Analysing thermal data...
Select Your Component Above
Choose your CPU or GPU, enter the temperature you saw in HWiNFO or Afterburner, and click Check Temperature.
Is 90°C Too Hot for a CPU?
The short answer: it depends entirely on the CPU. For an Intel Core i9-13900K or 14900K, 90°C under sustained load is completely normal and within Intel's published specification. For an AMD Ryzen 5 5600X, 90°C would be alarmingly high and a clear sign that something is wrong with the cooling setup.
Every CPU has a TJ Max (thermal junction maximum) — the temperature at which the processor begins reducing clock speeds to protect itself. On modern Intel desktop CPUs this is 100°C. On AMD Ryzen 7000 and 9000 series it is 95°C. On Ryzen 5000 series AM4 chips it is typically 90°C (Tdie). The throttle point — where actual performance loss begins — is usually 5°C below TJ Max.
What matters for performance is not the absolute temperature but the headroom to the throttle point. A CPU running at 88°C with a 95°C throttle point has 7°C of headroom and is warning-level territory. A CPU running at 88°C with a 100°C throttle point has 12°C of headroom and is performing normally. Without knowing your specific CPU's TJ Max, a raw temperature reading is meaningless.
What Temperature Should My CPU Run at While Gaming?
Target ranges for peak gaming load (not stress tests). These are comfortable operating temperatures with a quality aftermarket air cooler or AIO at 22°C ambient.
| CPU | TJ Max | Throttle Point | Target Gaming Temp | 90°C Safe? |
|---|---|---|---|---|
| Core i9-13900K / 14900K | 100°C | 95°C | 75–90°C | Yes |
| Core i7-13700K / 14700K | 100°C | 95°C | 72–85°C | Yes |
| Core i5-13600K / 14600K | 100°C | 95°C | 65–78°C | Marginal |
| Ryzen 9 7950X / 7900X | 95°C | 90°C | 75–88°C | Yes |
| Ryzen 7 7700X / 9700X | 95°C | 90°C | 68–82°C | Marginal |
| Ryzen 7 5800X | 90°C | 85°C | 70–83°C | Warning |
| Ryzen 5 5600X | 90°C | 85°C | 60–75°C | Too Hot |
| Core Ultra 9 285K | 105°C | 100°C | 70–88°C | Yes |
Target ranges assume a quality aftermarket cooler at 22°C ambient. Stress tests (Cinebench, Prime95) will push temps higher than gaming.
CPU vs GPU Thermal Throttling: How They Differ
CPU and GPU thermal throttling are mechanically similar — both reduce operating frequency to lower heat output — but the temperature thresholds, monitoring sensors, and context are quite different.
Modern Nvidia GeForce RTX 40 series GPUs throttle at 83–84°C by default through the GPU Boost algorithm. This is intentional: the boost clock is set to maintain this temperature target, not to avoid it. A GPU sitting at 83°C is performing exactly as Nvidia designed. Below the throttle point, higher temperatures actually allow higher boost clocks within the power limit.
AMD RDNA2 and RDNA3 GPUs introduce a further complication: the GPU junction temperature (hotspot) versus the GPU edge temperature. HWiNFO and Radeon Software will show junction temperatures reaching 100–110°C on a 7900 XTX under load. This is AMD's published thermal design maximum — it is the temperature of the hottest point on the die, not the die average. The GPU edge temperature on the same card will typically be 20–30°C lower. AMD GPUs begin throttling when the junction reaches the throttle threshold, so monitoring junction temp is the correct sensor to watch.
For CPUs, Intel publishes a single TJ Max and the throttle mechanism is straightforward. AMD Ryzen adds the Tctl vs Tdie distinction. On Ryzen 5000 series, some CPU models add a 10°C reporting offset to Tctl for firmware fan curve purposes — HWiNFO reports both, and Tdie is the accurate thermal measurement. On Ryzen 7000 and 9000, this offset was removed, so Tctl and Tdie are the same value.
Why Is My CPU Temperature So High at Idle?
A healthy idle temperature is roughly ambient room temperature plus 15–20°C. If your room is 22°C, you should see 37–42°C at idle on most desktops with decent cooling. Significantly higher idle temperatures point to one of a few common causes.
Dried or cracked thermal paste is the most common culprit on systems older than 2–3 years. Paste degrades over time, creating microscopic air gaps between the CPU die and heatsink that drastically reduce heat transfer. Replacing a $5 syringe of paste can lower temperatures by 10–20°C on an aging system.
Incorrectly mounted CPU cooler is the most common cause of suspiciously high temps on a new build. If the cooler's backplate is not fully seated or the mounting screws are not evenly tightened, one corner of the heatsink will have poor contact with the CPU integrated heat spreader. The result is normal-looking hardware that runs 20–30°C hotter than it should. Remounting with fresh paste fixes this immediately.
Background CPU load from antivirus scans, Windows Update, or poorly behaved software can disguise itself as a cooling problem. Open Task Manager, sort by CPU%, and ensure the system is genuinely idle before measuring baseline temperatures. Some monitoring software (including HWiNFO itself when first launched) causes a brief CPU spike that inflates the first few readings.
