[SOLVED] Rx 580 normal temps ? And undervolting ?

Status
Not open for further replies.
Dec 30, 2018
24
0
10
As the title suggests i recently bought an msi armor rx 580 8gb (there is only that model sadly)
And i noticed that it runs hot
Hotter than than my comfort zone.
I ran some tests like heaven benchmark and the temperature maxed at 89°C which is basically the temp limit on the card thanks to msi after burner.
At first i freaked out then i undervolted it with after burner by reducing power consumption to -20% and the temps are between 75-85 in benchmark and in other games like warframe i get about 65-70.
So in conclusion is undervolting it the way i did is safe ? And are temps before and after the reduction are reasonable ?
Oh and ly specs are
CPU :Intel core i5 6500
MB : asus prime z270 p
PSU : corsair cx550 80+ gold
GPU : MSI armor rx 580 8gb
 
Last edited:
Solution
Where and when did you buy it? Can you RMA/return it for refund?

It's quite possible that the GPU has simply died. Looking at reviews over the last year or two for the MSI Armor series, they've had many thermal issues and users generally are not happy with that line. Biggest AMD manufacturing mistake they'd ever made when they buckled under pressure from NVIDIA and their short-lived programs that forced the Gaming X series design to go NVIDIA solely. I see probably as many complaints about thermals on the Armor cards (but not the Armor Mark II cards, mind you) as most XFX cards.
I would recommend not undervolting like you did (power limit adjustment) since you're giving up performance/ frequency in that method.

Here's a guide using AMDs WattMan which is already installed and running on your machine inside the driver software. By undervolting per that method, you can save power/heat without sacrificing performance. The idea is to get away from "auto" voltage because it's overly conservative.
I can provide voltages to enter if you tell me the 7 frequencies of your card.

Also, my RX480 was hitting 85C @ 2400rpm fans while undervolted until I re-applied thermal paste (Noctua NT-H1) to the GPU. Now it runs 73C @ 1100rpm. It's 4 screws, exactly like doing a CPU (just less TIM because the GPU chip is smaller)

Also, what monitor do you have?
 
  • Like
Reactions: Metal Messiah.
Dec 30, 2018
24
0
10
My monitor is a Condor TV 32 inches wide couldnt get a normal screen sinxe they go way up high in my country
Though are you sure i gave up performance ?
In after burner frequency looks the same tbh
 
Last edited:
Dec 30, 2018
24
0
10
Anyways the 7 frequencies are
STATE 1 : 600
STATE 2 :920
STATE 3 : 1165
STATE 4 : 1240
STATE 5 : 1280
STATE 6 : 1325
STATE 7 : 1365
Quick note if push comes to shove 8f i reapply thermal paste do i need to change the pads too ?
 
Dec 30, 2018
24
0
10
State / MHz / mV
STATE 1 : 600 / 840
STATE 2 :920 / 890
STATE 3 : 1165 / 930
STATE 4 : 1240 / 970
STATE 5 : 1280 / 1015
STATE 6 : 1325 / 1070
STATE 7 : 1365 / 1120

Set memory to 2000MHz / 950mV

The thermal pads are reusable. They should just stay stuck to the heatsink or PCB.
in the wattman settings Memory
under memory timing the frequency (MHz) says 2000MHz max do i change it or leave it be ?
 
...it...stopped? Like, you get black screen for as long as you leave the computer on? Restarting the PC doesn't fix the issue?

FYI, nothing you changed in WattMan would've caused this behavior. The worst that can happen if you input an unstable value is that you'll see a black screen for 1-3 seconds, then WattMan will reset everything back to Auto.
 
  • Like
Reactions: WildCard999
Dec 30, 2018
24
0
10
...it...stopped? Like, you get black screen for as long as you leave the computer on? Restarting the PC doesn't fix the issue?

FYI, nothing you changed in WattMan would've caused this behavior. The worst that can happen if you input an unstable value is that you'll see a black screen for 1-3 seconds, then WattMan will reset everything back to Auto.
I restarted pc 6 times and it doesnt work
What do i do
I was ingame when all the sudden the card stopped and the screen says no connection
 
Take the GPU out and run on the integrated graphics (plug display cable into mobo IO panel) would be the first step.

I'm trying to think, even if you mistakenly added a digit to one of your voltage values (10,000mV instead of 1,000mV) the card should/would still impose a voltage upper limit within safe margins.
 
If the computer doesn't boot with the graphics card physically removed (sitting on a table outside the case). Remove all but one stick of RAM and try again. If that doesn't work, swap out for another stick.

If you have more than 2 sticks of RAM (maybe you bought two kits of 2 sticks) make sure to keep the pairs together.

Is your TV working properly? Have tried a different HDMI cable?
 
  1. Entering too low of a voltage (I guarante all the voltages I gave you were lower than the auto values) will not harm your GPU in the slightest. Therefore, nothing I told you to change has/ had ANY chance of "burning" your GPU.
  2. I don't have control of any user errors you may have made in this process, but as I said, the GPU has fail-safes to prevent damage from too much voltage.
  3. From the time I gave you those voltages till the time the machine stopped working, did you have have the side panel off and physically touch anything?
  4. Have you now tried reinstalling the GPU?
 
Dec 30, 2018
24
0
10
  1. Entering too low of a voltage (I guarante all the voltages I gave you were lower than the auto values) will not harm your GPU in the slightest. Therefore, nothing I told you to change has/ had ANY chance of "burning" your GPU.
  2. I don't have control of any user errors you may have made in this process, but as I said, the GPU has fail-safes to prevent damage from too much voltage.
  3. From the time I gave you those voltages till the time the machine stopped working, did you have have the side panel off and physically touch anything?
  4. Have you now tried reinstalling the GPU?
  1. The number you gave me i wrote them to the letter that i am sure of
  2. I did have the side panel off though not the one that covers the wires then one where gpu and mb is on display but i never touched anything
  3. I did try i formatted the pc , put the gpu in another slot , even went in the bios to see if the motherboard registers the gpu but theres nothing
 

Tigerhawk30

Distinguished
Dec 16, 2015
221
15
18,765
Where and when did you buy it? Can you RMA/return it for refund?

It's quite possible that the GPU has simply died. Looking at reviews over the last year or two for the MSI Armor series, they've had many thermal issues and users generally are not happy with that line. Biggest AMD manufacturing mistake they'd ever made when they buckled under pressure from NVIDIA and their short-lived programs that forced the Gaming X series design to go NVIDIA solely. I see probably as many complaints about thermals on the Armor cards (but not the Armor Mark II cards, mind you) as most XFX cards.
 
Solution
Status
Not open for further replies.