My Vega 64 card had been working phenomenally for the past few months but now it seems that it's kaput...
I recently noticed that I was unable to do a normal un-installation of the Radeon drivers on Windows 10 (which is fully up to date at the time of this posting). Didn't really bother me much as I was still able to eventually purge them; I never ran into any issues during the actual installation. Anyway, the graphics card crashes with a black screen almost immediately after it starts to render 3D, which to me can't mean anything good. This happens across many different game titles, and GPU stress tests with both DirectX and OpenGL. I've tried the latest Adrenalin drivers as well as rolling back to a previous set (figured maybe the monitoring tools had a bug) but the problem persists.
I've already opened a support ticket with Sapphire but I can't really think of what could have caused the card to simply stop working properly. The card was receiving sufficient power and temperatures were nominal at all times; card was not overclocked and the problem is still present if I abandon CPU and memory overclocking as well.
What's the rest of your hardware like? It could be a possible power delivery issue only cropping up now as the RX Vega card is being pushed to new limits. List your mobo etc (seeINFORMATION REQUIRED WHEN POSTING A QUESTION.)
When your screen goes black, can you still recover or is your whole system crashed? May indicate an issue with hardware listed above.
Also did you use the most current version of DDU before switching to Radeon Adrenaline - especially since it's a who new release. Before making any changes to Radeon Adrenaline (or any other version) use the defaults first and try it out as is. IF for some reason you feel the need to make changes, do so incrementally and test with each step. Making too many changes at once makes it harder to pinpoint any errors.
Nope, once the screen goes black the system essentially locks up and I need to do a hard reset. I boots back up completely fine though.
System is as follows:
Ryzen 7 1700
ASRock X370 Professional Gaming (with latest 3.10 BIOS)
16GB DDR4-3200 FlareX
Sapphire Vega 64
Seasonic X-750 power supply
Windows 10 Pro 64-bit
Luckily, I have a couple other machines I can try the card in so I plan on doing that but this would be my primary machine. I was thinking that maybe since my PSU is at the minimum per Vega's requirements that it would now be inadequate for the card but that doesn't really explain why the older driver doesn't work now either.
Thanks for your response; I will continue to troubleshoot, for now.
In theory, if you have a fully/properly functioning PSU, 750 should be able to handle a single RX Vega 64. It's cutting it close especially if you're OC'ing (CPU+ GPU). Try it out on another PC if you could and report back.
Ok, so Installed the Vega 64 in another machine with slightly older hardware:
Intel 4670k, Z87 G1 Sniper 5 motherboard, 16GB DDR3 2400... also Windows 10 64 bit
but most importantly, this machine uses a 1000w power supply so I could eliminate that variable. Upon installing the latest drivers, I ran a couple of quick rendering tests. Both times the card didn't crash like they did about 10 seconds in on the other computer. I'm going to run a couple additional tests with more stressing API's though, so I will post back with the results of those shortly. We might be on to something in regards to the card not getting enough juice.
Hope you figure it out. Unless there are other bad components or configuration in your system (motherboard/ram/PSU), 750W should be sufficient.
One of my systems has the following:
AMD Ryzen 7 1700X @ 3.95GHz
16GB Ballistix Sport ram @ 2400MHz
1 SSD and 1 HDD
Sapphire Radeon RX Vega 64 @1650MHz GPU core, 1100MHz HBM2
3 case Fans
Windows 10 Pro, Radeon Adrenaline.
All of the above is powered by a Thermaltake 700W Gold PSU. The CPU and GPU with OC's do not get above 650W at peak combined usage.
Also look into Windows Event Viewer to get more clues as to what caused the crash. It could indicate power issues or other software errors. Type eventvwr from the Windows Run dialog or command prompt, or just type Event Viewer after pressing the Windows key. Under Windows Logs -> Application and Windows Logs --> System look for any errors right after you recover from a crash.
I ran a few DX11 and DX12 benchmarks on the i5 haswell, just to hammer on the card for awhile, but I never experienced any problems. the only thing I noticed was that the heavy power draw was causing a small desk lamp nearby to flicker, though the system remained completely stable. I'm not sure what to make of it other than to just keep troubleshooting the other Ryzen computer--but it seems like the Vega card may be fine after all.
I will reinstall the card in the original machine tonight after work and have a look at the EV. It might just be something simple too, like a loose connection or power saving setting that isn't agreeing with the driver. I can also test the power supply further; hopefully I won't have to source a new one.
I've reached what I believe to be a pretty solid verdict...
The graphics card, drivers and power supply are fine; it turns out that my 4TB game drive has failed, thus resulting in a full system crash and/or lockup. The thing that turned me toward the graphics card was the monitoring software losing the GPU data and the fan on the card itself spinning down, but it would seem that those were simply secondary symptoms. Looks like my video card will live to see another day, but my games will need to be reinstalled.
Thanks for the help!
My previous conclusion has been proven incorrect I'm afraid, thus I've marked the question as not answered (sorry whiskey)...
I was able to retrieve all of my data from the drive that was flagging trouble; replaced it with a brand new Seagate Barracuda but the crashing continues. I've basically tried all I could think of and I've even confirmed that a Polaris card with the latest drivers works fine on the same system, so I think I may need to have the Vega warranty replaced. Also turns out that my PSU is 850 watts and not the previously 750w that I had first brought to attention. I'm not sure why the Vega card worked in the other machine but I can't think of anything else to try.
All was good up until about a week ago and now it just refuses to work as it should.