PC Graphics

[moose7924]

Whenever I play games my GPU (RX 580) reaches 70°C and I think it's pretty normal but the problem is that it shuts down when it does. I wanted to increase the limit to 95°C but don't know how to do so. can u guys please help me fix this? 

[black_zion]

An overheating GPU does not shut a computer down, it locks up the VPU which results in a black screen and, usually, a message "the graphics driver has stopped responded  and has been restarted", or just locking up the computer if it's unable to restart. An overheating CPU can shut down the computer, however, as can insufficient amperage at +12v.

[moose7924]

The processor must be the issue because i only have a 500 watt power supply

and is not 12 pin one. So will upgrading my power supply work ?

[executablefiles]

Most likely.

[moose7924]

Oh and btw it is not happening because of the temperature issue on my

processor

On Fri, 26 Apr 2019, 4:02 pm Mystic <evilkiller7924@gmail.com

[obts]

I have the same problem too. My Solution

Its temperature problem.

Dont use Auto Fan Speed - The Fan dont start as it should.

Use Custom Fan Speed. 

I start the fan at 18%.

Tested up to GPU clock 1460Mhz

I use Shadow of the Tomb Raider.

System :

Ryzen 5 2600

MSI B450 , 16GB Dual Channel

Sapphire Nitro RX 580 8GB

Dual monitor.

Window 10 Home 64 bit

Store MI

PSU : 600W 

[moose7924]

I am already using custom fan speed. I am thinking of upgrading my PSU for

overclocking purposes as well

>

[pokester]

I agree that these cards tend to thermally throttle but what the OP is explaining is not the same as the normal throttle which will cause hangs to a black screen or for the GPU to reset. So yes a custom fan curve along with setting the power limit to 50 would be a normal course of action I would recommend, but this issue sounds different.

[hardcoregames_]

adding more exhaust fans will work to cool the machine

[moose7924]

My pc has 6 fans bruh

[obts]

When I upgrade my GPU from RX560 to RX580 also same problem like u.

U pointed out at 70°C it shuts. Just like mine.

Now my GPU temp only goes up to 65°C - playing DMC 5.

Before I solved it. Playing CSGO also will black screen.

My fan curve :

0-40°C - 18% , 50°C - 50%, 68°C - 100%.

When my GPU temp hit 65°C , my fan running at 93% (3100rpm)

GPU power draw : min : 35.6w max:182.3w

Memory Controller load: max 95%

CPU temperature : max 70.8°C

Now I set my GPU to 1460Mhz. Always looking at how the fan speed fight the temperature .

BTW. PSU 500W is good for RX580 overclocking though I use 600W.

Fan :

System

Front 2 x 120mm at full speed. Exhaust 2 x 80mm at full speed.

CPU : Stock fan

GPU : Stock fan

[hardcoregames_]

set the fans to 100% at 50C and see if that works better

[moose7924]

But it's the processor for me because the motherboard has 12 pins for

powering the processor and my PSU only has 8

[hardcoregames_]

moose7924 wrote:

But it's the processor for me because the motherboard has 12 pins for

powering the processor and my PSU only has 8

get a better PSU with the appropriate number of cables

my AX860i has cables for all occasions

[goodplay]

Some more INFORMATION REQUIRED WHEN POSTING A QUESTION‌ might help someone, to offer further suggestions.

Some basic troubleshooting you can do,

[ardamir]

First of all, I'm an experienced overclocker. Your CPU is not the problem, it doesn't need any extra power cables. The 8pin EPS connector is good for up to ~400W, good luck on making an R5 2600 draw that much... 
I bet that your motherboard VRM can't even handle 2 power conectors (around 600W tops) and it's there purely for marketing purposes. 

Don't change your PSU for this...

As for the GPU it could be the VRM that's getting too hot because of improper heatsink placement.

[moose7924]

So what do I do ?

[ardamir]

You check your card on another PC. Maybe put a different PSU in your PC to check if it's dying (completely unrelated to the CPU "problem" you mentioned above).

And the last thing you can do (and the one I would recommend you to try first), is flash your GPU with the latest BIOS, provided by Sapphire. Probably the thermal limits are jumbled up in the version your card has. Contact Sapphire, tell them the problem you are facing and ask them for a compatible BIOS for your GPU. Then you need to flash it which is pretty easy, but kinda risky. 

Risky because it could brick your card and you'd need another one to flash it back in order for it to work again.

No need to be scared though. Almost all Sapphire cards have a BIOS switch on them (if you have your gpu in your PC right now, look at it, it should be on the left top side of your GPU) so if the flashing goes wrong in any way, you can just switch to the second BIOS and all will be fine! So no risk for you. 

