Quick answer
Restore your RAM to stock JEDEC speeds (turn off XMP/DOCP), reseat the GPU in a different PCIe slot, and reseat the RAM sticks. If the artifacting stops, you found the culprit — it's not the GPU, it's the motherboard or RAM setup.
Why this happens
When you see colored squares, flickering lines, or weird shapes on screen, your first instinct is „GPU is dying.“ And sometimes it is. But what's actually happening here is often a memory or bus problem. The GPU uses your system RAM indirectly through the PCIe bus, and the VRAM on the card itself. Both can get corrupted if the motherboard's memory controller or PCIe lane is unstable. Real-world trigger: you just installed new RAM, enabled XMP, or swapped a GPU. The board can't handle the higher memory frequency and starts sending bad data to the GPU. The driver then reports artifacting.
Fix steps
- Reset RAM to stock speed. Enter BIOS. Disable XMP (Intel) or DOCP (AMD). Set memory frequency to the default 2133 MHz or 2400 MHz. Reboot and test. If artifacting stops, your RAM speed or timings were the issue. The reason step 3 works is that overtight timings or high frequency past the CPU's memory controller limit causes data on the memory bus to flip bits — and those bits end up in the GPU's frame buffer.
- Reseat the GPU in a different PCIe slot. Power off. Take out your GPU. Put it in the second PCIe x16 slot (if your board has one). Boot. If artifacting goes away, the first slot's electrical connection is dirty, damaged, or the slot itself has a bent pin. This is more common than you'd think, especially on boards that had heavy GPU sag for years.
- Reseat both RAM sticks. Take them out, blow on the slots (carefully), reinsert until they click. Boot again. This fixes intermittent contact — oxidation on gold contacts builds up over 6 months of sitting in a dusty case.
- Update motherboard BIOS. Go to your board manufacturer's site. Download the latest BIOS for your exact revision. Install via USB stick in BIOS flash mode. Updated memory reference code often fixes stability with certain RAM kits.
Alternative fixes if main steps fail
Test with a single RAM stick. Remove one stick. Try each slot. Some boards have a bad memory channel that only shows up under load — artifacting is a clear symptom. You'll see the issue only when both channels are active.
Lower memory voltage slightly. If you have high-end RAM at 1.45V, try 1.35V. Too much voltage can cause signal reflection on the memory bus.
Check your PSU. A dying power supply can't keep the 12V rail clean, especially under GPU load. If you have a multimeter, check the 12V pin on the GPU power connector — it should be between 11.8V and 12.2V under load. Below that, you're looking at random artifacting.
Prevention tip
Stick to RAM that's on your motherboard's QVL (Qualified Vendor List). Not all DDR5 or DDR4 kits play nice with every board. Buy a kit that's explicitly listed for your model. Also, never tighten timings beyond what your CPU's memory controller can handle — a Ryzen 5 5600 won't run DDR4-4000 stable no matter what voltage you feed it. For PCIe slots, use a GPU support bracket to prevent sag and slot damage over time.