If your graphics card starts fast and then slows down after 10–20 minutes of gaming, thermal throttling is the likely culprit. To protect itself, a GPU automatically reduces its clock speed — and therefore your frame rate — when it gets too hot. Understanding how it works helps you tell the difference between normal behavior and a cooling problem worth fixing.
How throttling works
Modern GPUs boost their clocks opportunistically based on temperature, power, and voltage headroom. As the core heats up, the card steps clocks down to stay within a safe thermal limit — often somewhere around 83–90 °C for the core, with a lower limit for hotspot and memory junction temperatures. A little clock variation is completely normal; a large sustained drop that tanks your FPS is throttling worth addressing.
Normal vs problematic
A card that dips its boost clock slightly while staying near its rated performance is fine. A card that falls well below its base behavior, drops frames, and pins its temperature at the limit is throttling hard.
| Temperature (core) | What it usually means |
|---|---|
| Under 70 °C | Excellent cooling, full boost |
| 70–80 °C | Normal for many gaming GPUs |
| 80–85 °C | Warm; near throttle point on some cards |
| 85–90 °C+ | Likely throttling; investigate |
| Hotspot 100 °C+ | Poor paste/pad contact; act |
How to spot it
Monitor while gaming
Run a free monitoring overlay (many GPU utilities and hardware monitors do this) and watch core temperature, hotspot temperature, and core clock during a demanding game. The tell-tale sign of throttling is a clock speed that steadily falls as the temperature climbs and flatlines at the limit. If FPS drops in lockstep with rising temps, you’ve confirmed it.
Check the log after a session
Tools that log min/max/average clocks and temps let you review a whole session. A big gap between peak clock (first minute) and sustained clock (after 20 minutes), combined with temps parked at the ceiling, points squarely at thermal throttling.
What to do about it
Common fixes, cheapest first: improve case airflow (add intake/exhaust fans, clean dust filters), raise the GPU fan curve for more cooling at the cost of noise, undervolt the GPU to cut heat with minimal performance loss, or — for older cards with dried-out paste and pads — repaste and replace thermal pads. A vertical GPU mount too close to a glass panel can also choke airflow.
FAQ
Is thermal throttling bad for my GPU?
Throttling itself is a protective feature, not damage — it exists to keep the card safe. But if your GPU throttles frequently, it means it’s running hot, and sustained high temperatures shorten component life over the long term. It’s better to fix the cooling than to let the card constantly ride its thermal limit.
What temperature is too hot for a gaming GPU?
It varies by model, but core temps in the 70s °C are healthy, low-to-mid 80s are acceptable on many cards, and sustained upper-80s or a hotspot above ~100 °C signal a cooling issue. Check your specific card’s rated limits, since some are designed to run hotter than others.
Throttling vs other slowdowns
Not every mid-game slowdown is thermal throttling, and misdiagnosing it wastes effort. If your clocks stay high but FPS drops, the cause is more likely a CPU bottleneck, a game-engine issue, or a background process — not heat. Thermal throttling has a specific signature: temperature climbs to the limit and clock speed falls in response. There’s also power throttling, where the card hits its power limit and reduces clocks even while cool; monitoring tools often label the active limit (thermal, power, voltage, or utilization) so you can tell them apart.
Undervolting: the enthusiast’s fix
For a card that runs hot but is otherwise healthy, undervolting is the most elegant solution. By lowering the voltage the GPU requests at a given clock speed, you cut heat and power draw — often 10–20 °C cooler and quieter — with little or no performance loss, and sometimes a small gain because the card no longer throttles. It takes some testing to find a stable voltage-frequency curve, but the payoff is a cooler, quieter card that holds its boost clocks longer. If undervolting sounds intimidating, simply raising the fan curve or improving case intake and exhaust airflow solves most throttling without any tuning at all.
Bottom line
Thermal throttling is your GPU protecting itself by slowing down when it overheats. Spot it by watching clocks fall and temps pin at the limit during long sessions. A little clock variation is normal; a big sustained drop means it’s time to improve airflow, raise the fan curve, undervolt, or refresh the thermal paste.
