How Hot Should My GPU Run While Gaming? Normal Temps Explained

In this guide (9 sections)

• The three temperatures your graphics card reports
• How to measure it properly
• 1. The fans don’t speed up as the card heats up
• 2. Dust is blocking the heatsink
• 3. The case is short of fresh air
• 4. The card is working harder than it needs to
• 5. The thermal paste has dried out
• 6. The cooler itself is faulty
• Real cases: what other users ran into

You check your graphics cardGPU (graphics card): The graphics card: the part that creates the picture you see on the monitor. More in our PC basics glossary →Close ✕’s temperature while gaming and the number looks high, maybe 75°C, or a reading called “hot spotGPU hot spot (junction temperature): The temperature of the hottest point on the graphics chip. It always reads higher than the normal "GPU temperature". More in our Graphics cards & display glossary →Close ✕” near 100°C. Don’t panic: graphics cards are designed to run much hotter than people expect, and they protect themselves long before damage is possible. What matters is reading the right number and knowing the few signs that mean the cooling needs attention.

The three temperatures your graphics card reports

Free tools like HWiNFO, GPU-Z or MSI Afterburner show several temperatures for the graphics card. Windows Task Manager (Performance → GPU) shows just the main one. They mean different things:

GPU temperatures while gaming: a rule of thumb

Typical readings after 15–20 minutes of a demanding game. Every card is different: your card's official limit always wins.

GPU temperature

also "GPU", "core" or "edge"

Fineup to ~80°C

Watch it80–85°C

Too hotabove ~85–90°C

Hot spot

also "junction" or "hot spot"

Fineup to ~95°C

Watch it95–105°C

Too hot105–110°C and up

Memory

also "memory junction" or "VRAM"

Fineup to ~90°C

Watch it90–100°C

Too hotabove ~100–105°C

The gap matters too: a hot spot 10–25°C above the GPU temperature is normal. A gap of 30°C or more, or one that keeps growing over the months, usually means the cooler isn't touching the chip properly.

• GPU temperature (also called “core” or “edge”) is an average across the chip. It’s the main number to watch.
• Hot spot is the single hottest point on the chip. It always reads higher. AMD calls it “junction temperature”, and has said that for some of its cards up to 110°C during gaming is “expected and within spec”.
• Memory temperatureMemory temperature (memory junction): The temperature of the graphics card's own memory chips, which sit around the main chip. More in our Graphics cards & display glossary →Close ✕ is the card’s own memory chips. They’re built to run hot, and readings in the 80s or even 90s°C under heavy load are common on some cards.

Your card’s official limit is the final word. NVIDIA lists a “maximum GPU temperature” on each card’s spec page, and both NVIDIA and AMD cards slow themselves down automatically, called thermal throttlingThermal throttling (slowing down from heat): When a chip slows itself down because it's too hot, to protect itself. More in our Cooling & noise glossary →Close ✕, if they get close to it.

At idle, on the desktop, 30–50°C is typical. Up to about 55°C is normal on cards that stop their cooling fans when cool (why your fans may not be spinning).

How to measure it properly

A single glance mid-game doesn’t tell you much. To get a reliable reading:

1. Open HWiNFO (sensors only) or GPU-Z, and leave it running. Our guide on how to monitor temps while gaming also shows how to see them on screen during the game.
2. Play a demanding game for 15–20 minutes, so the card and case reach their working temperature.
3. Look at the maximum value for each reading, and at the gap between GPU temperature and hot spot.

If everything sits in the “fine” range, you’re done. If not, work through the causes below. They’re ordered from free and easy to more involved.

What you see	Most likely cause	Where to look
The temperatures are high and the fans barely spin.	Fan settings, or a fan problem	Cause 1
The temperatures have crept up over months, and the fans are louder than they used to be.	Dust in the heatsinkHeatsink: A block of metal fins that pulls heat away from a chip. Fans blow air through it. More in our Cooling & noise glossary →Close ✕	Cause 2
The card runs much cooler with the side panel off.	Not enough airflow through the case	Cause 3
The card is hot even in light or old games.	The card is making more frames than you need	Cause 4
The hot spot is 30°C or more above the GPU temperature.	Thermal pasteThermal paste: A thin gray paste between a chip and its cooler that helps heat pass from one to the other. More in our Cooling & noise glossary →Close ✕ dried out or poor cooler contact	Cause 5
The hot spot is very high on a brand-new card.	A faulty cooler	Cause 6

1. The fans don’t speed up as the card heats up

If the temperature climbs but the fans stay slow or stopped, the card can’t get rid of its heat. Often it’s a fan setting saved in an app like MSI Afterburner, sometimes a driverDriver: The software that lets Windows talk to a piece of hardware, like the graphics card. More in our Graphics cards & display glossary →Close ✕ issue or a failed fan.

Fix: reset fan settings to default in every fan app you use, and watch whether the fans speed up as the card warms. Our guide on a GPU fan not spinning walks through every cause.

2. Dust is blocking the heatsink

Dust packs into the thin metal fins of the heatsink under the fans, so less air gets through. It happens slowly: temperatures and fan noise creep up over months, and you may not notice until games start to freeze briefly.

Fix: clean the card and the case dust filters. Our step-by-step guide to cleaning GPU fans shows how to do it without damaging anything.

3. The case is short of fresh air

A graphics card can only be as cool as the air it breathes. A case with a solid front, too few fans or cables in the way can trap hot air around the card.

Quick test: play for 15 minutes with the side panel off. If temperatures drop by several degrees, airflow is part of the problem. Our guide to a loud graphics card fan explains how to set up case fans.

