i7 8750 H temperature issues

[Linx182]

Just got a new P52 - i7 8750H nvidia p1000 . Have a few questions regarding cpu temperature, hope some one more experienced can clarify. I ran xtu and noticed that the cpu temperature is quite high under load. While performing cinebench the temperature was around 90- 97 °C and thermal throttling was a yes. The multicore score was 942 single core 158 and open GL 112 fps. During the xtu stress test the temperature was roughly the same and thermal throttling. So I undervolted the cpu, got stable results at -0.160 (I ran a few stress tests, prime 95 for a couple hours, exported a few heavy projects in premiere pro). After undervolting, cinebench scores improved significantly around 1200 multicore, 171 single core, 126 opengl. But the temperatures remained the same , above 90°C. During the xtu stress test the throttling was gone and cpu temperature was from 75 to 80°C. On idle the temperature stays between 50 - 60°C. After that i ran the audio software with two 1080p monitors (I work with video) a midi controller and loaded a relatively heavy session. The temperature hovered from 78 to 98°C the whole time I had the monitors connected. The cpu clocked at around 3.8 GHz. I also got different cpu utilization readings in xtu and task manager 44% in xtu and 78% in task manager. Ive read tha most 8th gen cpus are prone to high temperatures but im worried about the cpu running so hot. How dangerous is running the cpu on 90 degrees? I'd really appreciate it if someone would help me make sense of these readings

[DakotaWebber]

obviously it will be high temperature in a stress test load, laptop cooling solutions generally throttle unless its a bulky gaming laptop or a low powered chip, it wont hurt your system for it running at 90c all the time, after all the manufacturer knows their systems in and out, and considering its a workstation laptop they know businesses will be using it and need it to run at an acceptable standard

the t junction for the chip is 100c anyway, and obviously there are safety factors such as throttling to keep your chip from cooking itself

[Turtle Rig]

17 minutes ago, Linx182 said:

Just got a new P52 - i7 8750H nvidia p1000 . Have a few questions regarding cpu temperature, hope some one more experienced can clarify. I ran xtu and noticed that the cpu temperature is quite high under load. While performing cinebench the temperature was around 90- 97 °C and thermal throttling was a yes. The multicore score was 942 single core 158 and open GL 112 fps. During the xtu stress test the temperature was roughly the same and thermal throttling. So I undervolted the cpu, got stable results at -0.160 (I ran a few stress tests, prime 95 for a couple hours, exported a few heavy projects in premiere pro). After undervolting, cinebench scores improved significantly around 1200 multicore, 171 single core, 126 opengl. But the temperatures remained the same , above 90°C. During the xtu stress test the throttling was gone and cpu temperature was from 75 to 80°C. On idle the temperature stays between 50 - 60°C. After that i ran the audio software with two 1080p monitors (I work with video) a midi controller and loaded a relatively heavy session. The temperature hovered from 78 to 98°C the whole time I had the monitors connected. The cpu clocked at around 3.8 GHz. I also got different cpu utilization readings in xtu and task manager 44% in xtu and 78% in task manager. Ive read tha most 8th gen cpus are prone to high temperatures but im worried about the cpu running so hot. How dangerous is running the cpu on 90 degrees? I'd really appreciate it if someone would help me make sense of these readings

Your temp on full load shouldnt pass 85c.  Your at almost 100c.  Not good.

[NineEyeRon]

3 hours ago, Turtle Rig said:

Your temp on full load shouldnt pass 85c.  Your at almost 100c.  Not good.

While I do agree is that benchmark testing load only or normal use load?

 

If during temps only exceed 85 during benchmarking and stress tests then the answer is don’t run those tests.

 

What must be considered for laptops is use conditions, a under 85 on a desk, propped up in a cold room is great. A warm lecture theatre propped up on ones lap may result in different performance. If one isn’t asking much of the system then a bit of throttling isn’t an issue.

 

I only say these things so people don’t throw out or replace perfectly functional items which are suitable for 99% of use because they “fail” a benchmark or stress test that isn’t replicating actual use.

 

Finally we must define “dangerous”, do we mean loss of life or life changing injuries to the user? Do we mean loss of critical data, money or time? Or do we just mean a reduced life of a component or system but not to within a timeframe which will cause any hardship or loss? So what if the device I replace every 5 years only would last 7 instead of 9?

[Turtle Rig]

1 hour ago, NineEyeRon said:

While I do agree is that benchmark testing load only or normal use load?

 

If during temps only exceed 85 during benchmarking and stress tests then the answer is don’t run those tests.

 

What must be considered for laptops is use conditions, a under 85 on a desk, propped up in a cold room is great. A warm lecture theatre propped up on ones lap may result in different performance. If one isn’t asking much of the system then a bit of throttling isn’t an issue.

 

I only say these things so people don’t throw out or replace perfectly functional items which are suitable for 99% of use because they “fail” a benchmark or stress test that isn’t replicating actual use.

 

Finally we must define “dangerous”, do we mean loss of life or life changing injuries to the user? Do we mean loss of critical data, money or time? Or do we just mean a reduced life of a component or system but not to within a timeframe which will cause any hardship or loss? So what if the device I replace every 5 years only would last 7 instead of 9?

My friend you need to stop running these synthetic tests or stress tests.  In the real world your CPU will not get taxed like Prime95 taxes it or other synthetic tests.  You might as well minus 10c from your bench or synthetic test and that is the value you would get under real world.  Try Aida64 and note how Prime is 10 to 12c warmer.  Stop running those tests or one day you might botch your CPU.  Go by real world.  Your idle temps should be in 30's c and regular usage should be 65c ish. imo

[Linx182]

22 hours ago, NineEyeRon said:

While I do agree is that benchmark testing load only or normal use load?

 

If during temps only exceed 85 during benchmarking and stress tests then the answer is don’t run those tests.

 

What must be considered for laptops is use conditions, a under 85 on a desk, propped up in a cold room is great. A warm lecture theatre propped up on ones lap may result in different performance. If one isn’t asking much of the system then a bit of throttling isn’t an issue.

 

I only say these things so people don’t throw out or replace perfectly functional items which are suitable for 99% of use because they “fail” a benchmark or stress test that isn’t replicating actual use.

 

Finally we must define “dangerous”, do we mean loss of life or life changing injuries to the user? Do we mean loss of critical data, money or time? Or do we just mean a reduced life of a component or system but not to within a timeframe which will cause any hardship or loss? So what if the device I replace every 5 years only would last 7 instead of 9?

Thanks for the answers. Here are some real world readings. When using my  audio software with a couple virtual instruments the temperature was from around 75°C degrees to 93°C with cpu utilization at around 60%. When building lighting in unreal the temperature went from 82 to as high as 97°C. On idle the temperature is from 46 - 55