My NVIDIA RTX 2070 Super (8 Gb VRAM) can easily go over 80 deg C (176 deg F) and as high as 84 deg C when running FS 2020 in a house 26.7 deg C (80 deg F) or so.
I’ve read PC Gamer’s article on How to Lower Your Graphics Card Temperature (6/6/2019) by Jared Walton. He says that although both NVIDIA and AMD claim that their cards are made to take temperatures in the 80 deg C range (even up to 92 deg C on some cards) that the experience of Bitcoin miners(!) running PC’s 24x7 with GPU’s cranked up all the way into that temperature range shows that a GPU isn’t going to last long into the future.
I have a Dell XPS 8930 Special Edition. Reading the Dell forums some users diss the ventilation design of the 8930 case and recommend installing an extra fan to control overall internal heat. In Jared Walton’s PC Gamer article, he suggests using MSI Afterburner and playing around with GPU voltages and clock speed. I’ve read that taking the sides off a computer case may hurt rather than help.
A somewhat related question to GPU performance and temperature is that for machine learning purposes, I’ve installed CUDA drivers and since I was working with MS’s machine learning framework, CNTK 2.7, I only installed an earlier version, CUDA 9.x as required by CNTK, not 10.x or whatever it’s up to now. Is having CUDA associated with my GPU and an older version at that going to affect my GPU performance (and its temperature?). Currently thru GeForce Experience I have my NVIDIA Driver set to be the latest GeForce Game Ready driver (452.06, 8/17/2020) but am unsure whether having any version of CUDA around affects gaming.
In several different threads on this forum, GPU temperature has been suggested as a possible cause of crashes and performance problems. So I thought a thread might be helpful specifically addressing analyzing and correcting GPU temperature and possibly software related issues that cause a GPU not to perform with optimum efficiency and temperature control. One thread already makes a good start on this issue with some particular posts but it’s filled with a number of other posts not specifically focusing on temperature or having more diffuse comments on computer-specific parameters.
Another thread, a very helpful wiki guide on graphics card settings doesn’t really say which settings will have the most impact on GPU temperature. It would be very helpful if comments about temperature were added to the guide and what might have the biggest impact on reducing cooking one’s GPU while still giving a beautiful high-res simulation to view.
Undervolting your GPU can reduce temperatures and increase performance too! as in the guide linked above,
the higher temperature your GPU, the more prone to errors it may be. although usually there is lots of headroom on modern GPU’s, I can even get a stable +200/475 mhz overclock at 80 degrees celsius!
Just add fans to the case if you can. I had issues with overheating CPU and GPU shutting computer down periodically running FS. I have a water cooled CPU. 3 Fans added and I no longer get the temperature spikes I was previously having. Don’t be afraid to take sides off the case, just be prepared to have to clean the interior more frequently from dust.
Running a Intel 9900-KS and GEForce RTX 2080 Super
@PHLMerc9914 Any good reference(s) as to where to add fans? I’ve replaced and added lots of parts to computers over the years but have never “modded” a computer by adding a fan. I’m still only a few months into my Dell warranty, too, so I’d hate to have Dell using the fan addition as an excuse to invalidate the warranty. But thanks very much for any further advice on the adding-extra-fans score.
Or perhaps I should go to Dell support directly(since I’m still under warranty support) and ask them for advice on how best to add one or more fans?
In your case I would check with Dell. I checked the documentation that came with my case for the size of the fans I could fit, probably want to check your motherboard documentation for the number of fan controller outlets you have as well. There are a few good videos on you tube for computer case fan installation.
My card (1080ti) dropped about 10 degrees after changing the fan settings. I was getting up to 84-85 degrees C which was making me a little nervous. It was only really getting to 60% fan speeds with the factory settings…I guess so they could brag about how quiet it was. I set a custom curve so it ramps from 85-100% when it’s 70-90 degrees C now.
Thanks for the suggestion. What card and what rpm is 100%? My RTX 2070S is turning something like ~3K rpm (don’t remember exact) when I’m at 80 to 81 deg C. Normal computing it’s around 1K.
hi, you don’t need to know rpm, just open afterburner and it show you temp limit, then open option found fan and set any you prefer curvature with your temp limit temperature work at 90-100% of fan speed, you can start curve from 50* this is default settings, as before 50* fan just don’t work at all, this is nice, for less dust in your vcard, from 50* you can start curve to 50% fan, then get it up by hyperbola or line or whatever you want for your gpu never will higher than temp limit
I have a Gigabyte AORUS RTX 2080 running MSFS @ 2560 x 1440, I use Afterburner to display GPU temps and FPS for tweaking the games. But I rely on Gigabyte’s AORUS app to run my cards fans. With most FPS games I choose a target temperature (68 C / 154 F) that I want maintained (my preferred Max temp) and my cards fans never have to work that hard to hold that temperature, until MSFS came along.
