I want to chronical my recent experience with a card swap with the though that it may help others experiencing similar issues. I am a gamer of 3+ decades and used to build systems (starting way back in the days that AMD came out with their K62-300 which we typically paired with an ATI 2D card and a Diamond Monster 3D).
To set the stage, I have always enjoyed a rock-solid stable system. My last 4 GPU iterations were a 3850 Gigabyte IceQ3 Turbo (crossfire), a Sapphire 4850x2, a Sapphire RX480XT, a Sapphire RX5700XT. This AMD RX6750 was the first refence card I ever bought.
Here we go. I uninstalled my 22.3.1's using AMD's driver cleaner utility and then proceeded to shutdown my machine and swap out the card for the RX6750XT. On start up I installed the 22.5.2's. This is where my stability issues began.
I play Lost Ark generally at a minimum of 1-2hrs to get my dailies done, etc, My system began immediately to power itself down with GPU fans going to 100% (1-2x per gaming session).
TLDR;
Over the next three days, I used the driver cleaner & swapped between the 22.5.2's & 22.5.1's which both exhibited same issue
I created a warranty ticket with AMD (with several communications)
I verified my BIOs version on my mobo - reset to optimal settings, validated chipset drivers
+++keep in mind my previous card RX5700XT ran 24/7 without a hitch+++
I ran some controlled dox' (designed experiments, changing one variable at a time. The only thing I did that I struggled with justifying was I ordered a new Corsair RM850 power supply to replace my dated Cooler Master 750. It had some miles on it, I wanted something modular plus I wasn't comfortable with how solid it's 12v rails were at its age. That coupled with the fact that the 6750 can draw another 25% under load and that's even before talking the spikes it gets.
I had eliminated that it was just isolated to Lost Ark as I could get it to do the very same thing in New World and Wolcen as well.
I was at my wits-end and that when it hit me. 3 decades of working on systems and in so many troubleshooting cases they often come down to heat. I never overclock...ever. I did notice some crazy temperatures with the temps in mid-80's and the juncture going beyond mid-90's!!! My place has A/C and it is frosty since my spouse works shift (my beer does not even sweat).
I immediately took a screenshot of the default fan curve settings, unlocked them and made the adjustments. I left P5 at 73 degrees max since anything above is quite audible. I did however adjust the curve to react to rising temps sooner. One week to figure this out, but my shutdowns are gone.
In Summary, I have a bad card. The GPU may not meet the 70% yield of 7nm Navi specifications. It may not have good contact for heat conduction between the GPU & heatsink (which I am not going to open to find out). Or a multitude of other reasons with the cards hardware that cooling will remedy.
In retrospect, buy all those cards from AMD's partners where they take the reference design and essentially improve on the cooling solution may have been well worth the extra few bucks as I have lost a week on this card. Below are my fan settings.
While I appreciate the help AMD sent via email, I cannot believe they had me almost to the point of taking them up on their suggestion to format and reinstall Windows 10 on systems that was rock-solid with their last gen card a week ago.
1 Solution
