25 - 10 - 2024
Login Form



 


Share this post

Submit to FacebookSubmit to TwitterSubmit to LinkedIn

TEST BED

 

 

 

 

 

 

 

 

 

TESTING METHODOLOGY

    To accurately test the performance of a CPU its clocks need to be set manually to a specific frequency/speed so to that end every model in my charts will be pushed near the maximum boost clock possible according to each manufacturer. I'll also try to use the same clocks and the same hardware configuration (motherboard, drive, memory, graphics card, cooler and power supply - even the same thermal paste will be used) when testing CPU models from the same series and as for operating system since till this day Windows 10 Pro is still far more popular among consumers than Windows 11 I’ll be using a fresh Windows 10 Pro installation fully updated up to the day I begin my tests. Needless to say, the latest bios and drivers will be installed prior to recording my results.


    As for the tests themselves* they include the latest versions of both popular synthetic benchmarks and games. For synthetic benchmarks I’ll be using UL 3DMark Professional (Steel Nomad & CPU Profile), AIDA64 (GPGPU), Maxon Cinebench R20, CPUID CPU-Z, Passmark Performance Test, GeekBench AI, Sisoftware Sandra Titanium and ASUS RealBench and as for games I’ll be using Dying Light 2, Cyberpunk 2077 and Forza Horizon 5 (both at 1080p and 2160p resolutions with max settings - the latest version of Cyberpunk 2077 with everything maxed out - minus RT - seems to be significantly heavier than the one I've been using for GPU tests so do keep that in mind). Each test is repeated a total of 3 times and then the average numbers are recorded in the charts. Also, since I’m positive many consumers and professionals alike would like to know I’ll also be using AIDA64 (stability test), CPU-Z (stress test) and HWMonitor to record the minimum and maximum temperatures and power consumption both at idle and after 5 minutes of heavy stress load (needless to say the cooler used will be set at 100% speed/performance – these results are only valid for the set clocks, with lower and/or higher clocks you can expect lower/higher power consumption and temperatures).


* As of October 2024, I slightly changed my testing methodology (primarily because I no longer have the GPU I used for past CPU tests) and also introduced IGPU tests (1080p resolution at low presets for Dying Light 2, Cyberpunk 2077 and 3D Mark Steel Nomad Light).