• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

GPU temp, hot spot?

Soldato
Joined
3 May 2012
Posts
9,957
Location
Wetherspoons
Hi guys,

Just wanted to look at the temps of the 3070 I got from OCUK on Fridays deal.

hotspot.png


What does hot spot mean?

I am not surprised its running warm today, its like 30c ambient temperature in there room today.
 
The hot spot sensor should be on the back of the card I believe, where there's no active cooling. As far as I know it's related to memory temperate limits, which are - dependent on the make/model - around 110-120 C.

I'm not sure if all cards throttle based on this temperature, but the FE 3090s (and I believe some 3080s) end up throttling when run hard due to what is essentially insufficient thermal interface material on the memory chips.

Someone will come along in a second and correct me no doubt :p
 
Ok thanks.

I'm not worried at all, seems to be working perfectly I'm very happy with it. Good day to test it today as it's just tipping into 31c in here which is about as hot as it will ever get in this room.
 
It's the max GPU core temp. Explanation below.

Nvidia's GPUs currently report the temperature to monitoring programs as a single value — GPU Temp, or whatever your program may call it. In reality though, the GPU doesn't have just one temperature sensor on it smack in the middle. Instead, the GPU has many sensors spread across its topography, and it averages the results for the reported number.

As you can imagine, the temperature across the sensors can vary wildly depending on the load on the specific area, the flatness of the GPU surface and the cooler, the mating between the two, the thermal grease used — you get the idea. As such, it's actually quite interesting to know what the peak temperature is that's recorded on the GPU

https://www.tomshardware.com/amp/news/nvidia-hwinfo-hotspot-detection

90° does seem quite high. It might be worth looking at using a custom fan curve in Afterburner for example.
 
Last edited:
The hot spot sensor should be on the back of the card I believe, where there's no active cooling. As far as I know it's related to memory temperate limits, which are - dependent on the make/model - around 110-120 C.
I believe you're mixing it up with the Memory Junction Temp reading. The 3070 doesn't support that, higher spec cards only probably due to them being GDDR6X.
 
Hi guys,

Just wanted to look at the temps of the 3070 I got from OCUK on Fridays deal.

hotspot.png


What does hot spot mean?

I am not surprised its running warm today, its like 30c ambient temperature in there room today.
GPU-Z is your friend, set it to log the sensor output and game away.
get an app called Generic Log Viewer to view the log in graph form.

Apart from being able to monitor the clock speed, temps etc over a longer period of time, there's a flag called Perfcap Reason which will tell you, if the card is throttling and why.

Here's an example that I sent to someone over the weekend. In Green is a well set up card, steady, high clock speed, no throttle triggers
in Red is a badly set up card, constantly bouncing between voltage, temperature and power limits.

Fk4RXsM.png


To Translate
16 is ideal
8 is voltage limited
2 is temperature limited
1 is power limited
4 is probably the worst since that is "the core is not stable at this speed at this voltage" and you're likely to have crashes.

As you can see above, 13 is quite common on this trace...that basically means that it's throttling for a number of reasons :cry:, unstable, power and voltage
 
Last edited:
GPU-Z is your friend, set it to log the sensor output and game away.
get an app called Generic Log Viewer to view the log in graph form.

Apart from being able to monitor the clock speed, temps etc over a longer period of time, there's a flag called Perfcap Reason which will tell you, if the card is throttling and why.

Here's an example that I sent to someone over the weekend. In Green is a well set up card, steady, high clock speed, no throttle triggers
in Red is a badly set up card, constantly bouncing between voltage, temperature and power limits.



To Translate
16 is ideal
8 is voltage limited
2 is temperature limited
1 is power limited
4 is probably the worst since that is "the core is not stable at this speed at this voltage" and you're likely to have crashes.

As you can see above, 13 is quite common on this trace...that basically means that it's throttling for a number of reasons :cry:, unstable, power and voltage

Thanks for that.


looks-good.png


I guess the solid green bit means it is running well? That was testing out PUBG 1440p, 144hz almost max settings, I dont put the AA all the way up at 1440p I dont think its necessary. The blue parts towards the start was launching the game/loading the map etc, once in the game all solid green.
 
Back
Top Bottom