Keeping an eye on your graphics card’s temperature while gaming is one of the simplest ways to catch overheating, thermal throttling, or a failing cooler before they cause real damage or ruin your frame rate. You don’t need to spend anything to do this properly: a combination of free tools can show you real-time temperature, clocks, and load directly on your screen while you play, plus deeper sensor logs if you need to diagnose a problem. Here’s exactly how to set it up and what the numbers actually mean.
The easiest method: MSI Afterburner with RivaTuner Statistics Server
MSI Afterburner is the most widely used free GPU monitoring and overclocking tool, and it works with graphics cards from any brand, not just MSI. It’s free to download from MSI’s website and installs alongside RivaTuner Statistics Server (RTSS), which is the companion tool that actually draws the on-screen display (OSD) overlay during gameplay; Afterburner handles monitoring and control, while RTSS renders that data on top of your game.
To set it up: install both Afterburner and RTSS (they usually come bundled in the same installer), open Afterburner, click the “Settings” gear icon, go to the “Monitoring” tab, and in the list of sensors find “GPU temperature.” Check the box under “Show in On-Screen Display” for that sensor, and repeat for anything else you want visible, like GPU usage, clock speed, or frame rate. Close the settings window, launch your game, and use the default hotkey (usually Ctrl+F to toggle it) to show or hide the overlay. You can also reposition the overlay, change its font size, and add a graph rather than just numbers, all from the same Monitoring tab. As of early 2026, Afterburner has continued receiving updates, including beta builds adding explicit support for newer NVIDIA RTX 50-series cards.
HWiNFO: deeper sensor data and logging
HWiNFO is a free system monitoring tool that goes further than Afterburner’s overlay, exposing granular sensor data including core temperature, hotspot temperature, memory temperature, fan speeds, power draw, and per-adapter breakdowns on multi-GPU systems. It’s the better choice when you need to log temperatures over time to diagnose an intermittent problem, since it can export sensor data to a file for later review, and it supports advanced telemetry many other tools don’t expose, such as per-chip VRAM temperatures on newer AMD cards. HWiNFO isn’t built for a slick in-game overlay the way Afterburner is, but it can feed its sensor data into RTSS to appear in the same overlay, giving you the best of both: HWiNFO’s sensor depth with Afterburner’s on-screen display.
GPU-Z: quick checks and hardware identification
GPU-Z is a lightweight, free utility focused on showing your GPU’s specifications and real-time stats, including temperature, core and memory clock speed, GPU load, and fan speed, all in a compact window. It’s not designed to run as an overlay during gameplay; it’s better suited for a quick check before or after a session, confirming your card’s exact model and specs, or verifying that reported clock speeds match what you expect. It’s a good complement to Afterburner or HWiNFO rather than a replacement for either.
Comparison table
| Tool | Best for | In-game overlay | Logging | Notes |
|---|---|---|---|---|
| MSI Afterburner + RTSS | Real-time overlay while gaming | Yes | Basic | Works with any GPU brand, most popular option |
| HWiNFO | Deep sensor data and diagnostics | Via RTSS integration | Yes, detailed | Best for logging and troubleshooting |
| GPU-Z | Quick spec and temp checks | No | No | Lightweight, good for verification |
Safe GPU temperature ranges by type
Temperature targets vary by manufacturer and even by chip design, so “normal” isn’t one fixed number. As a general baseline, idle temperatures typically sit around 30-50°C, while gaming loads commonly push cards into the 65-85°C range, with most GPUs designed to operate safely up to around 90°C under sustained load. Staying under roughly 80°C is a reasonable target for long-term reliability and consistent boost clocks, though modern cards are engineered to run hotter than that without harm.
For NVIDIA’s RTX 40 and 50 series, the core temperature limit (TJ Max) is around 89°C, and these cards are built to boost aggressively up to roughly 83-84°C before clocks start backing off meaningfully. NVIDIA cards also report a separate “hotspot” temperature that typically runs 10-20°C above the core reading; this is normal and expected, not a fault.
For AMD’s RDNA 2, RDNA 3, and RDNA 4 cards, the junction (hotspot) temperature is designed to safely reach up to around 110°C, though the edge (core) temperature should realistically stay under about 90°C, with throttling typically kicking in somewhere in the 90-95°C core range. AMD’s hotspot-to-edge gap is larger than NVIDIA’s by design, often 20-35°C, so seeing a 100°C hotspot next to a 70°C edge temperature on an AMD card is normal rather than alarming.
Memory temperature matters too, particularly on high-end cards using GDDR6X, which can tolerate temperatures up to around 110°C, though keeping it under about 95°C is better for long-term lifespan.
What the different temperature readings mean
Modern GPUs report more than one temperature sensor, and mixing them up is a common source of false alarms. “Core” or “edge” temperature is the main GPU die reading most people are used to watching. “Hotspot” or “junction” temperature reflects the single hottest point on the die and always reads higher than the core, sometimes significantly so on AMD cards by design. “Memory” or “VRAM” temperature tracks the video memory chips separately and has its own, generally higher, safe ceiling. When troubleshooting, look at all three where available rather than just the core number, since a normal-looking core temperature can still mask an overheating hotspot or memory module.
What to do if your GPU is running too hot
If your temperatures are consistently above the safe thresholds for your GPU type, or if you’re seeing sudden frame rate drops that line up with temperature spikes (a sign of thermal throttling), start with the basics: check that your case has adequate airflow and that intake and exhaust fans aren’t clogged with dust, since dust buildup on the heatsink and fans is one of the most common causes of rising GPU temperatures over time. Reseat or clean the card if you can safely open your case, and confirm the fan curve in Afterburner is actually ramping up under load rather than staying passive too long; you can set a custom, more aggressive fan curve in Afterburner’s fan tab if needed. If temperatures remain high after cleaning and adjusting fan curves, check ambient room temperature, case placement, and whether the GPU has enough clearance from other components. As a last resort, high hotspot or memory temperatures specifically, especially with an unusually large gap from the core reading, can indicate degraded thermal paste or pads that may need reapplying, particularly on cards that are a few years old.
Conclusion
For most gamers, installing MSI Afterburner with RivaTuner Statistics Server is the fastest way to get a real-time temperature overlay while playing, and it works regardless of which brand made your graphics card. Add HWiNFO if you need deeper sensor logging to diagnose an intermittent issue, and keep GPU-Z on hand for quick spec checks. As a rule of thumb, treat sustained core temperatures above roughly 85-90°C, or hotspot readings well beyond the normal gap for your GPU brand, as a signal to check airflow, dust buildup, and fan curves before assuming something is seriously wrong.