Open any monitoring app and your graphics card reports at least two numbers: the edge (or core) temperature and the hotspot (junction) temperature. New builders often panic when the hotspot reads 15-25C higher than the number they expected. In reality both are normal, and understanding the gap between them tells you a lot about how well your cooler is mounted and whether your card is aging.
As an Amazon Associate we earn from qualifying purchases at no extra cost to you.
Our Top Picks
Top 3 picks at a glance
What Each Reading Means
Edge temperature comes from a sensor near the edge of the GPU die and is the conservative, widely quoted figure. Hotspot is the single hottest point anywhere across the die, picked from a grid of on-chip sensors. Because it tracks the worst spot, it is always higher.
| Reading | Typical gaming value | What it reflects |
|---|---|---|
| Edge / core | 60-75C | General die temperature |
| Hotspot / junction | 75-95C | Hottest single point on the die |
| Memory junction | 70-95C | VRAM temperature (separate sensor) |
Why the Gap Matters More Than the Number
A healthy delta
On a well-mounted cooler with fresh thermal paste or pads, the hotspot typically sits about 10-15C above the edge reading. That spread is expected and nothing to worry about. Modern GPUs will happily run a hotspot in the mid-80s to low-90s and only start throttling as the junction approaches roughly 100-110C depending on the model.
A warning-sign delta
When the gap grows to 20-30C or more, it usually points to a mounting or paste problem: uneven cooler pressure, pump-out of the factory paste after a couple of years, or a warped contact surface. The card may still hold an acceptable edge temperature while one spot on the die runs much hotter, which drives fan noise up and clocks down.
What To Do About a High Hotspot
If your hotspot delta is small, leave it alone; the card is working as designed. If the delta has crept up over time, improving case airflow helps a little, but the real fix for an old card is a repaste and fresh thermal pads, which can pull a bloated delta back down toward the 10-15C range. Undervolting is another effective lever: trimming voltage can drop both edge and hotspot temperatures by 5-15C with little or no performance loss.
Tracking the Delta Over Time
The most useful habit is to log your hotspot delta when the card is new and healthy, then check it every few months. A fresh, well-mounted card typically shows a 10-15C spread between hotspot and edge under a steady gaming load. If that gap slowly widens to 20C, then 25C over a year or two, it is almost always the factory thermal paste pumping out or drying, concentrating heat at one point on the die. Catching that trend early means a cheap repaste restores the card instead of living with rising fan noise and creeping throttling.
It also helps to know your specific card’s throttle points, since they vary by generation. Many designs let the junction temperature climb into the high 90s or low 100s before they start trimming clocks, and a few tolerate even more. That means a 90C hotspot is often nowhere near the limit, and reacting to the raw number alone can send you chasing a problem that does not exist. Combine the absolute reading, the delta, and a light undervolt, which can pull both figures down 5-15C with little performance cost, and you have a complete picture of whether your cooling is genuinely healthy or quietly degrading.
FAQ
Is a 90C GPU hotspot dangerous?
Not on its own. Many cards are specced to run junction temperatures into the 90s and only throttle near 100-110C. A 90C hotspot with a normal 10-15C delta is fine; the same 90C with a 30C delta over a low edge temp is the sign worth investigating.
Which temperature should I use to judge cooling?
Watch both. Use the edge temperature to compare against typical values and the hotspot delta to judge mount quality and aging. A rising delta over months is the most useful early warning that a repaste is due.
Bottom line
Edge temperature tells you how hot the card is overall, while hotspot tells you about the worst single point and, through its delta, the health of the cooler mount. Do not fixate on the raw hotspot number. Watch the gap between the two, and treat a growing delta rather than a high absolute reading as your cue to repaste or undervolt. Log the delta when the card is new so you have a healthy baseline to compare against later.
