Surface Power Management Overview: How battery saver impacts CPU performance

[AlTech]

So, I was curious as to how Surface Power Management options in Windows 10 affected CPU performance.

My initial hypothesis was that Battery Saver would cause worse CPU performance than normally.

 

During my time testing, I discovered that there is a lot more going on.

 

So I decided to use Cinebench R15 (although Maxon should really update Cinebench), Blender, and my very own CSMark benchmark.

 

Testing notes:

• Testing conducted on 2016 Production Surface Pro 4 i7 SKU with Intel core i7-6650U (15w 2C/4T) @ 2.2-3.2GHz with 64MB eDRAM.
• System was running Windows 10 Build 15063.608
• Blender 2.79 was used due to 2.79 being released somewhat recently.
• Cinebench R15 v15.038 was used
• CSMark Pre-release Dev Build 0.18.1709.1 was used
• Latest Surface Firmware was used.
• Background programs consisted of Calculator App, FileExplorer and notepad.
• Each benchmark was run 3 times in each power management option to provide accurate results.

 

In Blender, I decided to use the Ryzen Graphic for testing as it's quite a uniquely demanding render for CPUs.

 

Performance was definitely interesting with Blender. We can see that Blender clearly rendered the Ryzen graphic faster when using "Best Performance". This isn't too surprising since Battery Saver disables Turbo Boost. Best Performance appears to run at full boost all the time when stressed. Recommended seems to provide a conservative boost which lands somewhere between Better Performance and Battery Saver.

 

 

Moving onto Cinebench R15, we can see similar results although the difference between the recommended and Battery Saver options aren't nearly as large here.

 

 

Moving on to CSMark. the single threaded performance is definitely stronger when using anything above Recommended. That being said, all options above Battery Saver provided somewhat respectable scores.

 

 

Last but not least we have the CSMark Multi-threaded score and it's here where we see one potential outlier in "Better Performance". The scores of "Best Performance" and "Better Performance" are within margin of error and so it would seem that the performance is about the same which is fairly different from what we saw with Cinebench R15 and Blender.

 

 

In conclusion, running a CPU using Battery Saver and/or any power option below Best Performance can impact CPU performance.

 

Anyhow, I hope it was fun to read as it was to do .

 

Thanks for checking this out.

[Guest]

Is the "High performance" preset a good power plan option to optimise hardware better, or do I choose the recommend settings.                       

[RGProductions]

1 hour ago, TheBeastPC said:

Is the "High performance" preset a good power plan option to optimise hardware better, or do I choose the recommend settings.                       

ALLL OF THE SPEEED

[AlTech]

1 hour ago, TheBeastPC said:

Is the "High performance" preset a good power plan option to optimise hardware better, or do I choose the recommend settings.                       

Depends on if Laptop or Desktop. If on Desktop then I always use High Performance.

 

If on laptop it depends on what I'm doing.

 

The plans tested above are in Windows 10 on select devices and they don't use Control Panel for it though.

 

20 minutes ago, RGProductions said:

ALLL OF THE SPEEED

rly bruh?