cancel
Showing results for 
Search instead for 
Did you mean: 

Graphics Cards

jokertug2211
Adept I

radeon vii is getting to 100 celicius easy some times 110 degrees is this normal?

This is running 3 monitors as well. they are all Samsung monitors one has 144hz but using at 60hz the other two are only 60hz as they cant run at 144hz. I found that in Battlefield V the frame rates set at 60fps fixes any temperature problems. it was normally run at 100+ degrees is this normal for radeon vii to run 100+ degrees? I haven't noticed any performance issues but did notice a burning smell playing anthem when hitting 110+.

13 Replies
bearcat22
Miniboss

I'm not sure what the problem is but this about the 10th post I've seen here and on other forums relating to the Radeon Vii running too hot. Must be something going on. Maybe AMD has got them clocked too high so they can compete with Nvidia's offerings. Don't know.

0 Likes

im wondering if it has to do with the different hardware monitors? like hwmonitor, logitech, gpuz, coretemp, ect.. because the official amd link and rivia tuner monitor doesn't show high temps like the other ones. Maybe they all need updates to show the proper temp while the amd app or rivia tuner show a more accurate reading? I read that there are a ton of temperature sensors on the radeon vii, so like the amd app or rivia should be more accurate. using 3d mark it dosen't even register a temp. Another thing that works is limiting frame rates but i don't know if that's practical loosing performance for safety.

0 Likes

The frame limiting is a good suggestion. You could try a high value like 120 fps and see how the system performs and how the games look. Maybe even a little lower, perhaps 110 fps. You might notice anything in the game but the hardware thermals could be much better. As for software reading temps, I used GPU-Z and it does not appear to accurately monitor my Radeon VII.

0 Likes
khun_doug
Elite

You don't mention what version of Adrenalin you are using (i.e. what driver revision). I have the Radeon VII and I do not see it getting that hot. I am using the latest driver, 19.3.1, I did some testing and discovered that with the card set for Auto Overclock GPU, a benchmark made the fans sound like the machine was going to lift off the floor. If you are overclocking the card, that could likely be a culprit.

The other thing to consider is the temp inside the case. Since the fans draw their air from inside the case, and expel the heat into the same area, there should be enough airflow to keep the interior of the case cool.

0 Likes

I am using the current driver, not overclocking, and my air cooled cpu isn't getting too hot. What temperature monitor are you using? I did try the auto undervolt and temps were better by about 10 degrees.

0 Likes
harbi
Adept III

Did you undervolt your Radeon Vii?

.

Mine is stock 1075 mV, manual undervolt to 989 mV
Edge Temp went down from 75c to 64c
Junction temp from 105c to 88c
.
^ No overclock + manual fan curve ( max 59% / 2750 rpm ) when hitting 85c Junction temp.
.
I don't know why the Vega 56, Vega 64 and the Radeon Vii ship with high voltages. I had a Vega 64, undervolted from 1200 mV to 995 mV for about a year, no issues. My Radeon Vii is running perfectly without throttling or crashing for 2 weeks now. Undervolted from 1075 mV stock to 989 mV
.
Dropped it even further, testing 979 mV now with Anthem, 1440p, uncapped fps, so far no issues.

I tried to undervolt my Radeon VII and the system immediately went unstable. Black monitors, had to use the reset button to get a desktop. However, this ONLY started with Adrenalin 19.3.1. On rev 19.2.3 the undervolt seemed to be working well. AMD changed something regarding the Radeon VII and undervolt with the newer driver. What I have settled on for the time being is using Manual, and just below the Current Speed on the left I set Frequency/Voltage to automatic. The system is stable and no wild junction temperatures.

I didn't know how too undervolt so I was using MSI afterburner, sorta worked sorta didn't. I finally figured out to use wattman. The auto undervolt lowered temps by 10 degrees. So I'll have to look more into that later. I think that the vii has 64 temperature sensors and a lot of the temperature monitoring software isn't supported yet until they roll out updates. That's why I'm thinking afterburner and AMD link should be accurate.

