Rx 6800 issues in games (only one so far) along with benchmark and stress test results.
Rx 6800 issues in games (only one so far) along with benchmark and stress test results.
Computer Specifications: GPU: XFX SPEEDSTER SWFT 319 RAD RX6800
CPU: AMD Ryzen™ 7 7800X3D Desktop Processor
Motherboard: MSI B650 GAMING PLUS WIFI BIOS Version: >wmic bios get smbiosbiosversion SMBIOSBIOSVersion 1.F0
RAM: CORSAIR Vengeance DDR5 32GB (2x16GB) at 6000MHz
Power Supply: Corsair RM750e (2023)
Case: NZXT H5 Flow - CC-H51FB-01
Operating System & Version: Windows 11 64-bit Pro
GPU Drivers: Version 32.0.11029.1008 (Adrenalin 24.7.1)
Chipset Drivers: AMD Chipset Software 6.07.22.037 installed; chipset version approx 10.0.22621.1 (if more data)
Hello team, I’m facing issues with my XFX Radeon RX 6800. The card freezes during certain games and specific benchmarks, making it unclear if the fault lies with the GPU or another component in my new tower.
Description of Problem:
- Crashes occur, displaying a blank screen.
- Fans may overheat or stop completely when power is pressed.
- System becomes unresponsive and restarts on its own.
- Identified issue: Game "Guardians of the Galaxy" (gotg) on Steam.
Testing:
- Stress tests during gameplay caused instability.
- System stayed stable during some heavy workloads.
Actions Taken:
- BIOS updated to latest release.
- Adjusted PCIe mode in BIOS for improved stability.
- Reinstalled AMD Radeon software via DDU in Safe Mode.
- Paused Windows updates to resolve driver conflicts.
- Ran stock settings without overclocking.
Monitoring:
- No overheating observed during crashes.
- HwInfo logs captured the crash events but lacked detailed guidance.
Additional Tests:
- Prime95 (CPU): 12-hour continuous run without issues.
- Cinebench (CPU): Consistent performance; occasional slowdowns in multi-core tests.
- FurMark & Heaven (GPU): Stable over 4 hours without crashes.
- Aida64, Prime95, FurMark, Heaven, Memtest_Vulkan: Experienced significant slowdowns but no complete failure.
Notes:
- Disabling Smart Access Memory (SAM) extended game runtime.
- Driver version 32.0.11029.1008 was used; further checks needed if more details are required.
Any suggestions would be appreciated. Thanks!
Check if lowering the GPU TDP to 80% resolves crashes. If it works, the issue may be with the GPU or power supply. If not, it’s likely a hardware problem beyond just the GPU/PSU. Also, confirm if you’ve downloaded chipset drivers from amd.com.
I'm not familiar with the x3D CPUs but the symptoms you describe look awfully similar to an underpowered PSU. Do you have another PSU you can use for testing ? Preferably one with a higher wattage, but any PSU can do for a quick test. I know that the rx6800 had a recommended PSU of 750W, but this was before the introduction of Ryzen 7800X3D CPU. 750W may be enough for most operations but if a spike on power usage occurs, the PSU may fall into overcurrent protection and briefly shuts down then back up. As always, if you can get your hands on spare components to test in your system, that's the best way to find which one is faulty. Good luck !
Verify your memory using Memtest86+ or Passmark Memtest86. Run it without Expo or use Expo profile 2 (below 6000). This looks like a memory issue.
I had completed the memtest86 verification and it confirmed success after several hours of operation. That made me question my assumptions. Initially, I turned off expo and it didn<|pad|> to help, but that wasn't enough. Now I've reactivated it, and I can play Doom Eternal at full settings (including ray tracing) for extended periods. Appreciate the advice!
Thanks for the feedback. The PSU seems adequate, but a test using an extra PSU would be useful to eliminate any issues. I wish I had one myself! Perhaps I could ask a friend for a spare later.
This advice seems reasonable, but I’m a bit concerned about underclocking. It might be seen as overclocking, which could affect the warranty. If I need to return the card, I don’t want to spend time testing it to the point of losing my chance for a refund. OCing has become so simple that it’s easy to do unintentionally. Just one click XD
Normal OCing doesn’t cancel any warranty. You’re not changing anything except the TDP%, just limiting its power usage. In practice, it’s safer than running at full TDP%. It’s likely the simplest test to perform because it poses minimal risk and only slightly impacts gaming performance while generating a lot of data.
Update from October 1, 2024 – I’m reaching out to share my experience with the RX 6800 GPU. After facing ongoing issues, I chose to return it within the first 30 days. The Amazon reseller confirmed the defect and issued a refund of €390, minus €10 for shipping.
During the return, I upgraded to the RX 7800 XT from Gigabyte for about €500. Despite being pricier, I’m pleased with its performance and ray tracing features. It’s more powerful than the older model and has passed rigorous stress tests, though the fan can be noisy. Overall, it’s been a great upgrade.
I also tested running the original RX 6800 at lower voltage and reduced TDP, but it still failed. Many factors could have caused the problems, but ultimately it was a defective unit. Thank you for your help and advice. If others face similar problems, consider returning the GPU within the warranty window to get a refund. Acting quickly is key if issues arise within the first 30 days.