4. The card is working harder than it needs to

If a game runs at 200 frames per second on a 60 Hz or 144 Hz monitor, the card is making frames you never see, and turning them into heat. Game menus without a limit can be the worst.

Fix: set a frame rate capFrame rate cap (FPS limit): A setting that stops the graphics card making more frames per second than you choose, so it works less hard. More in our Graphics cards & display glossary →Close ✕ a little below your monitor’s refresh rateRefresh rate (Hz): How many times per second the monitor updates the picture, for example 60 Hz or 144 Hz. More in our Graphics cards & display glossary →Close ✕, in the game or in the NVIDIA or AMD driver settings. For more, undervoltingUndervolting: Running a chip at slightly lower voltage so it makes less heat and noise, often with no loss of speed. More in our Cooling & noise glossary →Close ✕ can lower temperatures with little or no loss of speed; see step 8 of our loud fan guide.

5. The thermal paste has dried out

Between the chip and the cooler sits a thin layer of thermal paste that carries the heat across. After several years it can dry out or spread unevenly. The typical sign is a hot spot that runs much further above the GPU temperature than it used to, 30°C or more.

Fix: replacing the paste can bring temperatures down a lot, but opening the card can void the warranty. If the card is still covered, contact the maker first. If not, and you’re comfortable with small screws, follow a guide for your exact card model, because some cards also need new thermal padsThermal pads: Soft, squishy strips that carry heat from memory chips and other small parts to the cooler. More in our Cooling & noise glossary →Close ✕ of the right thickness.

6. The cooler itself is faulty

Occasionally a new card simply has a bad cooler. In early 2023, AMD confirmed that a limited number of its own reference RX 7900 XTX cards had a cooler problem causing unexpected throttling, and asked affected owners to contact AMD Support.

Fix: if a new card runs far hotter than reviews of the same model, or its hot spot hits about 110°C while the GPU temperature is moderate, contact the shop or maker for an RMARMA (warranty return): Sending a faulty part back to the shop or maker under warranty to be repaired or replaced. More in our PC basics glossary →Close ✕ (a warranty return) rather than opening it.

Real cases: what other users ran into

These are real reports from PC forums, summarized in our own words, with a link to each original thread.

Real caseZotac RTX 4070 Super TrinityTurned out to be normalA hot spot in the high 80s that was perfectly normalRead what happened

During demanding games and a stress test, the card ran at about 68–69°C, with the memory around 65°C and the hot spot at 86–87°C. The owner worried because reviews showed a hot spot nearer 79°C. Forum members said a gap of about 16°C is normal and that readings vary between samples of the same card. The owner compared more reviews and decided the card was fine.

Takeaway: a hot spot in the 80s or 90s isn’t a problem on its own. Look at the gap. See the three temperatures.

Source: TechPowerUp Forums

Real caseASUS TUF RTX 4070 TiSolvedA growing hot spot gap fixed by new paste, at the cost of the warrantyRead what happened

After about six months, the card ran at 72–73°C with the hot spot at 95–97°C, and the gap had grown from 16°C to over 22°C. The owner had good airflow and no dust. ASUS support said 95°C was normal and that opening the card would void the warranty. The owner repasted it anyway: temperatures fell to about 60°C, with a hot spot of 70°C, but as they put it, they now had no warranty.

Takeaway: new paste can work wonders on a growing gap, but check your warranty first. See cause 5.

Source: ASUS ZenTalk

Real caseSapphire Pulse RX 6700 XTTurned out to be normalThe maker said 100°C was normalRead what happened

The card ran at 68°C, with the hot spot between about 95 and 100°C, sometimes 102°C. The owner contacted Sapphire, which said these temperatures were normal for this card. The owner decided to keep using it as it was.

Takeaway: AMD cards in particular can show high hot spot readings by design. When in doubt, ask the card maker. See the three temperatures.

Source: TechPowerUp Forums

Real caseAMD Radeon RX 7900 XTX (reference)Solution not confirmed by the user110°C hot spot on a new card: a known cooler faultRead what happened

Owners of AMD’s own reference RX 7900 XTX reported the hot spot hitting about 110°C while the main GPU temperature stayed moderate, with the card slowing down. One owner who re-mounted, repasted and undervolted saw no change. AMD later confirmed a cooler problem on a limited number of these cards and told affected owners to contact support. The thread doesn’t record individual outcomes.

Takeaway: if a new card’s hot spot is extreme and home fixes don’t help, it’s a warranty case. See cause 6.

Source: TechPowerUp Forums

Frequently asked questions

Is 80°C too hot for a GPU?

No, not for most modern cards while gaming. It’s on the warm side, so it’s worth checking for dust and airflow, but it’s well within what graphics cards are designed for.

Is a 100°C hot spot dangerous?

Not on its own. The hot spot is the hottest single point of the chip and always reads higher. What matters is the gap: 10–25°C above the GPU temperature is normal. Much more than that suggests the cooler isn’t making good contact.

What temperature should my GPU be at idle?

Usually 30–50°C on the desktop. Cards that stop their cooling fans when cool often sit a little higher, up to about 55°C, which is normal.

Can high temperatures damage my graphics card?

Cards protect themselves by slowing down and, at worst, shutting off before damage occurs. But running very hot for years can shorten the life of fans and other parts, so it’s worth fixing poor cooling.

Why is my GPU hotter than it used to be?

Usually dust building up in the heatsink, a warmer room, or thermal paste drying out after several years. Cleaning the card is the first thing to try.

And how hot should my processor run?

Processors have different limits from graphics cards, and some modern ones are designed to reach 95°C under load. See what CPU temperature is safe while gaming and why a CPU can run hot at idle.