0 Likes

khun_doug wrote:

However, this ONLY started with Adrenalin 19.3.1. On rev 19.2.3 the undervolt seemed to be working well.

That explains it then. I'm still running 19.2.3 

I was going to update to 19.3.1 today but I hesitated. Told myself why update if everything is working without issues.

Did you clean install the newest driver with DDU? it might help, if not, go back to 19.2.3.

0 Likes

Yep, I used DDU and did it from Safe Mode. Installed 19.3.1 from as though Adrenalin had never been installed. It installed clean and I went in and set the Auto Undervolt. I opened Talos Principle and the machine went black. I immediately knew what that meant. Restarted the system and was just doing some browsing and it went black again. After the next reset and startup I set it back to Manual and the machine is stable.

It is worth mentioning that AMD has made errors in drivers before. When they released 19.2.3 I noticed that 19.2.2 suddenly seemed unavailable, while 19.2.1 was still available. And the other thing I did was to use Valley Benchmark with different settings. When I used the Auto Overclock option, I swear the fans on the Radeon VII were going so hard I thought it might lift the Fractal Design case a bit off the floor. Using Auto Memory overclock and Manual (the default curve and with the voltages and frequency set to auto), the fans run but not at such a feverish speed.

I expect 19.3.2 or later will resolve the Auto Undervolt issue. For now, I will stay with Manual auto voltage/frequency. it seems stable and runs cool enough.

khun_doug wrote:

I expect 19.3.2 or later will resolve the Auto Undervolt issue. For now, I will stay with Manual auto voltage/frequency. it seems stable and runs cool enough.

To be honest, I haven't used auto undervolt once, I'll give it a try.

khun_doug wrote:

When I used the Auto Overclock option, I swear the fans on the Radeon VII were going so hard I thought it might lift the Fractal Design case a bit off the floor. 

Buwahahahahaha oh man thanks for that.

Unfortunately, with Auto Overclock, we can't adjust voltage or fan curve. So it's running at high voltages and the last two high points in the auto fan curve settings are, 60% at 94c and 75% at 105c.

For reference, 58% is 2750 rpm.

All Vega cards ship with high voltages, it baffles me how I'm stable at manual undervolt 989mV, down from stock 1075mV. I even applied a manual overclock with my undervolt, 1830 MHz frequency + 1100 MHz HBM frequency. I get extra 5-7 fps but in my opinion, its not worth the increased heat & fan speed. 

0 Likes
jokertug2211
Adept I

 HI after some testing tonight I’m wondering if it had something to do with my power supply being faulty. I got a new psu today and I ran some games like battlefield v and on a single monitor or with three there seems to be no overheating. I think my problem is solved.  Before I was using one cable from a corsair 850 for both of my 8pin slots on the vii. I think the problem of the psu might have had something to do with the sticker on the bottom of the fan but I’m not sure. It was creating friction and my psu would turn off really hard like lets say everything was plugged in and running and then pulling the power cord out form the wall. But im using two cable this time and I got a new 750w psu of the same brand and model it had a sticker too but I was sure to remove it. two different cables for the 8 pins on the vii. I think the problem was mainly or all because the stupid sicker on the fan of the psu.

0 Likes

I would turn on the option to display the FPS while in game play and make sure one of the crashes didn't change a setting that you weren't aware of. I have an EVGA SuperNova 850 and two cables to the Radeon VII. Today I did some testing with Assetto Corsa. At one point the junction temperature was over 100C. In game play I had the FPS display running and at times the FPS dipped under 100. As a test I setup the WattMan profile to activate Auto OverClock for the game. That clearly prevented the FPS from going under 100, but the PC case sounds like it morphed into a drone. I changed it back to Manual with automatic frequency/voltage.

I think what we need is a way to lower the voltage with overclock. There really are some extra frames per second to grabbed in complex displays. I believe the GPU is being overvolted, causing some of this heat.