You can find guides on how to do it on the internet, if you want any help with the process though, do consult me on it. I know it must be scary and complex looking to a newbie. 

[moose7924]

really appreciate your help my man

[moose7924]

hey man, let's just assume that it's not a BIOS issue or another software

issue let's just assume that this is VRM heat sync issue how am I going to

fix this issue because there are no service centers within 500 km of my

city. So I will have to fix this issue by myself but I don't know how to do

so. can you please help me out.

[ardamir]

Just remove all the screws from the back of the GPU and start removing parts of the heatsink. You'll get down to the VRM eventually, depending on the number of screws  
It's really easy actually.
Try that after the BIOS though, because I'm almost sure that this is where the problem lies. 

[moose7924]

what do I do after removing all the screws? and I also wanted to ask does

the VRM look like a processor ?

[ardamir]

You'll need some thermal paste to replace the old one if you are removing the heatsink. So don't do it if u dont have any.

Also watch this vid : https://www.youtube.com/watch?v=iizl9tJfj24

[moose7924]

thank you so much mate, I really appreciate your help.

[pokester]

70 degrees is well within the operating range of that chip. So if you are having issues, you should talk to the maker of your card about an RMA. This is not a typical problem. That is assuming you meet all requirements for using the card to begin with such as a proper power supply.

[qwixt]

Power issues also shutdown the computer. If you can try the card in another computer, you should. I suspect the PSU could be going bad.

[obts]

Get a 700W PSU and see tthe difference

[moose7924]

apreciate your help

[ardamir]

A 700W PSU? Why would he do that? My RX 480 has a hard power limit at around 250W (the voltage controller shuts down after that). With a shunt mod I run it at about ~330W (at 1610MHz) on a 650W PSU. I can't imagine an RX 580 drawing more than 200W, while overclocked for 24/7.

And don't forget that his card shuts down when reaching 70C. Power consumption does increase with temperature, since the resistance increases too. But the increase in power consumption is nelgible, especially in such a small node (14nm LP) that the Pollaris chips are using. Maybe a couple of Watts or so.

Through my testing though, I have noticed that the GPU core gets unstable at around 58-60C, although this is only for extreme scenarios. Can't see how this would affect everyday prerformance.

As I said above, my best guess is that its the BIOS of the card. Let him try that first, instead of spending any money. 

[hardcoregames_]

try maximizing the fans at 50C

[obts]

Read between the lines and all posts.

He wants to buy a new PSU. 

[moose7924]

yeah but I don't want to buy it if it's not necessary

[ardamir]

Well, he answered that himself.... 
So yeah.

[obts]

Bruh....RX 580 have 6pin + 8 pin power connector. 

U connect all ?

[ardamir]

Not really concerning if he has only one of those connected. The 6pin and 8pin (with adequete cables from the PSU, at least 16g) can provide around ~400W each. "No the 6 pin is rated for 75W and the 8pin is for 150W".

Well on paper, yes they are. Their true capabillities though are much more. Also the part about 6pin < 8pin is complete bull... Electronically the 6pin is 3 pins +12V and 3 pins GND (ground). The 8pin is 3pins +12V, 4 pins GND and a sense pin which tells the GPU that "there is an 8pin connected here". The power comes through the +12V lines, so no difference between a 6pin and an 8pin there.

This is not true for really old PSU's though, because their 6pins are 2 pins +12V, 3 GND and a blank one. I doubt that there are still PSU's that old though out there, we're talking about 14-15 year old units. 

And that's not just theory, I have tried it. My RX 480 draws Vcore only from the 8pin connector and the memory power (along with some other stuff, like the PLL rail and VDDCI) comes from the PCI-E slot. I have drawn up to ~330W from that 8pin for extended periods of time and my card is working fine.

So yeah, only a single 6pin is enough to power his GPU, even if it had twice the power consumption. 

[moose7924]

so I contacted sapphire and they provided me with a VBIOS I flashed my

VBIOS and it didn't work so it must be the VRM heatsink issue right?

[obts]

Still not solved ? 

Sent for RMA or get a New one. 

Dont waste too much time on this. 

[moose7924]

there aren't any service centers in around me and the when seller I bought

it from is not taking back the products. he is saying contact sapphire but

there aren't any sapphire service centers in my country so I'm kinda stuck

with this. I have to fix it by myself.

[obts]

Did u test This card on another PC? 

Crossfire Game Compatibility List | AMD Crossfire Wikia | FANDOM powered by Wikia

If the games you play is on the list above. Get a new card to solved your problem and in the mean time solved the problematic card. Once you solved it, u got two rx580 as crossfire.

Rumour that Crossfire 2xRX580=1080TI

Then you will be the envy of the town. 

[moose7924]

buying second Rx 580s is a really good option but what I heard that two RX

580s heat even more