How to Check if a Used GPU is Healthy
Fans spin freely
Give each fan blade a gentle spin with your finger. They should rotate smoothly with no grinding or resistance.
No burn marks or corrosion
Check the PCB (circuit board) and connectors for any black scorch marks, rust, or white residue from liquid damage.
Non-original cooler
Aftermarket coolers aren’t necessarily bad, but they suggest the card has had work done. Ask why the cooler was replaced.
Bent or damaged PCIe connector
A bent power connector means the card was likely removed or installed carelessly. Walk away — connector damage can cause instability or fire risk.
FurMark
Stress tests the GPU under full load. Watch for artifacting — random coloured pixels or screen glitches — which indicate VRAM or shader issues.
FreeOCCT
Runs a VRAM-specific test that catches bad memory cells. Mining cards often fail here. Run for at least 15 minutes.
FreeGPU-Z + HWiNFO
Shows real-time GPU temperature, fan speeds, and clock speeds. Under load, most GPUs should stay below 85°C. Above 95°C is a red flag.
Free3DMark Time Spy
A gaming-realistic benchmark. Compare the score to the same GPU model online — a significantly lower score suggests throttling or damage.
Free tierArtifacting under load
Random coloured pixels, screen flickering, or corrupted textures during FurMark or a game. This usually means VRAM is damaged — often from mining. Not fixable.
Temperatures above 95°C at stock settings
GPU running this hot means the thermal paste is dry, the heatsink is clogged, or the cooler is faulty. Fixable but factor in the cost of a repaste.
Seller won’t let you test it
Any legitimate seller will let you run a benchmark before paying. If they refuse or rush you, assume something is wrong.
Vague answers about mining history
“I only used it for gaming” is the most common lie in second-hand GPU sales. Ask for proof — original purchase receipt, photos of the rig it was in.
No display output at all
Could be a dead card, a faulty port, or a driver issue. Don’t pay full price for a card that won’t display until you’ve diagnosed the cause.
Passes all tests
Clean visuals, temps under 85°C, no artifacting, VRAM healthy. Pay fair market price.
Minor issues
High temps, noisy fans, non-original cooler. Factor in S$30–60 for a repaste and negotiate down.
Artifacting or VRAM errors
VRAM damage is permanent. No price is worth a card that will fail in months.
Don’t want to test it yourself?
Bring any used GPU to Respawn.sg and we’ll run a full health check — display output, VRAM, thermals, and fans. Free assessment, no obligation.