This ASUS TUF F15 FX516PE often experiences crashes when running many games.
This ASUS TUF F15 FX516PE often experiences crashes when running many games.
Details:
Laptop model: ASUS TUF F15 FX516PE
Processor: Intel i7-11370H
Graphics: NVIDIA RTX 3050Ti
Memory: 16GB DDR4 at 3200 MHz
Operating System: Latest Windows 11
Issue description: When launching tough games (Rainbow Six Siege, CS2, Peak), the system freezes. Audio keeps playing but the game closes after 10-30 seconds. Works on both Windows and Linux. 2D or light 3D games mostly function normally.
Tried solutions:
- MemTest86 overnight – passed all tests
- FurMark and Heaven benchmarks ran well
- GPU drivers reinstalled using DDU Nvidia studio drivers
- Switched to a different USB-C charger
- Monitored temps: CPU ~75°C, GPU ~60°C; fans at 100%
- Adjusted voltage curve in MSI Afterburner – above 700 MHz unstable
- Replaced thermal paste and used putty instead of pads
- Reset CMOS, disabled NVidia HD Audio, changed power plans
- Disabled CPU boost in MSI Afterburner
- Capped FPS in-game to avoid crashes
- Tested with laptop plugged in; automatic cap to ~750 MHz when powered on
- Ran without any software during crashes
Notes: No benchmarks detected the problem except 3DMark. BIOS is locked, so many settings can’t be changed.
Experiment without the NVIDIA HD Audio driver occasionally causes issues when under stress. Reduce the GPU power target slightly in MSI Afterburner, for example to 90–95%, which may help steady power output. Perform a VRAM test OCCT or use diagnostics like MemtestG80/CL. Since FurMark doesn’t fully challenge VRAM like recent AAA games do, this might expose problems. If possible, run tests on Linux using custom NVIDIA drivers versus open-source ones to compare behavior. Keep in mind your GPU remains stable at 700 MHz but struggles beyond it, suggesting a voltage/frequency issue. This could stem from hardware variability or PSU/charging problems during sudden spikes. If none of these steps clarify the matter, consider testing with a different PSU/charger long-term to confirm.
This might indicate a hardware issue, though it seems like faulty Nvidia drivers could be the cause. Dealing with recent problems from Nvidia can be frustrating, but switching to an older driver might improve your GPU performance. If your games need the latest drivers (like Rainbow Six and CS2), you have limited options. For testing older drivers, consider trying them before the Series 5000 release. I’m still using the November 2024 version for my 3060ti and haven’t encountered any issues so far. Hope this helps.
VRAM performance lasted over 20 minutes, but extending the test to at least an hour would help confirm reliability. Some issues only emerge after prolonged use, especially with different chargers and audio drivers disabled. The problem seems tied to BIOS/firmware settings, particularly on laptops where vendor configurations can affect power delivery. Next steps include updating ASUS support firmware and Nvidia drivers via DDU. Comparing stability between Nvidia Studio and Game Ready drivers may reveal differences. If instability persists above 700 MHz, using Nvidia Inspector to adjust voltage/frequency could stabilize the GPU. If that fails, the hardware may be defective and an RMA might be necessary.
The test completed successfully. It seems the BIOS may have been updated via Windows updates. I’ve used DDU and Nvidia Studio drivers before. Since I’m not very experienced with the Nvidia profile inspector, I can adjust the curve in MSTG afterburner. What voltage/frequency settings do you recommend trying?
I also attempted the Nvidia 566.14 drivers, but it didn't resolve the issue. Worth a try.
Secure the core at 700–900 MHz with a voltage range of 0.75–0.8 V. In Afterburner, use the curve editor (Ctrl+F) to select points above your target voltage and adjust them to flatten the line. Apply the settings and test with OCCT or a known crash-prone game. If it remains stable, incrementally increase by 50 MHz until failure occurs—this defines the silicon’s maximum safe frequency. This approach combines undervoltage with a frequency limit, sacrificing some performance for reliability. If crashes persist even at lower settings, it likely points to a silicon defect or an unresolved RMA issue.