RX 580 Pulse Issues

[P4in]

Hello all,

So I just upgraded my Mac Pro (4,1 flashed to 5,1) with a RX 580 Pulse 4GB and I'm running into a strange issue. For some reason the fan that cools the PCI section is louder than usual. When I launch a game (in Windows) after a small amount of time the machine cuts out and it sounds like all the fans are running at maximum speed. (Namely the PSU section)

Tried resetting the SMC with no positive results. Anyone know what could be up or what I could try?

Greatly appreciate your help.

 

[bookemdano]

Do you have the side door on? Have you confirmed that the fans on the card itself are spinning when playing the game?

And when you say the machine cuts out do you mean it shuts down on its own, restarts on its own, or something else?

 

[P4in]

Do you have the side door on? Have you confirmed that the fans on the card itself are spinning when playing the game?

And when you say the machine cuts out do you mean it shuts down on its own, restarts on its own, or something else?

Side door is open. Upon further inspection it’s the gpu fans that ramp up to max. What happens is the system cuts out with a black screen but there’s still power going through it, the GPU fans are running at max when this happens.

I have to hold the power button to power off completely.

Looks like it may be a faulty GPU.

 

[bookemdano]

I wonder if the heatsink on the card isn't making good contact. So as you stress the card it's correctly detecting a high temperature and spinning the fans up but the heatsink isn't properly conducting the heat away from the card. Just a theory, and if you can still exchange or RMA the card it's probably easier/better to just do that.

 

[h9826790]

You can use this command in terminal to monitor the RX580's parameter in macOS.

ioreg -l |grep \"PerformanceStatistics\" | cut -d '{' -f 2 | tr '|' ',' | tr -d '}' | tr ',' '\n'|grep 'Temp\|Fan\|%\|(W)\|Hz'

In Windows, GPU-Z or even Wattman itself can easily monitor the GPU's temperature and clock speed.

If should be very obvious if the GPU is overheating. (Clock low, fan high, temperature high)

Anyway, OP, how you power the 8pin?

 

[P4in]

It’s hitting around 85C before cutting out, that’s within about 1-2 minutes of Titanfall 2 gameplay. Powered by 2 x 6 pin mini pci to 6 pin then and adapter to convert both 6 pin to 8 pin for the card.

The heat sink coming loose could be a viable issue possibly from when the card was delivered hmm (it is second hand from eBay)

 

[thornslack]

Titanfall 2 is dope af. Sorry you're running into this.

I would second gpu z for monitoring in Windows. But something definitely seems awry. Does it do the same thing if stressed under OS X? Cutting to black seems bad. Maybe reinstall drivers? Unfortunately I only have experience with nvidia gpus. Ebay offers strong protections at least for faulty purchases if this proves to be the case. Good luck!

 

[Matty_TypeR]

It’s hitting around 85C before cutting out, that’s within about 1-2 minutes of Titanfall 2 gameplay. Powered by 2 x 6 pin mini pci to 6 pin then and adapter to convert both 6 pin to 8 pin for the card.

The heat sink coming loose could be a viable issue possibly from when the card was delivered hmm (it is second hand from eBay)

There was an influx of RX580 cards into ebay that were used for mining, as the mining get rich quick bubble burst a lot of cards and some with modded bios became available and of course these cards were worked real hard. I would remove the heatsink and check the contact to the GPU and re apply new thermal paste. I even do this to new GPU cards as I don't like the cheap thermal paste which is used from the factory.

Also check to see if the bios has been modded in the past. and like said above if it still overheats then return to seller with your issues.

Good luck, hope you get it sorted out.

 

[P4in]

Thanks for the responses everyone. I think I will try replacing the thermal paste. The card was used for mining but bios was not modified. I noticed something just now, if I change my display rr to 144hz (gaming monitor) this causes the GPU fans to go crazy and cause this cut out issue. If I leave it at 59hz and play the game, the fans come on but the temps are sticking around 60C (fans not as loud either) and there is no cutting out.

So maybe running at 144hz just requires too much power to the card and the 2 PCI cables can't supply enough?

 

[thornslack]

Power isn't the problem. It it were, either the card would not work at all (not enough power) or the Mac Pro would trigger a hard shut down (over drawing power and engaging voltage protection).

The symptoms sort of behave like a card that's being over taxed. It's like playing an old game with out v sync on. The card goes from pushing 60 fps to pushing hundreds of frames per second. This causes everything to heat up and fans increase drastically to compensate. But a gpu is supposed to throttle down once it hits its thermal limit. For example my Titan will start to throttle after 83 c and I've seen it as high as 84 or 85. But it doesn't crash. It doesn't go black. It just dynamically clocks itself to stay functional. So the fact that your card is cutting out points to some other issue. Either the card is overheating to such a degree that it becomes non functional, or the drivers or something else are to blame. Could be worth repasting the cooler as it seems thermally related. However unless I got a stellar deal on the item from eBay I would consider returning it as defective and looking for a better alternative. Although I guess you could still do that after checking the cooler.

 

[h9826790]

Power isn't the problem. It it were, either the card would not work at all (not enough power) or the Mac Pro would trigger a hard shut down (over drawing power and engaging voltage protection).

The symptoms sort of behave like a card that's being over taxed. It's like playing an old game with out v sync on. The card goes from pushing 60 fps to pushing hundreds of frames per second. This causes everything to heat up and fans increase drastically to compensate. But a gpu is supposed to throttle down once it hits its thermal limit. For example my Titan will start to throttle after 83 c and I've seen it as high as 84 or 85. But it doesn't crash. It doesn't go black. It just dynamically clocks itself to stay functional. So the fact that your card is cutting out points to some other issue. Either the card is overheating to such a degree that it becomes non functional, or the drivers or something else are to blame. Could be worth repasting the cooler as it seems thermally related. However unless I got a stellar deal on the item from eBay I would consider returning it as defective and looking for a better alternative. Although I guess you could still do that after checking the cooler.

The throttling point of the PULSE RX580 is 84C, so touch 85C is somehow normal.