How to Lower CPU Temperatures: Cooling Upgrade Path
Improvements for CPU temperatures scale roughly in this order of cost-to-impact ratio:
1. Reapply thermal paste (Free–$10). If your system is 2+ years old, this is always the first step. Remove the cooler, clean both surfaces with isopropyl alcohol, apply a pea-sized dot of quality paste (Noctua NT-H1, Thermal Grizzly Kryonaut, or Arctic MX-6), and remount. Expect a 5–15°C improvement on aged paste.
2. Add case fans ($15–50). A front intake fan dramatically reduces the ambient temperature inside the case, which reduces the delta between your cooler's input air and the heat being shed. One 140mm intake fan on a previously fanless case can lower CPU temps by 8–12°C. Aim for positive pressure (more intake CFM than exhaust).
3. Upgrade the CPU cooler ($30–100). The step from a stock cooler to a mid-range tower air cooler (Noctua NH-U12S, be quiet! Dark Rock 4, DeepCool AK620) is the biggest absolute temperature gain. On a 65–125W CPU, this typically reduces load temps by 15–25°C versus the stock cooler.
4. 240mm or 360mm AIO ($80–180). For high-TDP CPUs (Intel i9 13th/14th Gen, Ryzen 9 7000/9000) that cannot stay below 90°C with even the best air coolers, a 360mm AIO is the practical ceiling for consumer cooling. The larger radiator surface dissipates heat faster than any single-tower air cooler can.
Frequently Asked Questions
For most modern CPUs, 80°C under gaming load is comfortable and safe. On Intel 13th and 14th Gen desktop CPUs (TJ Max 100°C, throttle point 95°C), 80°C provides 15°C of headroom — solidly in the "Good" range. On AMD Ryzen 5000 series with a 90°C TJ Max and 85°C throttle point, 80°C is only 5°C from throttling — warning territory. Always interpret temperatures relative to the throttle point of your specific CPU, not as an absolute number.
There are two stages. First, thermal throttling: when the CPU hits its throttle point (typically TJ Max minus 5°C), it reduces its operating clock speed and voltage to lower heat output. Performance drops but the system stays running. Second, if temperatures somehow continue rising beyond TJ Max — which is extremely rare with intact hardware — the CPU triggers a thermal shutdown to cut power entirely and protect the silicon. Modern systems rarely reach actual shutdown in practice; throttling resolves the situation first.
Thermal throttling itself does not damage the CPU — it is a protective mechanism designed to prevent damage. However, sustained operation at very high temperatures (above 90–95°C) over months and years can accelerate electromigration in the CPU's transistors, potentially reducing the component's long-term lifespan. This effect operates over very long timescales and is unlikely to cause noticeable degradation within a typical ownership period of 3–5 years. The bigger practical concern is performance: a CPU that is throttling is delivering less performance than you paid for.
Task Manager reads the CPU Package temperature, which is a single averaged value across all cores. HWiNFO reads individual core temperatures, the CPU die temperature (Tdie), the control temperature (Tctl), and sometimes a distance-to-TJ Max figure. On AMD Ryzen 5000 series "X" chips, Tctl includes a firmware-imposed offset of up to 10°C above the actual silicon temperature (Tdie) — this is intentional and used for fan control logic. HWiNFO shows both; use the Tdie or "CPU Die Average" figure for accurate thermal analysis. On Intel and AMD Ryzen 7000+, Tctl and Tdie are the same value.
For Nvidia RTX 30 and 40 series GPUs, 83°C is the GPU Boost target temperature. The GPU actively adjusts clock speeds and power draw to maintain this temperature. It is not a sign of a problem — it is the algorithm working as designed. If your GPU is hitting 83°C and Boost clocks are stable, cooling is adequate. If you see clocks dropping below the rated boost clock while at 83°C, the GPU is thermal throttling and a better-airflowed case or more aggressive fan curve would help.
Every 2–3 years is a reasonable interval for most users. Polymer-based pastes (the common grey compounds) dry out and crack over thermal cycles, reducing conductivity. Liquid metal compounds last longer but require more care during application and are not compatible with aluminum heatsinks. If you have not replaced paste in 3+ years and your temperatures are elevated, this is the first maintenance step to try. Signs that paste needs replacing: idle temps have risen noticeably over time, temperatures spike rapidly under light load, or there is a large delta between core temperatures on a single chip.
Related Calculators
Custom Water Cooling Calculator
Plan your custom loop: radiator size, pump, reservoir, and flow rate
AIO vs Air Cooler Decision
Get a clear recommendation based on your CPU TDP, case, and budget
CPU/GPU Bottleneck Calculator
Find out if your CPU is limiting your GPU's performance in-game
RAM Capacity Checker
Check if 16GB or 32GB RAM is enough for your games and apps
GPU FPS Estimator
Estimate your GPU's frame rate output across resolutions and games
PC Build Budget Planner
Allocate your build budget optimally across every component