Detailed analysis
Your GPU might be running at half its bandwidth right now and you'd have no idea unless you checked. Open GPU-Z (free from techpowerup.com) and look at the "Bus Interface" field. If it says x8 or x4 instead of x16, your GPU is bandwidth-starved and you're losing 10-20% FPS in demanding games. The fix is usually moving the GPU to the correct motherboard slot.
What's actually happening
Your GPU communicates with your CPU through PCIe lanes — think of them as highway lanes for data. A full x16 connection gives your GPU 16 lanes of bandwidth. An x8 connection cuts that in half, and x4 cuts it to a quarter. Modern GPUs are designed to use all 16 lanes, and when they don't get them, they can't receive texture data, geometry, and shader instructions fast enough to keep up with demanding games.
The tricky part is that most motherboards have multiple full-length PCIe slots that physically look identical. They're all the same size, they all accept your GPU, and the system boots fine in any of them. But electrically, only one of those slots — almost always the top one closest to the CPU — is wired for the full x16 bandwidth direct from the CPU. The others run through the chipset at x8 or x4, or they share lanes with M.2 SSD slots. Your GPU works in those secondary slots, but it works slower, and nothing tells you about it.
The most common causes (in order of likelihood)
GPU installed in the wrong PCIe slot — This is the cause 80% of the time. Someone built the PC and put the GPU in the second or third slot for aesthetics or cable management, not realizing only the top slot provides full x16 bandwidth. Prebuilt systems sometimes ship this way from the factory.
BIOS PCIe lane bifurcation setting — Some motherboards have a BIOS option that splits the primary x16 slot into two x8 slots for multi-GPU configurations (SLI/CrossFire). If this is set to "x8/x8" instead of "x16" or "Auto," even the correct slot only provides half bandwidth. This setting is only useful if you're running two GPUs, which almost nobody does anymore.
M.2 NVMe SSD sharing lanes — On many mid-range motherboards, installing an NVMe SSD in certain M.2 slots steals PCIe lanes from the GPU slot. Your motherboard manual has a chart showing which M.2 slots share lanes with which PCIe slots. A Gen 4 NVMe in M2_2 might drop your primary GPU slot from x16 to x8.
Dirty or damaged PCIe slot contacts — Dust or physical damage to the gold contacts can prevent some lanes from making proper electrical contact. The system works but negotiates a lower lane width.
CPU PCIe lane limitations — Lower-end CPUs provide fewer PCIe lanes. If you populate all M.2 slots and expansion slots, something has to share. AMD Ryzen CPUs generally provide 24 lanes, but laptop CPUs and older models may offer less.
How to fix it
First, confirm the problem. Download GPU-Z from techpowerup.com (free, no install needed). Launch it and look at the "Bus Interface" line near the bottom of the main window. It shows two values: the maximum your GPU supports and what it's currently running at. For example, "PCIe 4.0 x16 @ x8 4.0" means your GPU supports x16 but is only getting x8 right now.
Important: GPU-Z may show a lower width at idle because NVIDIA and AMD GPUs dynamically reduce PCIe width to save power when not under load. Click the question mark button next to Bus Interface in GPU-Z to run a render test that forces the GPU to full load. The "@ x__" value should jump to x16 during the test. If it stays at x8 or x4 under load, you have a problem.
Check which slot your GPU is in. Power off your PC, open the side panel, and look at your GPU's position on the motherboard. On almost every ATX motherboard, the top full-length PCIe slot (closest to the CPU socket) is the primary x16 slot. If your GPU is in the second or third slot, move it to the top one. This requires removing the GPU (release the PCIe retention clip, unscrew the bracket, gently pull straight up), and reinstalling it in the top slot. Make sure the PCIe retention clip clicks when the card is fully seated.
Check your motherboard manual for lane sharing. Find the PDF manual on your motherboard manufacturer's website (search your exact board model). Look for a section on PCIe configuration or a lane allocation chart. It will show exactly which M.2 slots share lanes with which PCIe slots. If your top M.2 slot sharing diagram shows that M2_2 reduces PCIEX16_1 to x8, and you have an NVMe drive in M2_2, either move the NVMe to a different M.2 slot or accept the lane sharing.
Check your BIOS for PCIe lane settings. Restart into BIOS (press Delete during boot) and look under Advanced > PCIe Configuration or similar. Find any setting related to "PCIe Lane Configuration," "Link Width," or "Bifurcation." It should be set to "Auto" or "x16" for single-GPU setups. If it says "x8/x8" or "2x8," change it to Auto.
Clean the PCIe contacts. If the GPU is in the correct slot and BIOS settings are right, the issue might be physical. Power off, remove the GPU, and inspect the gold edge connector. If you see dark spots, oxidation, or dust, gently clean the contacts with a pencil eraser (yes, really — it removes oxidation from gold contacts) followed by isopropyl alcohol and a lint-free cloth. Also inspect the PCIe slot on the motherboard for debris. Reseat the GPU firmly.
After any change, reboot into Windows, open GPU-Z, and run the Bus Interface render test again. Confirm it now shows x16 under load.
Is this a hardware or software problem?
This is a configuration problem — your hardware is fine, it's just not set up optimally. Moving the GPU to the correct slot and checking BIOS settings is all it takes in most cases. The only scenario where this is a genuine hardware limitation is if your CPU doesn't provide enough PCIe lanes for your desired configuration (multiple NVMe drives plus a GPU all demanding full bandwidth). In that case, upgrading the CPU or motherboard is the only option, but that's rare for typical gaming builds. If you're not sure, Crashless can check your drivers, temps, VRAM, and 400+ known patterns automatically — just use the chat above.
Games commonly affected
The FPS impact of x8 vs x16 is most noticeable in GPU-bound games at high resolutions and high frame rates. At 4K, the difference is smaller (2-5%) because the GPU is the bottleneck. At 1080p with a powerful GPU targeting 200+ FPS, x8 can cost 15-20% performance. Games where it matters most: Cyberpunk 2077, Call of Duty: Warzone, Fortnite (at competitive settings), Counter-Strike 2, Apex Legends, and any game where you're pushing very high frame rates. Ray-traced games are also more bandwidth-sensitive.
Frequently asked questions
Q: GPU-Z shows x16 at idle but x8 under load. Isn't it supposed to be the other way around?
A: Something is wrong. Normal behavior is x1 or x8 at idle (power saving) and x16 under load. If it drops to x8 under load, you likely have a lane-sharing conflict with an M.2 SSD or a BIOS bifurcation setting. Check your manual's lane allocation chart.
Q: I only have one PCIe x16 slot on my Mini-ITX board. Can it still run at x8?
A: Yes — if you have an M.2 SSD that shares lanes with the PCIe slot, or if there's a BIOS setting limiting it. Mini-ITX boards are especially prone to lane sharing because of their compact design. Check your specific board's manual.
Q: Does PCIe generation matter more than lane width?
A: Both matter, but lane width has a bigger impact. PCIe 4.0 x8 provides the same bandwidth as PCIe 3.0 x16, so if you're stuck at x8 on a Gen 4 system, it's equivalent to full Gen 3 speed — not ideal but not catastrophic. But PCIe 3.0 x8 is genuinely bandwidth-starved for modern GPUs and you should definitely fix it.