I'm experiencing repeated TDR timeouts while playing demanding games on my PC. I've replaced my power supply with a 1050W unit and switched from an NVIDIA 3070 to an AMD 9070XT, but the issue persists. I've also used Display Driver Uninstaller (DDU) multiple times and ran RAM tests through the Windows diagnostic tool.
The crashes seem to differ by GPU: with NVIDIA, my system shuts down completely, but with AMD, it only crashes the GPU drivers and not the entire system. It typically happens after playing for a while, especially in games like Half Sword and Valheim. I noticed that after increasing the TDR delay, I encountered a different error linked to nvlddmkm.sys, but the crashes continue.
Next, I'm considering replacing my DisplayPort cables and connecting my PC directly to an outlet instead of using my UPS. I might also disable Resizable BAR to see if it helps.
Does anyone have additional suggestions for diagnosing these TDR timeouts? Could analyzing the dmp files or Radeon crash files be beneficial, or would it pose any security risks? This issue started after I ran a task through WAN2.1 using SwarmUI for nearly 40 hours straight.
1 Answer
It sounds like you've done quite a bit already! Testing your RAM with both OCCT and Memtest86 could give you more insights. I recommend running at least 4 passes per stick to be thorough. Doing them one at a time is a good call to pinpoint any faulty sticks.
Also, don't forget to check which specific tests you use; OCCT is great, but Memtest86 is often recommended for thoroughness. It might take a while, but it's worth it to rule RAM out as a cause!
Testing one stick at a time is ideal for isolating issues. Pairs may still work fine but could mask problems with individual modules.
Should I test them as pairs or one stick independently? I have four sticks in total.