Hello everyone,
I am working on an ASUS ROG G513QY laptop with RX 6800M dGPU.
Originally, this laptop was completely non-functional. I replaced the motherboard (part number 6050A3254002-MB-Axx), and with the replacement board it started working fine.
For a long time, the dGPU worked perfectly stable. I played several hundred hours on very demanding games such as Cyberpunk 2077 without any crashes, instability, or detection issues. Performance was excellent and there were no warning signs of failure.
However, after some time, the replacement motherboard began to develop problems. Now, the dGPU sometimes disappears (not detected in BIOS, Windows, or Linux). And when it is detected and I launch a game, after a certain random period of time under load, the system will crash abruptly and reboot. After this reboot, the dGPU is missing again (not enumerated until another long EC reset).
I have also tried DDU (Display Driver Uninstaller) several times. DDU shows the RX 6800M entry, but as explained to me, this is only because of registry remnants, not because the hardware is really being detected.
---
What I have already checked:
- +NVVDD (GPU core) ≈ 0.94 V (present on all R15 coils near GPU)
- +FBVDDQ (VRAM) ≈ 1.3 V
- +PEX_VDD (PCIe interface) ≈ 0.9 V
- VBIOS EEPROM (25xx, 8-pin) is powered with 3.3 V (VCC, CS#, WP#, HOLD# all high)
- GPU gets warm when powered on.
- After a long EC reset (holding power button), sometimes the dGPU comes back and is detected again.
---
Symptoms:
- When missing, the dGPU is completely absent from Device Manager and from lspci under Linux.
- All GPU power rails are present and stable.
- When detected, the GPU can run games, but after some time under load, the system will randomly crash and reboot.
- After this reboot, the dGPU disappears again until another reset.
- This suggests a PCIe sequence issue (PERST# or REFCLK), or possibly instability in reset/clock circuitry, rather than a missing power rail.
---
My tools and limitations:
- I do not have an oscilloscope or thermal camera.
- I only have a multimeter, a hot air rework station, and basic electronics tools.
- So I cannot probe REFCLK waveforms directly, but I can measure DC rails, resistances, and replace small components (resistors, MOSFETs, buffers, etc.).
---
My questions:
1. On this board (6050A3254002-MB-Axx, G513QY), where exactly can I measure PEX_DGPU_RST (PERST#) easily (TP or PR reference)?
2. Could this be a common failure of the reset buffer / pull-up resistor keeping the GPU in reset?
3. Or is it more often the PCIe REFCLK buffer / ESD array on the REFCLK lines?
4. Any known weak components on this ASUS board that could cause the dGPU to be powered but invisible, or cause random GPU crashes/reboots under load followed by the GPU disappearing?
5. Am I going in the wrong direction with my assumptions, and could this problem be related to something else entirely? Has anyone experienced a similar issue (even on different laptop models) that could help point me in the right direction?
---
Planned checks:
- Measure PEX_DGPU_RST (expect ~0 V if held in reset, ~3.3 V if released).
- Measure NVVDD_PG / PWRGD near U67400 (GPU core VRM controller).
- Check 3.3 V supply of the PCIe REFCLK buffer.
- Inspect series PR resistors and ESD arrays on REFCLK and PCIe lanes.
---
Any help locating the right test points, confirming typical culprits on this platform, or sharing experience from similar GPU disappearing issues on other laptops would be greatly appreciated.
Thanks in advance!
