CPU temperature is one of the most useful health metrics on any PC — it tells you whether your cooling is working, whether thermal paste needs replacing, and whether your case airflow is adequate. Checking it takes thirty seconds with the right tool. Here is what to use, what the numbers mean, and what to do if they are too high.
What changed in 2026
- Intel Core Ultra (Arrow Lake) and AMD Ryzen 9000-series have higher Tjmax limits than previous generations (up to 105°C for some Intel parts) — context matters when reading raw temperatures.
- Intel P-core vs E-core temperatures differ within the same package; per-core monitoring is more useful than a single package average.
- Laptop thermal compounds dry out faster as thinner chassis leave less thermal headroom — 3-year repaste cycles are realistic for gaming laptops.
- HWiNFO64 v8.x added better support for Arrow Lake and Zen 5 sensor layouts.
The right tools
| Platform |
Best tool |
Free? |
| Windows |
HWiNFO64 |
Yes |
| Windows (simpler) |
Core Temp |
Yes |
| Windows (gaming overlay) |
MSI Afterburner + RivaTuner |
Yes |
| macOS |
iStatMenus |
~$12 one-time |
| macOS (free) |
Hot (menu bar app) |
Yes |
| Linux |
sensors (lm-sensors) |
Yes |
| Any OS (BIOS) |
UEFI hardware monitor |
Built-in |
How to check in Windows: HWiNFO64
- Download from hwinfo.com — install or run portable.
- Click Sensors at the main screen.
- Scroll to the CPU section. Look for:
- CPU Package — overall package temp.
- CPU Core #X — individual core temps.
- CPU (Tctl/Tdie) on AMD — the actual die temp (subtract ~10°C offset on older Ryzen for Tdie vs Tctl).
- Let it run for a few minutes at idle, then run a load (a game, a benchmark, or Prime95 for stress testing) and observe the peak.
How to check on macOS
- iStatMenus: installs a menu bar item with live CPU, GPU, and fan readouts. Most accurate on Apple Silicon and Intel Macs.
- Hot (free): simpler menu bar display — shows the maximum temperature across all cores.
- Terminal:
sudo powermetrics --samplers smc -n 1 shows raw SMC sensor data including CPU die temperature.
How to check on Linux
sudo apt install lm-sensors
sudo sensors-detect # accept defaults
sensors
For more detail: watch -n 1 sensors updates every second. For per-core temps on modern CPUs, also try sudo apt install i7z && sudo i7z.
Safe temperature ranges
| State |
Good |
Acceptable |
Concerning |
| Idle (desktop) |
30–45°C |
45–55°C |
>60°C |
| Light load (browsing, office) |
45–65°C |
65–75°C |
>80°C |
| Heavy load (gaming, render) |
65–85°C |
85–95°C |
>95°C |
| Peak burst (Intel Turbo) |
Up to 95–105°C |
— |
Sustained >100°C |
Note: "Tjmax" (thermal junction maximum) is the temperature at which the CPU starts throttling. Most modern CPUs throttle at 95–105°C. Operating below this is safe — running at it constantly is not ideal for longevity.
If temperatures are too high
- Clean the heatsink and fan. Use compressed air to clear dust from heatsink fins and fan blades. This is often enough to drop temps 10–20°C.
- Replace thermal paste. On a desktop, this is a 15-minute job. On a laptop, it requires removing the bottom panel and heatsink. Thermal paste (Noctua NT-H1, Thermal Grizzly Kryonaut) costs ~$5–10.
- Improve case airflow. A desktop with no exhaust fan has hot air pooling. Add a rear 120mm exhaust fan.
- Check fan operation. A fan that has stopped spinning is obvious in HWiNFO (fan RPM reads 0). Replace a failed fan.
- Reduce power limits. In the BIOS, lower PL1/PL2 (Intel) or PPT (AMD) if the CPU runs at full power limits indefinitely. Many motherboards ship with unlocked power limits that boost performance but spike temps.
Common mistakes
Checking temperature only at idle. Idle temps tell you little. Monitor under realistic load — your typical workload or a 10-minute gaming session.
Confusing Tctl and Tdie on AMD Ryzen. Older Ryzen CPUs (1000–3000 series) report Tctl at +10°C above actual die temp. HWiNFO shows both. Use Tdie.
Re-applying thermal paste without cleaning off the old. Always clean with isopropyl alcohol (90%+) before applying new paste. Layering makes things worse.
Using too much thermal paste. A pea-sized blob in the center is correct for most CPUs. Too much squeezes out and can insulate instead of conduct.
What to skip
- Thermal paste with silver or metal particles on a laptop — risk of shorting nearby components if it spreads. Use non-conductive paste like Noctua NT-H1.
- Ambient temperature monitoring as a proxy for CPU health — room temp affects readings but a 10°C hotter room is not the same as a clogged heatsink.
- Liquid metal on consumer laptops unless you are experienced — it is corrosive to aluminum and permanent if misapplied.
FAQ
What is a normal CPU temperature for gaming?
70–85°C is normal and safe for most desktop CPUs under gaming load. Laptops run 5–10°C hotter due to thinner cooling. Above 95°C sustained means your cooling needs attention.
Why does my CPU spike to 95°C briefly?
Modern CPUs boost aggressively in short bursts (Intel Turbo Boost, AMD Precision Boost). A 95°C spike lasting 1–2 seconds under a burst workload is by design. Sustained 95°C under steady load is the concern.
How often should I replace thermal paste?
Desktop: every 3–5 years or when temps rise noticeably. Laptop: every 2–4 years, sooner for gaming laptops that run hot regularly.
Can high CPU temps damage the CPU permanently?
Modern CPUs throttle before reaching damaging temperatures and have emergency shutdown. Long-term operation near Tjmax accelerates electromigration but rarely causes outright failure in consumer use. The bigger risk is system instability and crashes.
Where to go next