Pick a tool by the job you need
A PC has many temperature sensors, and the right one depends on your question. Is the processor slowing down in a long game? Is case airflow poor? Is the graphics card running hot? One free tool can answer most of these, but you need to know which reading to check.
Keep it simple. Read the CPU and GPU sensors under real load, check the motherboard sensor for airflow problems, and log the results so you can compare them later.
Free monitoring tools
| Tool | Best for | Notes |
|---|---|---|
| HWiNFO64 | Almost everything: CPU, GPU, VRMs, chipset, drives and fans | Use Sensors-only mode. The free version limits shared-memory export, which overlays and panel apps use, to 12 hours per session. Pro removes that limit. |
| HWMonitor | A simple view of CPU, GPU, voltages and fans | Made by CPUID |
| Core Temp | Per-core CPU temperatures | CPU only |
| GPU-Z | GPU sensors and the reason the card is limiting its clocks | GPU only |
| MSI Afterburner with RivaTuner Statistics Server | In-game overlays | RTSS installs with Afterburner |
| NVIDIA App and AMD Adrenalin overlays | Quick on-screen GPU readouts | Alt+R for NVIDIA, Ctrl+Shift+O for AMD |
On Linux, lm-sensors (the sensors command) and psensor read board and CPU sensors, and nvidia-smi reports NVIDIA GPU temperatures.
Motherboard vendors ship their own utilities that also read board sensors. ASUS, Gigabyte and ASRock temperature monitoring tools compares them with HWiNFO64.
Which sensor reads what
- CPU: Intel chips report a “CPU Package” reading plus per-core readings. AMD Ryzen chips report “CPU (Tctl/Tdie)”.
- GPU: The edge (main) temperature is the number most tools and spec sheets quote. The hotspot (junction) is the hottest point on the die. Some cards also report memory junction temperature. GPU hotspot vs edge temperature explains how to read the gap between them.
- Motherboard: A sensor from the board’s Super I/O chip usually reports “Motherboard” or “System”. Other sensors may cover the CPU socket area, the VRM (power delivery) and the chipset. Labels vary by board and are often unhelpful, so check the board manual.
- Drives: NVMe SSDs should stay under about 70°C. Many throttle somewhere around 70–80°C, depending on the controller.
Watching temperatures while gaming
To see temperatures on screen in a game with MSI Afterburner:
- Install MSI Afterburner and make sure RivaTuner Statistics Server (RTSS) is included in the setup.
- Open Afterburner’s settings, go to the Monitoring tab, and select the CPU and GPU temperature sensors.
- For each sensor, tick Show in On-Screen Display.
- Keep Afterburner and RTSS running, then launch the game.
NVIDIA’s overlay (Alt+R) and AMD’s Adrenalin overlay (Ctrl+Shift+O) are lighter options if you only need GPU numbers.
For a long session, HWiNFO64 in Sensors-only mode can log readings to a CSV file. Watch clock speeds as well as temperatures, because a high number alone does not prove throttling. The thermal throttling guide explains how to tell the difference.
Avoid running several monitoring tools at once. Their sensor polling can conflict and cause stutter, and vendor apps can clash with third-party readers.
Motherboard and external sensors
Most boards have several temperature sensors, but few are labelled clearly. Use the board manual, or HWiNFO64, to find the sensor for the VRM or chipset, and watch it under load. Motherboard temperature is not CPU temperature, and a hot heatsink on the board is not automatically a fault. The motherboard temperature guide covers normal ranges.
Some mid-range and high-end boards have an external sensor header, often labelled T_SENSOR or TEMP. It takes a 2-pin probe, usually a 10K NTC thermistor, that you can place where you want to measure. Many boards’ BIOS fan pages can also use that probe as a fan-control source. Power off before installing a probe, and keep its lead away from fan blades.
Front-panel and desk temperature displays
If you want temperatures visible without opening a window, these options exist:
- AIO cooler LCDs: Some liquid cooler pump heads have an LCD that shows CPU temperature, such as the NZXT Kraken Elite or Corsair iCUE LINK LCD models. Check that the exact cooler model supports the display.
- Digital-readout air coolers: A few air coolers, including some Thermalright “Digital” models, show temperature on the cooler itself.
- Small USB screens: Compact IPS displays can show sensor data through software such as AIDA64 SensorPanel.
Check compatibility and required software before you buy any of these.
Safe temperatures at a glance
These ranges are typical, not limits. Check your part’s spec sheet for its exact maximum.
| Component | Idle | Under load | Reference limit |
|---|---|---|---|
| CPU | About 30–50°C | About 60–85°C in games | Varies by chip. See what TjMax means. |
| GPU (edge) | About 30–50°C | About 65–85°C | Card-specific. Many NVIDIA cards list around 90°C. |
| GPU hotspot | Close to edge | Usually 10–25°C above edge | Up to about 100°C can be acceptable. AMD lists 110°C. |
| Motherboard / System sensor | About 25–45°C | Around 50–55°C | Not a CPU limit |
| VRM | Not usually a concern | Up to about 90°C under load is often fine | Varies by board |
| NVMe SSD | Well below 70°C | Keep under about 70°C | Throttling starts around 70–80°C, depending on the drive |
For a fuller explanation of CPU and GPU numbers, see how hot is too hot for a computer.
Mistakes that make readings look wrong
- Comparing different sensors. Apps a few degrees apart often read different sensors or poll at different times. A small gap does not mean one tool is broken.
- Confusing board and CPU temperature. A “motherboard temp” reading is not the CPU.
- Reading normal spikes as faults. Some Ryzen chips show brief idle spikes of 10–15°C as boost kicks in. That alone is not a cooling problem.
- Trusting one extreme value. Readings of 120°C or more, with a cool heatsink and a stable system, usually point to a faulty or misread sensor. Check with a second tool.
- Judging by a short peak. A single spike, especially during a loading screen, tells you little. Look at averages over a sustained session.