rx6950 xtRX6950XT causes my PC to crash during specific games upon startup
rx6950 xtRX6950XT causes my PC to crash during specific games upon startup
I recently acquired a new Gigabyte RX 6950 XT GAMING OC 16G. Right away, I removed my previous 1080Ti and plugged in a third PCIe 8 Pin cable, beginning the tests. Everything functioned properly, so I removed all old NVIDIA drivers and installed "AMD Software: Adrenalin Edition." The first time I launched certain games, performance improved noticeably. Performance stayed stable until I tried Deathloop. Ironically, it caused a looping issue for my system. After restarting and trying again, the PC shut down abruptly—no blue screens, no error messages, just a faint hum from the PSU.
My initial concern was using AMD Software’s Auto-Tune to overclock my GPU and CPU. I reset both clocks to defaults and retried, but nothing changed. Eventually, the machine went completely black. I reinstalled drivers, updated Windows, cleared updates, adjusted game settings, but nothing resolved the issue.
A friend shared a YouTube video that helped: after it, my PC didn’t crash under moderate loads (around 720p) and the GPU drew only 289W—matching the card’s specs. Testing with ForzaHorizon 5 showed similar results, though it still peaked at 289W. We increased power limits with Afterburner, which boosted the GPU to about 350W—still above the card’s spec. Running Cyberpunk benchmarks gave 28-29 FPS at max settings, and even with Borderlands 3 at full settings, we achieved 40-80 FPS.
We tried different cables, BIOS versions, and even swapped power supplies, but nothing fixed the problem. Eventually, we contacted Gigabyte Support. The ordeal happened on June 8th, and it’s been eight days since then—no response yet.
In the meantime, I’ve listed my full setup: motherboard Gigabyte X570, Ryzen 5 5600X CPU, Radeon RX 6950 XT GAMING OC 16G GPU, 32GB VENGEANCE RAM at 3200MHz, 11kW beQuiet PSU, and other peripherals.
You're unlikely to get help from Gigabyte support—they rank low among Board partners after Asrock in terms of resolving issues. When you submit a return request, they often claim nothing is wrong because their staff lacks the expertise for more than basic troubleshooting. Complex problems beyond simple on/off checks are usually outside their scope. You might be better off returning the card to the original seller and trying a different one. After several customer support interactions, it seems like the situation is typical. It could still point to a driver problem or power supply issues, but you've already attempted fixes with some success, so outrighting it might not be the best approach.
Make sure your PSU is functioning properly, then test your old NVIDIA GPU for power usage. If everything looks good, contact the manufacturer who sold you the new card to request a refund or switch to another one.
I used just one 12pin to 2x8pin cable connected to my power supply. Because the new GPU required three 8-pin connectors, I had to switch to a second cable. We tested all three PCIe power cables that came with my PSU in various setups.
Initially, I returned the GPU to the vendor for replacement today. I’ll close this thread now, thank you for your help. Although I might not be the last person with this problem here are my solutions: First attempt failed, so I connected the GPU via two cablemod extensions on a 12pin→2x8pin cable plus one direct connection. After removing the extensions, I still used only two cables from the GPU. Then I plugged in a third cable and only used one 8pin port per cable. It still crashed with another PSU.