Cooling & ThermalsIntermediate8 min read · Updated 2025-11-20

Cooling & Thermals for Stable FPS

The most common cause of FPS that starts high and slowly decays isn't a bad game or an old GPU — it's heat. When a CPU or GPU gets too hot it deliberately slows itself down to stay safe, and that quietly costs you frames and smoothness. The good news is that the fix is almost always about getting heat out of the box, and every safe step is reversible and leaves the chip's built-in protection fully intact. This guide covers how throttling works, what temperatures are normal, and how to cool a desktop or laptop properly. Repaste and undervolt are included at the end as optional advanced tuning — with clear caveats — never as required steps.

How heat turns into lost FPS

Every modern CPU and GPU has a target temperature it won't exceed. As it approaches that limit, it lowers its clock speed — and therefore its performance — to reduce heat output. This is thermal throttling, and it's a protective feature working as designed, not a fault. The symptom is distinctive: performance is fine for the first few minutes of a session and then FPS gradually sinks as the components heat up and clocks step down.

Because throttling is the chip protecting itself, the correct response is to remove the heat, never to remove the safety limit. Improving airflow, clearing dust and keeping the intake unobstructed let the hardware hold its full clocks for longer, which restores the frames you were losing. Any tool or tweak that claims to 'unlock' more performance by disabling thermal protection is dangerous and should be avoided outright.

What temperatures are actually normal

It's easy to panic at a number that's completely fine. Gaming hardware is built to run hot: sustained GPU temperatures in the roughly 70–85 °C range under load are normal and safe for many cards, and gaming CPUs frequently spike into the 80s °C during heavy scenes without any problem. A warm chip is not a damaged chip — these parts are rated for it.

What matters is the pattern, not a single reading. Concern is warranted when clocks are visibly dropping while temperatures sit in the high 80s or 90s °C and stay there, when fans are roaring constantly, or when the machine shuts down to protect itself. Read temperatures with your GPU app or a reputable hardware monitor, watch them alongside clock speed, and judge whether cooling is actually limiting you before changing anything.

  1. 1Install or open a monitoring overlay (NVIDIA app / AMD Adrenalin or a trusted hardware monitor) and show temperature plus clock speed in-game.
  2. 2Play a demanding scene for 15–20 minutes and watch the trend, not just the peak.
  3. 3Treat sustained high-80s/90s °C with falling clocks as the signal to improve cooling; a steady 70–85 °C is generally fine.
  4. 4If the PC ever shuts off under load, stop and address cooling before continuing — that's the last-resort protection triggering.

Desktop airflow and fan setup

A desktop cools well when cool air flows in the front/bottom and warm air leaves the back/top in a clear path, with the components in between. Most thermal problems in a working PC come from blocked or poorly arranged airflow rather than an inadequate cooler: a case shoved against a wall, cables strangling the airflow, or fans set up so intake and exhaust fight each other. Fixing airflow is free and completely reversible.

You can also let the fans do more work through the vendor's own software. Raising the fan curve so fans spin up sooner trades a little noise for lower temperatures and is entirely safe — you're asking the cooler to work harder, not touching any limit. None of this requires opening the chip's safety settings.

  1. 1Give the case room to breathe — a few centimetres of clearance around intake and exhaust, and off carpet where possible.
  2. 2Confirm fan direction: intakes at the front/bottom pulling cool air in, exhausts at the rear/top pushing warm air out.
  3. 3Tidy cables out of the main airflow path so air can move front-to-back unobstructed.
  4. 4In your GPU app, raise the fan curve slightly so fans ramp earlier (louder but cooler) — fully reversible.
  5. 5Make sure case dust filters are present and clean so intakes aren't choked.

Laptop cooling

Laptops are far more thermally constrained than desktops, and heat is often the reason a gaming laptop can't hold its advertised performance. Two things dominate: keeping the air intakes (usually on the underside) clear, and keeping the machine plugged in. On battery, laptops deliberately throttle both CPU and GPU hard to save power, which looks exactly like thermal throttling but is a separate power cap — so plug in before you blame the cooling.

Because the intake sits underneath, soft surfaces like beds and couches suffocate it and cause rapid overheating. A hard, flat surface, a stand that lifts the rear, or a cooling pad that pushes air at the underside can all lower temperatures noticeably. Many gaming laptops also have vendor performance modes that adjust fan behaviour — worth enabling while plugged in.

  1. 1Keep the laptop plugged in while gaming — battery power caps CPU and GPU clocks by design.
  2. 2Use it on a hard, flat surface; never game with it on a bed, blanket or lap where the underside intake is blocked.
  3. 3Add a laptop stand or cooling pad to lift the chassis and feed cool air to the intake.
  4. 4Enable the vendor's 'Performance'/'Turbo' fan profile (Armoury Crate, Lenovo Vantage, MyASUS, etc.) while plugged in.
  5. 5Keep the rear/side exhaust vents unobstructed and clear of dust.

Dust cleaning, and optional advanced steps

Dust is the slow, steady killer of cooling: it clogs filters, fan blades and heatsink fins, and a badly caked laptop vent can add 10–20 °C over time. Regular, gentle cleaning is the highest-value maintenance you can do and carries no risk if done with the machine powered off. Use short bursts of compressed air and hold fans still so they don't spin; never use a vacuum's suction against delicate internals.

Two further steps can help older or thermally stubborn machines, but they are optional and carry real caveats. Repasting — replacing the dried thermal paste between the chip and its cooler — can drop temperatures on an aging PC, but it means opening the hardware and may affect a warranty, so it's best left to a confident builder or a professional. Undervolting — asking the CPU or GPU to hit the same clocks at slightly lower voltage using the vendor's own tools — can lower temperatures and even improve sustained clocks, but it must be tested carefully for stability and is fully reversible. Neither is necessary for good performance; treat them as fine-tuning, not requirements, and never pair them with anything that disables thermal or security protection.

  1. 1Power the machine off and unplug it before any cleaning.
  2. 2Use short bursts of compressed air on fans, filters and vents; hold fan blades still so they don't over-spin.
  3. 3Clean or replace case dust filters, and clear laptop underside vents where dust builds up fastest.
  4. 4OPTIONAL (advanced): consider a repaste only on an older machine, mindful of warranty — a job for a confident builder or a pro.
  5. 5OPTIONAL (advanced): try a mild undervolt with the vendor's tool, test thoroughly for stability, and revert immediately if you see crashes — never disable safety limits.

Key Takeaways

  • Thermal throttling is the chip protecting itself by lowering clocks — the fix is removing heat, never disabling the limit.
  • Sustained ~70–85 °C is normal; worry only when clocks fall while temps sit in the high 80s/90s and stay there.
  • Most desktop overheating is blocked airflow — give the case clearance, set intake/exhaust correctly, and raise the fan curve.
  • For laptops, plug in, use a hard surface or cooling pad, and enable the vendor performance profile.
  • Clean dust regularly (powered off); treat repaste and undervolt as optional, caveat-heavy advanced tuning, not requirements.