New discrete GPU crashes after launching game, switching to onboard fixes the issue?

Zii

Honorable
Feb 22, 2013
199
0
10,710
5
I had an RX480 8GB blower card that I sent back to XFX for an RMA. They sent me a "new" replacement card because the blower card was unrepairable. The replacement card is a dual fan RX480 8GB black edition, which I assume is better than the blower card for cooling anyways.

I ran userbenchmark on it, and the GPU shows to be slightly above expectations (67th percentile?) so I try to run DXMD to test it out.... and I encounter my first problem.

The game loads up fine! For about the first minute on the main menu, then I get a completely blank grey screen and the monitor says "no signal" but I can still hear the game running in the background. CTRL+ALT+DELETE sometimes gives me my display back, but sometimes it doesn't so I can't kill the game and need to reboot.

Since this has been happening each time I run the game, I tried ripping the DisplayPort cable from my GPU and connecting it to the onboard APU (4790k). This actually works and my display turns back on immediately! What's even better is that despite now running off my APU, the game is still using the GPU to render, and it may be at 98% usage in task manager but it runs and I get 30-50fps on ultra.

What I don't understand is why my screen dies out on me in the first place, and why the APU is able to handle the load and my "new" GPU is not. If I run the game from scratch while plugged into the APU, it will give me shit framerates (2-7fps on ultra) but it still RUNS. It doesn't kill the signal to my monitor.

What I have tried:

  • Updating AMD drivers
    Rebooting
    Clean install AMD driver, DDU
    A different DP cable
    APU graphics (which works)
    A different DP port on the GPU
    Changing game from Ultra to Low settings
The only reason I think the signal is being cut is because the GPU is nearing full load. However it should just give me poorer framerates or choppyness, and not kill the signal entirely?? I've never had this happen in a game before. This only occurs if I try to launch the game above a 1080p resolution, or above medium graphics. For comparison, my old blower card was able to handle 1440p at high without killing the display.

What else could I be missing here? I originally overclocked the new RX480 about 5% without issues and thought that maybe that's why the game was crashing, so I changed everything back to stock and it still kills the display. I really don't know what to try at this point and I would hate to RMA the card I JUST got a few days ago because I feel like there's a fix. And it's not like it gets super hot anyway, it kills the signal to my display at about 65c. That's quite warm, but I wouldn't say it's "crash immediately unstable warm". When using the APU I hit almost 80c and it still "runs".... So what's up here?

I'm also aware that in Windows 10 you can run off an APU, and have certain applications set to use a discrete graphics card, but I'm on a much older build of Windows 10 and this feature was not available. And I wouldn't want to use my APU by default anyway, just a thought.

My original RX480 warranty expires in 2 weeks so I'd like to get this sorted before then in case I do really have to RMA this 2nd card... Thanks guys!
 

Zii

Honorable
Feb 22, 2013
199
0
10,710
5


I can't update it for compatibility reasons. I'm stuck on build 16299.19 indefinitely. Honestly, it shouldn't matter anyway because DXMD was out for a while by the time that build was released, and so were the RX480 cards.

Especially since my faulty blower card ran this game, I don't see why the replacement wouldn't. The only difference is the clock speeds and an extra fan.

As far as I understand, yes. I have the 8pin all the way in. I don't understand why this card is 8pin only and the blower card was 6+8pin?
 

Zii

Honorable
Feb 22, 2013
199
0
10,710
5


i7 4790k
RX480 8GB
16GB DDR3
512GB SSD



scout_03.... Yes, it's a genuine version. I have software installed for my courses that does not work with newer builds for whatever reason.
 

Zii

Honorable
Feb 22, 2013
199
0
10,710
5


Dual... BIOS? I've never heard of that. How should I check it?
 

Zii

Honorable
Feb 22, 2013
199
0
10,710
5
Is my only option really throwing the GPU in another system, uninstalling Nvidia drivers to install AMD drivers again, and logging in to Steam only to re-download a 70GB game just to see if it runs? There's nothing else I can do to try to get it working properly on this completely different system?
 

Zii

Honorable
Feb 22, 2013
199
0
10,710
5


All 3 of my external drives failed in the past couple months. My 1TB seagate being the most recent in late Jan.

Either way, I did as you suggested and was able too run furmark on the RX480 while it was in a different PC.

So what could be the issue? My power supply in the PC that gives a blank screen is 550w. That seems to be more than enough according to every online PSU calculator I have tried. Any idea why I'm getting a dead screen when I try to run something GPU intensive? It even crash on me during a skype call!

I have tried every AMD driver version in the past 2 years, and DDU. I don't think it's a driver issue. I'm really out of ideas because the card doesn't appear to be faulty in my other system.

Also, it's not the display itself or the cables either. Have went through all of them.
 
the problem i could see if it work on others system would be the os that you could not update and made issue with the gpu driver in system see if they have a updated version of the os at the school or you might have to buy yourself and new os to install on the system and update it to last version .
 

Zii

Honorable
Feb 22, 2013
199
0
10,710
5


I see what you're getting at, but do you really think there is no possibility to get the GPU working on that older build of Windows 10? If I recall correctly, the 480 is actually older than build 16299.19, so by the time it came it the drivers should have been stable.

I feel like I'm still missing something else. The RX 480 blower card was an identical type of card as the one I'm using now, and there was never a display issue. Sure, different model but same generation.

Is there anything else I may have not tried? Updating isn't really an option because I would lose the ability to run certain 3D applications, which I need for my projects. But I also can't use them in their current state either because my GPU kills the display after about 8 seconds of rendering. Between a rock and a hard place here...
 

Zii

Honorable
Feb 22, 2013
199
0
10,710
5


Managed to get the newest AMD driver working with relive, 19.1.1 or something

But everything still crashes if the GPU goes over 50% load... This isn't right. I pulled it out and tested it again in the other PC, no problems

Also should not that the other PC is also not up to date and runs an older version of windows 10 but does not experience this issue. It's a DDR4 system and the PC it's not working in is a DDR3 system. Is that important?
 

Zii

Honorable
Feb 22, 2013
199
0
10,710
5
MSI Z97I Gaming AC LGA 1150
It's a mini itx, so sadly one one slot.

No other GPU has this issue, I even pulled out my old 7950 and it "tries" to run benchmarks in 4k but still doesn't cut the signal.
 

Zii

Honorable
Feb 22, 2013
199
0
10,710
5
Update:

I believe I resolved the issue (for the most part) by:

  • DDU all drivers again in safe mode
  • clearing CMOS
  • installing the most recent WHQL drivers
  • setting the GPU to a slight, stable overclock
  • running the AMD pixel patcher tool
I'm still occasionally getting crashes, flickers, and dead screens, but only about 1/7th of the time.
I'm still not quite sure why I can't recreate this issue in any other system, but at least it is tolerable now.
 

ASK THE COMMUNITY

TRENDING THREADS