[SOLVED] Are My RTX 2070 Super Temps Too High?

Heat_Fan89

Reputable
Jul 13, 2020
534
284
5,290
My RTX 2070 Super is an OEM GPU that came with my Alienware Aurora R10. It's still in it. It can play Dying Light in 1080p Max settings but my temps shoot way up after 15-20 minutes. The stock R10 fan curve which I cannot adjust idles, at 999 rpm and it only increases once the GPU hits 83c. Then it ramps up to around 2800 rpm. I had to install MSI Afterburner and even then if I run the fans at a constant 2500 rpms my GPU temps still hit 81-83c and my VRAM temps hit 97c.

So I took the side cover off and ran the fans at 75% which is around 4500 rpms and it ran much cooler. Here's a screenshot of what those temps look like.

https://drive.google.com/file/d/1SEIq9AAYKNcAWV_ZxcXpzdUBUsZ8wFJo/view?usp=sharing

Edit: The OEM RTX 2070 Super only has 1 fan. I guess Dell didn't want to blow hot air in the direction of the CPU and increase temps inside the case.
 
Last edited:
Solution
My RTX 2070 Super is an OEM GPU that came with my Alienware Aurora R10. It's still in it. It can play Dying Light in 1080p Max settings but my temps shoot way up after 15-20 minutes. The stock R10 fan curve which I cannot adjust idles, at 999 rpm and it only increases once the GPU hits 83c. Then it ramps up to around 2800 rpm. I had to install MSI Afterburner and even then if I run the fans at a constant 2500 rpms my GPU temps still hit 81-83c and my VRAM temps hit 97c.

So I took the side cover off and ran the fans at 75% which is around 4500 rpms and it ran much cooler. Here's a screenshot of what those temps look like.

https://drive.google.com/file/d/1SEIq9AAYKNcAWV_ZxcXpzdUBUsZ8wFJo/view?usp=sharing

Edit: The...
Seems to be normal to me, my 2060 used to get about the same temps under load. Its a bit cooler now that my side panel is gone.

try replacing the thermal paste and cleaning the heat sink for good measure too.
 
Last edited:
Seems to be normal to me, my 2060 used to get about the same temps under load. Its a bit cooler now that my side panel is gone.

try replacing the thermal paste and cleaning the heat sink for good measure foo.
So 81-83c for the GPU and 97c for the VRAM is within range? It dropped by 40% when I ran the GPU blower fan at 75% and the side cover off.

If you have any warranties on that changing the thermal paste will void it.
Yes, it's a two month old system and still under warranty with Dell. Thanks for mentioning that because I was about to order some thermal paste. A few days ago I was thinking about reapplying the thermal paste. I'm glad I waited.

Will I fry my GPU with those GPU temps of 81-83c and 97c VRAM?
 
So 81-83c for the GPU and 97c for the VRAM is within range? It dropped by 40% when I ran the GPU blower fan at 75% and the side cover off.


Yes, it's a two month old system and still under warranty with Dell. Thanks for mentioning that because I was about to order some thermal paste. A few days ago I was thinking about reapplying the thermal paste. I'm glad I waited.

Will I fry my GPU with those GPU temps of 81-83c and 97c VRAM?

i didn’t know you could check memory temps in the rtx 20xx series, but i only use afterburner to see so… I’ve read that gddr6 runs hotter than gddr5 so im guessing thats normal.

those core temps are a bit warm but still within operating range, considering that your gpu only has one fan I wouldn’t worry too much about it.

undervolt and fan curves if you don’t want to void warranty.
 
My RTX 2070 Super is an OEM GPU that came with my Alienware Aurora R10. It's still in it. It can play Dying Light in 1080p Max settings but my temps shoot way up after 15-20 minutes. The stock R10 fan curve which I cannot adjust idles, at 999 rpm and it only increases once the GPU hits 83c. Then it ramps up to around 2800 rpm. I had to install MSI Afterburner and even then if I run the fans at a constant 2500 rpms my GPU temps still hit 81-83c and my VRAM temps hit 97c.

So I took the side cover off and ran the fans at 75% which is around 4500 rpms and it ran much cooler. Here's a screenshot of what those temps look like.

https://drive.google.com/file/d/1SEIq9AAYKNcAWV_ZxcXpzdUBUsZ8wFJo/view?usp=sharing

Edit: The OEM RTX 2070 Super only has 1 fan. I guess Dell didn't want to blow hot air in the direction of the CPU and increase temps inside the case.

