[SOLVED] How to compare throttling

[guggas]

I have two of the same model of laptop(Thinkpad T580), one with an i5-8250u, and one with an i7-8650u.
Im curious to test how much performance boost, if any the i7 has over the i5. Word on the street is the i7 really isnt much better than the i5 due to thermal throttling. Is there a good way to test this? Im sure if I stress each cpu heavily they will both throttle, but is there a way to record or graph how much work each cpu is doing, its clock speed and temp in a way that each cpus performace can be compared?

Thanks

 

[sam2700]

The simplest thing you can do here is downloading prime95 and a temp monitor program, like hwinfo or open hardware monitor, open them up along with the task manager and run a torture test with small FFT in prime95. Now the things you may want to take into consideration are the running frequency on each processor and compare it with the maximum one, the average/highest temperature on each of those and the processor state (go in the performance tab in task manager and make sure each processor has 100% state, either way it's throttling). If you capped your processor maximum state (note that this is different that the one in task manager) from the power management settings, you have to set it back to 100%.

Also I don't think the processor has too much to do with the temperatures inside a laptop. It is more important that you have your laptop cleaned around once an year and you have a decent thermal paste. No laptop should be designed to throttle it's CPUw/GPU due to high temps. (I have an i7-4750HQ - this processor is supposed to run really hot - and in prime95 I have 67 - 70 degrees Celsius).
An i5 could be of course better than an i7, but you might want to consider other parameters different than the temperature or which one throttles or not. The temperature in your laptop is tightly bound to the way your laptop cooling system has been built and the dust inside your radiators. =))

I hope it helped you!

 

[CompuTronix]

guggas,

Although sam2700 has given you some excellent advice, it's absolutely critical to point out that you must configure Prime95 v29.8 to run only Small FFTs with all AVX test selections disabled. Without AVX, Prime95 Small FFTs is a true steady-state 100% workload with steady Core temperatures. However, if you do not disable all AVX test selections, then Prime95 can severely overload your CPU with an unrealistic and extremely brutal 130% workload, which will cause your laptop to throttle immediately.

Although this is a seemingly fine detail, I can not overemphasize the fact that it has a profound impact on workload, power consumption (Watts) and Core temperatures.

Also, use Hardware Info to monitor power consumption. When your CPU reaches its rated TDP, then you're at 100% TDP workload. Keep in mind that Windows Task Manager CPU % Utilization only indicates processor resource activity, NOT actual % TDP power consumption. Certain utilities such as Prime95 using AVX can push your CPU well above 100% TDP workload, which Task Manager is NOT capable of indicating.

Workload drives power consumption, which in turn drives Core temperatures. Accordingly, power consumption (Watts) is the best indication of your CPUs true workload.

You might also find ThrottleStop very useful, which is a utility that was written specifically for laptops by the author of Real Temp, Kevin Glynn.

sam2700,

On behalf of Tom's Moderator Team, welcome aboard! Excellent 1st post! Please let us know if you have any questions.

CT

 

[sam2700]

guggas,

Although sam2700 has given you some excellent advice, it's absolutely critical to point out that you must configure Prime95 v29.8 to run only Small FFTs with all AVX test selections disabled. Without AVX, Prime95 Small FFTs is a true steady-state 100% workload with steady Core temperatures. However, if you do not disable all AVX test selections, then Prime95 can severely overload your CPU with an unrealistic and extremely brutal 130% workload, which will cause your laptop to throttle immediately.

Although this is a seemingly fine detail, I can not overemphasize the fact that it has a profound impact on workload, power consumption (Watts) and Core temperatures.

Also, use Hardware Info to monitor power consumption. When your CPU reaches its rated TDP, then you're at 100% TDP workload. Keep in mind that Windows Task Manager CPU % Utilization only indicates processor resource activity, NOT actual % TDP power consumption. Certain utilities such as Prime95 using AVX can push your CPU well above 100% TDP workload, which Task Manager is NOT capable of indicating.

Workload drives power consumption, which in turn drives Core temperatures. Accordingly, power consumption (Watts) is the best indication of your CPUs true workload.

You might also find ThrottleStop very useful, which is a utility that was written specifically for laptops by the author of Real Temp, Kevin Glynn.

sam2700,

On behalf of Tom's Moderator Team, welcome aboard! Excellent 1st post! Please let us know if you have any questions.

CT

This is some really valuable information. I followed your steps and I noticed two things. The temperatures are ~4 degrees lower and the CPU frequency is stable and close to it's maximum frequency. It seems like AVX tests are meant to test your thermals and cooling system close to it's margins and disabling them tests your stability. Thank you very much!

 

[CompuTronix]

sam2700,

We have a Guide that covers this topic. It's a "Sticky" located at the top of the CPUs Forum where this Thread is posted. Stickies are helpful information resources which are permanently "stuck" in place so they're always available for everyone's benefit.

We encourage our Members to check the Stickies before posting Threads, as they frequently contain the information you need. This can save you time needlessly searching for answers, or waiting for others to responds with perhaps somewhat less than "well informed" answers.

You might like to read this Sticky: Intel Temperature Guide.

CT

 

[SamirD]

This is some really valuable information. I followed your steps and I noticed two things. The temperatures are ~4 degrees lower and the CPU frequency is stable and close to it's maximum frequency. It seems like AVX tests are meant to test your thermals and cooling system close to it's margins and disabling them tests your stability. Thank you very much!

Some very interesting results considering both processors are seen identical except for the clock speed:
https://www.cpubenchmark.net/compare/Intel-i5-8250U-vs-Intel-i7-8650U/3042vs3070

But they actually have quite a few differences including features:
http://www.cpu-world.com/Compare/86...5-8250U_vs_Intel_Core_i7_Mobile_i7-8650U.html

I wonder how many of those features factor into the temperature difference?

 

[sam2700]

Some very interesting results considering both processors are seen identical except for the clock speed:
https://www.cpubenchmark.net/compare/Intel-i5-8250U-vs-Intel-i7-8650U/3042vs3070

But they actually have quite a few differences including features:
http://www.cpu-world.com/Compare/86...5-8250U_vs_Intel_Core_i7_Mobile_i7-8650U.html

I wonder how many of those features factor into the temperature difference?

My CPU is i7-4750HQ

And I tested it with and without AVX enabled. The difference in temperature was around 4 degrees Celsius.

 

[SamirD]

My CPU is i7-4750HQ

And I tested it with and without AVX enabled. The difference in temperature was around 4 degrees Celsius.

Well, there's the difference then, lol.

But seriously, if there was a difference between the i5 and i7 of 4C, it would be interesting to know how much the additional instructions/features on the i7 factor into the difference.