I'm troubleshooting an ASUS ROG Astral RTX 5080 OC that consistently shows PCIe 5.0 x8 under load instead of PCIe 5.0 x16. The system uses an ASUS TUF Gaming Z790-Plus WiFi motherboard with BIOS 1836, an Intel Core i9-14900KF, 32 GB of DDR5-6000 with XMP enabled, and a Samsung 990 Pro 2 TB PCIe 4.0 SSD. The SSD is installed in the primary M.2 slot, which the motherboard manual says should not reduce the main graphics slot to x8. I have updated the BIOS and NVIDIA drivers, cleared CMOS, run GPU-Z's render test for more than a minute, benchmarked with Steel Nomad, tested several games including Remnant II, and monitored the system with HWiNFO. Windows reports no WHEA errors, temperatures and power draw are normal, and the SSD performs normally. The GPU support bracket is also installed. Could this be caused by a BIOS or PCIe bifurcation setting, an incorrectly shared M.2 slot, bent LGA1700 socket pins, poor CPU contact, a faulty motherboard, or the graphics card itself? I'd also like to know whether this hardware combination is known to run at PCIe 5.0 x16 and whether a particular BIOS version resolves the issue.
3 Answers
Since the BIOS and drivers are current and the link remains at x8 during sustained load, the motherboard or CPU socket is worth investigating. With the system powered off, inspect the LGA1700 socket under strong light and magnification for shifted or bent pins, then reinstall the CPU and cooler with even, moderate pressure. In BIOS, check the System Agent PCIe configuration and bifurcation settings. Testing another GPU in the primary slot would also help: if a second card trains at x8, the problem is much more likely to be the board, socket, CPU contact, or firmware than the RTX 5080.
There may be lane sharing depending on which M.2 slot is populated, even if the slot layout is not obvious. Try moving the NVMe drive to another supported socket, or temporarily remove all secondary drives and boot from a live operating system before checking GPU-Z again. If the card still negotiates at x8 with the other M.2 slots empty, shared storage lanes are unlikely to be the explanation.
Another owner reported the same x8 behavior with an ASUS high-end board and an RTX 5080, even after trying the recommended M.2 placements and removing extra drives. The card still performed well, so the practical performance loss may be small, but the persistent link width is understandably concerning. If the board is still under warranty and socket, settings, and alternate-GPU tests do not identify the cause, an RMA would be reasonable.
The computer is currently being checked by a technician, so I’ll ask them to inspect the socket, verify the PCIe and bifurcation settings, and test another graphics card if possible. If another card also runs at x8, I’ll focus on the motherboard, CPU contact, or BIOS.