RTX 2080 owner here. Generally speaking, most GPUs aren't designed to exceed temperatures of something in the high 80s, so a temperature of 80-something seems high to me. Since you're 2070 super only has a single fan, you'll need to run it at high RPMs to get reasonable temperatures. Letting it run for sustained periods in the 80s is not a very good idea.
 
Solution
For a single fan card, those look like ok temps. Could be better? Yeah. Could be worse? Absolutely. You can try to undervolt the card or put some hefty vsynch on your games (like 50 fps, should make most games playable).

Must be a genius whoever came up with the idea to make a damn single fan 2070 lmao. Usually single fan cards are reserved for lowtier models (ie 1030, 1050ti, rx550). Definitely not a 2070 super.

Oh ok i just read it's an alienware. Yeah mate, im not surprised. No offense to you, but they have been proven to be trash overpriced preassembled company for the most part.
 
For a single fan card, those look like ok temps. Could be better? Yeah. Could be worse? Absolutely. You can try to undervolt the card or put some hefty vsynch on your games (like 50 fps, should make most games playable).

Must be a genius whoever came up with the idea to make a damn single fan 2070 lmao. Usually single fan cards are reserved for lowtier models (ie 1030, 1050ti, rx550). Definitely not a 2070 super.

Oh ok i just read it's an alienware. Yeah mate, im not surprised. No offense to you, but they have been proven to be trash overpriced preassembled company for the most part.
Yeah, that's how bad the GPU scene is today that I resorted to buying an Alienware crap box. I paid $1250 for a decently spec'd system (Ryzen 3800X, 512GB Hynix SSD, RTX 2070 Super, Liquid Cooled w/ 1000W PSU, Wifi 6) when Alienware was trying to unload their back stock. I feel bad for those that spend $2000 or more thinking they are getting a top flight gaming rig. That thinking could NOT be further from the truth.

How much should I under volt and can I do that with MSI Afterburner?

Edit: Just want to give a BIG thanks to everyone else who replied and gave your opinion and advise, much appreciated !
 
Last edited:
How much should I under volt and can I do that with MSI Afterburner?
You can do this with Afterburner, but it's going to take a few steps to figure out how to do this for your particular video card:
  1. Download and install HWiNFO. When you run it, use "Sensors Only" and scroll down to the section where your video card's stats are
  2. Leave everything on stock, then run 3DMark
  3. On HWiNFO, note what the maximum value for the core clock was. This is to give us a starting point.
  4. Open MSI Afterburner and click on the "Curve Editor" button (Ctrl +F also works)
  5. At 1000mV, set the clock speed to the maximum value HWiNFO reported. Note how much of an offset this is from the baseline curve.
  6. Everything after 1000mV, set it to the maximum reported clock speed
  7. Everything before 1000mV, set it to the same clock speed offset as 1000mV
  8. Rerun 3DMark to verify the curve actually works and doesn't crash the software
    1. If everything is fine, repeat 5-8 but lower the voltage by the nearest 25mV. Repeat until the 3DMark run crashes, use the last voltage value that worked (you can narrow down further if you want)
    2. If it fails, add 25mV and repeat 5-8. Repeat until 3DMark doesn't crash
  9. Make sure to save this profile
When setting the VF curve, you can use tab, shift + tab, and the arrow keys to modify and move between points so you're not using the mouse for precise movements on 40+ points.
 
My RTX 2070 Super is an OEM GPU that came with my Alienware Aurora R10. It's still in it. It can play Dying Light in 1080p Max settings but my temps shoot way up after 15-20 minutes. The stock R10 fan curve which I cannot adjust idles, at 999 rpm and it only increases once the GPU hits 83c. Then it ramps up to around 2800 rpm. I had to install MSI Afterburner and even then if I run the fans at a constant 2500 rpms my GPU temps still hit 81-83c and my VRAM temps hit 97c.

So I took the side cover off and ran the fans at 75% which is around 4500 rpms and it ran much cooler. Here's a screenshot of what those temps look like.

https://drive.google.com/file/d/1SEIq9AAYKNcAWV_ZxcXpzdUBUsZ8wFJo/view?usp=sharing

Edit: The OEM RTX 2070 Super only has 1 fan. I guess Dell didn't want to blow hot air in the direction of the CPU and increase temps inside the case.

Do you have a picture of the inside of the case? When you remove a side panel and see temps drop it means the case is underperforming with regards to cooling. The Founders Edition is going to run a bit warmer.

