpler to do that with automation/algorithms).

In many cases, re-pasting the CPU and GPU with quality thermal paste such as Gelid GC-Extreme (along with cleaning the internals from dust) has done the trick in helping to run the laptop at lower temperatures, as some OEM's can be REALLY bad at applying them.

Additionally, undervolting both the CPU and GPU also tends to help temperatures quite a bit (though this can be trickier in NV case because they usually lock the voltages and frequencies)... but, even if you undervolt just one component (such as the CPU for example), it will likely affect the GPU as well to a degree because its a small space/enclosure and a lot of laptop CPU's and GPU's share the cooling assembly.

If the laptop is under the warranty, I would not advise re-pasting it as that frequently voids the warranty in question (since that would be easily recognised by the repair teams), and in that case, your best bet would be to focus on undervolting the CPU and GPU if you can.

 

That generation of GPUs had heaps of flaws in them and hence why they skipped that generation in the desktop line.

 

Desktop didn't skip it, 800M was simply Maxwell 1 and rebranded Kepler.

 

oh ok. but they never used the 8xx series on desktops officially and i believed it was due to the laptop GPUs having temp issues.

 

It was because there was no space left in the 700M series numbering scheme for the new Maxwell 1 parts, so they had to move up to 800M, while desktop simply slotted them into 750/750 Ti. The Kepler 800M cards had heat issues because they were pushed beyond the efficiency break point.

 

oh right, cool. thanks for that.

 

I can't speak for laptop GPUs but I ran an EVGA 1060 6gb card in my desktop for months and it stayed at around 82C during heavy use. This was not the SC card which has a real copper heat sink. This was the lowest version that has just a basic aluminum heat sink wrapped in plastic shroud, and could not be overclocked because of that. However, it ran like a top at that temperature and I still have it today in my family PC (upgraded to a 1080 Ti in the gaming PC.)

 

Before repasting and cleaning the fan my Kepler 750m reached temps up to 97C while gaming after over a year since the last cleanup and now its back in the 70C range. Its still hits around 80C at times but its still in the safe range. If 80C+ was really bad for a GPU my 750m would of died years ago.

 

The XPS 9570 has a 1050 TI, when it was released and reviewed the throttle limit was 78c, now that is is post review season Dell has reduced the throttle point to 74c in a BIOS update. To make matters worse you can not downgrade past this BIOS due to a intel boot guard lock.

 

my 1070 tops out in the mid 70s and i have never seen it reach 80.