At first I didn’t even have Afterburner’s monitoring displayed, when I turned it on to start tweaking the Sim’s settings, I almost freaked when I saw the 176 F temp. I opened the AORUS app, changed it from Auto to Manual, and set the fan between 85% - 90% of the cards full fan speed. That keeps it at about 140 - 155 F.
The weird thing is, I noticed when I quit flying, and enter the Main Menu, the temps go up, which makes no sense at all.
I’ve liquid cooled all my processors since 1996, but have never liquid cooled my graphics card, but my next card will definitely be liquid cooled for both the GPU and VRAM. I’ll just have to save a little longer. I’m just waiting for official RTX 3080 or 3090 cards prices to release, so I can fall out of my chair from from shock.
That happens to me, too, and a lot of other folks have noticed that as well. If you turn on developer options, FPS, you’ll see that you’re getting 60 FPS, too, when the computer and screen are actually doing nothing!!! Someone on the forum jokingly(?) asked if Asobo was doing Bitcoin mining while we think are computers are just idling!
Yeah! I’ve always tried to think positive. I got my present computer before FS 2020 was on my horizon and an RTX 2070S seemed like a good Goldilocks price/performance compromise for machine learning where unlike gaming, instant performance isn’t critical - you just wait longer for the same computation that would be shorter on a better GPU. Now with FS 2020, the thought has crossed my mind that I shouldn’t work too hard to avoid frying my “old” GPU as I’ll have the perfect excuse for the wife as to why I need the baddest RTX 30 that I can afford, sooner rather than later.
Did the same, was running 83-84 degrees (started to underclock itself at this point) constantly on my 2060 super, setting the fan speed has lowered this to about 75 degrees.
Because the screaming GPU fans did bother me after a while,
i turned on vsync and a frame rate limit of 30fps in the nvidia control panel for msfs.
(Don’t activate these ingame!)
Before this i had fluctuating fps between 50 and 70, and gpu or cpu load maxed out, depending on the airplane and situation.
Now i have 50% cpu load, 60% gpu load, medium fan speeds,
and a constant 30fps without tearing and without stutter. Runs perfectly fine.
PS: the easiest way to lower GPU power consumption i found was with MSI Afterburner to simply lower the power limit to between 60 and 70, and overclock gpu and memory (200MHz each for my GTX 1070). This results in nearly the same clock speeds as before, but a cooler GPU, as the gpu core voltage is now much lower.
I personally have never had removing the side panel from a case make my temps worse. Many times I have seen it not really make a difference but never made it worse. I have also in many cases have had adding fans do nothing as well. Even really high rpm fans that made a crazy amount of noise lol. Case airflow is a tricky thing, more isn’t always better.
The first thing I would start with is make sure your case has “adequate” airflow and that it has good ventilation with intake air somewhere near the bottom and exhaust somewhere near or out the top. And don’t set the case on carpeting. If you have to put it on the floor at least put a board or something under it.
Then I would use software like afterburner or EVGA’s to set a more aggressive fan curve or control it manually. As well as seeing if your Bios has controls for the case fans. If that doesn’t work move on to undervolting and under clocking the card. In an extreme case maybe look for a different aftermarket card that has better cooling. I rarely get the stock nvidia cards for this reason. Some are ok though. My founders edition 2070 did pretty well for as small as it was. But this is a rather expensive option if you already own the card lol.
As a last ditch effort you could pull the side panel off the case and aim a regular fan at it and see what happens. Oh and if your case has filters infront of the fans try removing them, they can restrict airflow. Especially if they are poorly designed.
Also better fans in general can sometimes make a difference. Some of the fans that come with cases are not the best things in the world.
You might be better off with a new case. I have mine in a Cooler Master M500P, with 3x 140mm fans in front, 2x 140mm fans on top, and 1x 140mm fan in the back. All set with more aggressive custom curves. Even overclocking my CPU (air cooled) it stays nice and cool. At least Dell uses standard parts so that it’s not as much of a pain as it sounds to move the guts to a new case; even I could do it.
The other helpful thing you can do is cap your fps at, say, 30fps. It’s still smooth and nice, but keeps your GPU running a lot cooler, too.
I considered opening my case but I read that removing the case can upset the intended airflow from intake over boards to out the exhaust vents. But if one is aware of this, one can just try it, watch closely, and reassemble the case if it’s not working. Best to look at all the components one can, though, as what I read implied there might be just a few components that got in trouble from missing the air flow they were intended to have - so whoever tries this wouldn’t want to just be looking at CPU and GPU temps but consider all the items something like CPUID’s HWMonitor can report on.
I think most of the time I see ppl take the side off because of heat, they set a regular house fan to blow at it, too. I think it’s better to have a case with good airflow like mine, or see channels like Gamers Nexus that do good case reviews that rate airflow.
Someone on the forum jokingly(?) asked if Asobo was doing Bitcoin mining while we think are computers are just idling!
, I lowered my GPU temp.