How many fans are in that case? Just the two at the rear? It looks like you could add a side panel fan if it doesn't already have one.
 
You can do this with Afterburner, but it's going to take a few steps to figure out how to do this for your particular video card:
  1. Download and install HWiNFO. When you run it, use "Sensors Only" and scroll down to the section where your video card's stats are
  2. Leave everything on stock, then run 3DMark
  3. On HWiNFO, note what the maximum value for the core clock was. This is to give us a starting point.
  4. Open MSI Afterburner and click on the "Curve Editor" button (Ctrl +F also works)
  5. At 1000mV, set the clock speed to the maximum value HWiNFO reported. Note how much of an offset this is from the baseline curve.
  6. Everything after 1000mV, set it to the maximum reported clock speed
  7. Everything before 1000mV, set it to the same clock speed offset as 1000mV
  8. Rerun 3DMark to verify the curve actually works and doesn't crash the software
    1. If everything is fine, repeat 5-8 but lower the voltage by the nearest 25mV. Repeat until the 3DMark run crashes, use the last voltage value that worked (you can narrow down further if you want)
    2. If it fails, add 25mV and repeat 5-8. Repeat until 3DMark doesn't crash
  9. Make sure to save this profile
When setting the VF curve, you can use tab, shift + tab, and the arrow keys to modify and move between points so you're not using the mouse for precise movements on 40+ points.
Wow, thanks for the detailed info, much appreciated. I'll give that a try as soon as I get the time. I'm dealing with other Gremlins in this POS Alienware R10, atm.

Do you have a picture of the inside of the case? When you remove a side panel and see temps drop it means the case is underperforming with regards to cooling. The Founders Edition is going to run a bit warmer.

How many fans are in that case? Just the two at the rear? It looks like you could add a side panel fan if it doesn't already have one.
The Dell XPS and G5 series as well as the Alienware desktops use an antiquated internal design. They look huge with the covers on but the inside of my case is a "Nano Chassis". There are two fans, one on top if you have a liquid cooled CPU like my R10 and a lower front fan blowing across the bottom towards the GPU.

The Alienware desktop is as "proprietary as you can get". Air flow is beyond lousy with temps rising fast under major loads such as playing "Dying Light" for 15-20 mins with Max Settings.

The PSU is on a swing out arm that covers the CPU. It's basically a tight fit or to use another analogy like trying to ask a 275 pound man to put on size 34 jeans. 😀

desktop_aw_aurora_r10_mlk_mod_vid.jpg
 
Last edited:
  • Like
Reactions: Roland Of Gilead
You can do this with Afterburner, but it's going to take a few steps to figure out how to do this for your particular video card:
  1. Download and install HWiNFO. When you run it, use "Sensors Only" and scroll down to the section where your video card's stats are
  2. Leave everything on stock, then run 3DMark
  3. On HWiNFO, note what the maximum value for the core clock was. This is to give us a starting point.
  4. Open MSI Afterburner and click on the "Curve Editor" button (Ctrl +F also works)
  5. At 1000mV, set the clock speed to the maximum value HWiNFO reported. Note how much of an offset this is from the baseline curve.
  6. Everything after 1000mV, set it to the maximum reported clock speed
  7. Everything before 1000mV, set it to the same clock speed offset as 1000mV
  8. Rerun 3DMark to verify the curve actually works and doesn't crash the software
    1. If everything is fine, repeat 5-8 but lower the voltage by the nearest 25mV. Repeat until the 3DMark run crashes, use the last voltage value that worked (you can narrow down further if you want)
    2. If it fails, add 25mV and repeat 5-8. Repeat until 3DMark doesn't crash
  9. Make sure to save this profile
When setting the VF curve, you can use tab, shift + tab, and the arrow keys to modify and move between points so you're not using the mouse for precise movements on 40+ points.
😗
That's a good deal easier than what I came up with. Nice!
 
  • Like
Reactions: Roland Of Gilead
After stewing on what I wrote, if you can't use the maximum default clock speed at 1000mV, you could also lower the clock speed instead by 25MHz or 50MHz, do a test run, and then lower the voltage if that passes. I guess it depends on where you want that sweet spot of performance and temperature to be.
 
So check this out. My temps plummeted even with the side cover on. Did the thermal paste break in?

I went from 81-83c (GPU) and (97c VRAM) with 70% on the blower fan. Now with 65% fan on, my GPU temps are 57c and the VRAM is 83c and the side cover on. That's a big difference with 5% less fan running. I still have not underclocked or under volt the GPU.