[SOLVED] RX 470 Graphics card unexplainable higher temps on intensive games.

[Loczx1]

So as the title says, recently my GPU's temps are incredibly high, to the point of where it starts throttling (on more intensive games).

Previously during benchmarks it would hit a maximum of 58C, that same benchmark atm pushes it to 85+ C. This causes it to get extremely hot, and start throttling decreasing preformance during gaming.

It previously worked fine however, so I assumed at first it was dust, which I had cleaned, but afterwards it only decreased a few C at most.

For the record, my PC was used when I bought it, and the thermal paste on the CPU was dried off in a few months of usuage, but I'm not sure if it would be the same case with the GPU.

It also does not reach that high temps on non intensive games, working fine. It would for example throttle on Jump Force or Civ 6, but not on League Of Legends or CSGO.

One final thing to note is that my CPU atm temps are high as well, due to bad thermal paste until the one I ordered arrives, not sure if that can affect GPU temps as well.

What would be the cause of this? The fans are spinning as well, so I'm not quite sure.

RX 470 AMD.

 

[rgd1101]

case fan?

 

[Loczx1]

case fan?

I'm not sure if you're asking if there's a case fan, but if that's the case yes, I have two (with one being by the power supply) case fans other than the GPU's and CPU's.

 

[rgd1101]

check them? are they running?

 

[Gzr]

Polaris cards usually run hot and the 400 series were sort of worse than their successors (ime). If you think you can, reapply thermal paste to the gpu, although whether that'll help or not is debatable. Also, try undervolting. That'll certainly lower your temps. You can also set a more aggressive fan curve. I find I prefer fan noise over a boiling card. Taking the side panel off usually works too, in a pinch.

The fact that it didn't go that high before is suspect, though.

 

[Loczx1]

check them? are they running?

Yes all of them are.

 

[Loczx1]

Polaris cards usually run hot and the 400 series were sort of worse than their successors (ime). If you think you can, reapply thermal paste to the gpu, although whether that'll help or not is debatable. Also, try undervolting. That'll certainly lower your temps. You can also set a more aggressive fan curve. I find I prefer fan noise over a boiling card. Taking the side panel off usually works too, in a pinch.

The fact that it didn't go that high before is suspect, though.

What's making me worried is as you said, it didnt use to be that high, I did try setting a higher fan curve, still same temps however. I have a suspicionit might be the thermal paste, as the one that came with CPU also wore off after a few months, and its a used computer.

do you have anything else that might be the cause of this issue?

 

[Gzr]

I don't, although to be honest, the 85C seems a more expected temperature for that card, rather than 60C. The latter seems more peculiar to me, really! In any case, repasting is pretty easy so you can give it a try and see how it goes.

 

[Loczx1]

I don't, although to be honest, the 85C seems a more expected temperature for that card, rather than 60C. The latter seems more peculiar to me, really! In any case, repasting is pretty easy so you can give it a try and see how it goes.

It normally would to be honest, but considering it reaches point of throttling (and decreasing ingame fps on multiple games), I think it's not supposed to be that hot.
I already ordered a decent thermal paste for my CPU, so im waiting on that to arrive, I can try reapplying it for the GPU as well.

It would also make sense, as when the thermal paste for CPU wore off it did similar things, regularly working fine, playing things increasing temps allot.

its also worth noting that its regular temps are a little higher as well. normally it would be around 40-42, atm it hovers around 46-50.

 

[Gzr]

I don't want to dissuade you from repasting, in fact quite the opposite, but when my 480 did pretty much the same thing, repasting did very little. Undervolting and better airflow or the side panel off produced the best results and the temps were still higher than I'd want. I had the nitro+ variant but it seems most of them encounter these problems anyway, so... yeah.

 

[Loczx1]

I don't want to dissuade you from repasting, in fact quite the opposite, but when my 480 did pretty much the same thing, repasting did very little. Undervolting and better airflow or the side panel off produced the best results and the temps were still higher than I'd want. I had the nitro+ variant but it seems most of them encounter these problems anyway, so... yeah.

Did you ever manage to fix the temps? Also, how exactly does undervolting work? Last thing is it normal for this issue to just crop up randomly?

 

[Gzr]

Did you ever manage to fix the temps? Also, how exactly does undervolting work? Last thing is it normal for this issue to just crop up randomly?

Better airflow made them better and the side panel off was the best, combined with undervolting. Still it was too high for my tastes and made pretty much everything else in there hotter as well so I got rid of it. Plus, it was a power hog.

Undervolting's pretty easy and you can find it under the performance tab on radeon settings, over to tuning. Enable the first upper left hand side setting to see the states, which consist of frequencies on the top and respective voltages just under those. There, you can input whatever value you want.

This is a trial and error process and, considering the silicon lottery, there are no hard and fast settings that will suit everyone. My logic is that you'll have to modify the values of the last three states. Say 1100mV for P7, 1080mV for P6, 1040mV for P5. Just an example, that.
A quick way to do this is to have heaven running while trying different values. When you get to the lowest stable setting, I'd add about 20-50mV just to be safe. Then you can test stability in games and adjust accordingly.

You'll know it's too little when you get artifacts in heaven or you'll have a black screen or even a reboot, in some cases. Don't worry, just start over with a higher number. For initial testing, I find 5mV increments to be best. You can play with lower values after you've established stability if you want to fine tune it.

 

[Loczx1]

Better airflow made them better and the side panel off was the best, combined with undervolting. Still it was too high for my tastes and made pretty much everything else in there hotter as well so I got rid of it. Plus, it was a power hog.

Undervolting's pretty easy and you can find it under the performance tab on radeon settings, over to tuning. Enable the first upper left hand side setting to see the states, which consist of frequencies on the top and respective voltages just under those. There, you can input whatever value you want.

This is a trial and error process and, considering the silicon lottery, there are no hard and fast settings that will suit everyone. My logic is that you'll have to modify the values of the last three states. Say 1100mV for P7, 1080mV for P6, 1040mV for P5. Just an example, that.
A quick way to do this is to have heaven running while trying different values. When you get to the lowest stable setting, I'd add about 20-50mV just to be safe. Then you can test stability in games and adjust accordingly.

You'll know it's too little when you get artifacts in heaven or you'll have a black screen or even a reboot, in some cases. Don't worry, just start over with a higher number. For initial testing, I find 5mV increments to be best. You can play with lower values after you've established stability if you want to fine tune it.

Is it safe to undervolt?

 

[Gzr]

Is it safe to undervolt?

Undervolting, yes. Overvolting, debatable.