Your GPU’s core temperature is one of the most important gaming performance stats you can track. When a card overheats, it automatically throttles its clock speeds — dropping frames and causing stutters — before you see a single crash. Monitoring it takes five minutes to set up and costs nothing. Here are four methods, from a zero-install Windows check to a full diagnostic overlay.
Method 1: Windows Task Manager (No Software Required)
Windows 10 version 1909 and later include GPU temperature monitoring directly inside Task Manager — no third-party tools needed. Press Ctrl + Shift + Esc to open Task Manager, click the Performance tab, then scroll down the left panel to GPU 0. Temperature appears at the bottom of the GPU panel in Celsius.
This works only with dedicated GPUs and WDDM 2.4-compatible drivers, which every modern NVIDIA and AMD card meets. The limitation: Task Manager shows the current temperature only — no logging, no history, no in-game overlay. It’s the right call for a quick pre-session sanity check, not live monitoring mid-game.
Method 2: MSI Afterburner + RTSS Overlay (Best for Live Gaming)
MSI Afterburner is the standard choice for monitoring GPU temperature while gaming. Its on-screen display (OSD) — powered by RivaTuner Statistics Server (RTSS) — draws temperature, framerate, and GPU usage figures directly onto your game. No alt-tabbing required.
Setup takes about five minutes:
- Download MSI Afterburner from MSI’s website. During installation, keep RivaTuner Statistics Server checked — the overlay won’t work without it.
- Open Afterburner and click the gear icon to open Settings.
- Go to the Monitoring tab. Find GPU Temperature in the sensor list, click it, then check Show in On-Screen Display in the Properties panel at the bottom.
- Navigate to the On Screen Display tab and assign a toggle hotkey — Scroll Lock works well.
- Click OK, launch your game, and press the hotkey to show the overlay.
You can stack multiple sensors into one compact overlay — add GPU usage, VRAM usage, and framerate alongside temperature. Font size, position, and color are all customisable.
Pro tip: Also enable GPU Memory Junction Temperature if you’re on an RTX 30/40-series or RX 6000/7000 card. VRAM overheating causes more unexplained crashes and artifacting than core temperature issues do, and most players never think to check it.
Method 3: HWiNFO64 (Most Detailed Diagnostics)
HWiNFO64 reads every thermal sensor on your GPU simultaneously: core temp, hotspot temp, memory junction temp, power draw, fan RPM, and voltage. It’s the right tool when you suspect a specific thermal problem and need data to confirm it.
Download it from hwinfo.com, run it in Sensors-Only mode, and scroll to the GPU section. You’ll typically see three key readings:
- GPU Temperature — the averaged reading across shader processors (the number most tools report)
- GPU Hot Spot — the single hottest measured point on the die, always higher than core
- GPU Memory Junction — critical on GDDR6X cards; RTX 3080 and 3090 routinely hit 100–105°C here by design
A hotspot reading 10–20°C above core temperature is normal. A gap above 25°C on a card in a well-ventilated case can point to dried thermal paste or poor airflow — worth investigating before assuming the card is faulty.
For in-game display, you can link HWiNFO64 to the MSI Afterburner OSD using the HWiNFO plugin, giving you HWiNFO’s sensor depth with Afterburner’s overlay convenience.
Method 4: NVIDIA App Overlay
If you’re on an NVIDIA GPU and prefer not to install third-party tools, the NVIDIA App (which replaced GeForce Experience in 2024) includes a built-in performance overlay with GPU temperature monitoring.
- Open the NVIDIA App, go to Settings > Features, and enable In-Game Overlay.
- In a game, press Alt+Z to open the overlay panel, click Statistics, then Configure HUD to add GPU temperature to the display.
- Press Alt+R to toggle the stats HUD on and off while gaming.
The overlay shows GPU temperature, usage, framerate, latency, and clock speeds in one panel. AMD users have an equivalent through AMD Software: Adrenalin Edition — open its overlay with Alt+Z from the system tray.
GPU Temperature Ranges: What the Numbers Mean
Most guides list temperature numbers without explaining when to actually act. Here’s what each range means in practice:
| Temperature (Core) | Status | What It Means |
|---|---|---|
| 30–50°C | Idle — Normal | Desktop use; fans may be fully stopped on modern cards |
| 60–75°C | Gaming — Excellent | Full load with ample thermal headroom |
| 75–85°C | Gaming — Normal | Within spec for most cards; nothing to fix |
| 85–90°C | Gaming — Warm | Check case airflow; sustained 88–89°C warrants attention |
| 90–95°C | Throttling Zone | Card is reducing clock speeds; expect lower FPS |
| 95°C+ | Investigate Now | Active cooling problem — act before long-term damage |
Brand ceilings differ. NVIDIA GeForce RTX 30-series cards begin throttling around 93°C core. AMD RDNA 2 and RDNA 3 cards are officially rated to 100°C edge temperature, but edge temps consistently above 90°C in a standard tower case suggest a real cooling problem rather than normal operation.
One figure most guides skip: the hotspot temperature. AMD and recent NVIDIA cards expose a hotspot reading that’s always higher than the core average — it measures the single hottest sensor on the die. On AMD RDNA 3 hardware, a hotspot of 100–110°C while core sits at 75–80°C is within spec by design. It’s not a warning sign — it’s how the boost algorithm works.
What to Do If Your GPU Is Running Too Hot
If core temperature consistently sits above 85°C during normal gaming sessions, work through this list in order before spending money on cooling upgrades:
- Check case airflow: Confirm front fans are pulling cool air in and rear/top fans are exhausting hot air out. Cable management blocking front intake is a surprisingly common cause of 5–10°C spikes.
- Clean the GPU: Dust buildup on heatsink fins can raise temps by 10–20°C in severe cases. Compressed air through the heatsink with the PC horizontal takes under two minutes.
- Adjust the fan curve: In MSI Afterburner’s Fan tab, create a custom curve that ramps above 70% fan speed at 75°C, rather than waiting for the GPU’s conservative factory curve to catch up.
- Repaste the GPU: Thermal paste loses conductivity after 3–5 years. Replacing it typically drops temperatures by 5–15°C. This requires opening the GPU cooler, which voids most warranties on cards still in the warranty period.
For a full look at settings that reduce GPU workload — and how resolution scaling, shadow quality, and ray tracing each contribute to thermal load — see our PC Game Optimization Guide. For a breakdown of what each graphics setting actually does, Game Settings Explained covers every major option.
Frequently Asked Questions
What is a safe GPU temperature while gaming?
60–85°C core is the target range for most desktop GPUs. Cards sitting at 90°C+ while gaming are thermally throttling and losing performance.
Does a higher GPU temperature mean lower FPS?
Yes. Once a GPU reaches its thermal limit, it automatically reduces clock speeds — called thermal throttling — which directly lowers framerate. You may not notice a hard crash, just gradual FPS drops during demanding scenes.
Can I monitor GPU temperature without installing any software?
Yes. Windows 10 version 1909 and Windows 11 both show GPU temperature on the Task Manager Performance tab with no additional software required.
Why is my hotspot temperature so much higher than the core?
The hotspot measures the single hottest sensor on the die; the core temp is an average of all shader processors. A 10–25°C gap is normal. On AMD RDNA 3 cards, hotspot readings of 100°C+ alongside a core of 75–80°C are within specification by design.