FPU stability test, AVX, and temperatures

[rexbinary]

Hi all,

I've noticed that when I run the FPU stability test alone there is little difference in maximum temperatures with or without AVX enabled. Actually, the test runs about 1-2 degrees cooler with AVX enabled. This is not what I expected. After working with other tools like Prime95 I expected there to be a significant difference in temperatures with AVX on verses AVX off.

Why is this? My CPU is an i5-9600k for reference, and I have the AVX offset (downclocking) option disabled.

 

 

[Fiery]

3 hours ago, rexbinary said:

Hi all,

I've noticed that when I run the FPU stability test alone there is little difference in maximum temperatures with or without AVX enabled. Actually, the test runs about 1-2 degrees cooler with AVX enabled. This is not what I expected. After working with other tools like Prime95 I expected there to be a significant difference in temperatures with AVX on verses AVX off.

Why is this? My CPU is an i5-9600k for reference, and I have the AVX offset (downclocking) option disabled.

AIDA64 has several codepaths to stress a CPU.  It has different, heavily optimized codepath for your processor that utilizes AVX, and another codepath that is also heavily optimized to stress your CPU without using AVX.  Since all our codepaths are optimized "to the teeth", there are situations when they can achieve quite similar results.  It depends on the CPU type, AVX offsets, factory default TDP, overclocking settings, and a lot of other factors.  We've added the ability to disable using AVX (and AVX-512) in the AIDA64 System Stability Test to let you adjust the test to a deeper degree than when everything works automatically.

[rexbinary]

Makes sense. Thanks for your help! I really enjoy using AIDA64.