Originally Posted by Awsan
I bench everyday with the same settings with the same temps and everyday I get a different score ranging between 9500-9900 and single of 480 to 520.
So either have an insanely bloat-free version on another drive or USB drive or live with it as it is.
Thanks A lot Microsoft
Its about background services, and optimizations.
Here is the difference you are talking about. This was during a benchmark session I had back in January, but this shows the difference between scores, when the only difference is Windows Optimization. This isn't a special boot, this is my Everyday Driver, Windows Installation. The 1st was my highest Score I could achieve with this overclock before going through the trouble of killing Windows Services. The second screenshot is the EXACT SAME Overclock, using the exact same Voltages and settings, the one difference being the Services and Programs I allowed to run in the background (And there was a day in between the Benchmark sessions, hence the 2 degree difference in Room Ambient). All Of the Programs, save for HWinfo (Which will take away from your Score as well) were opened after the fact.
Notice the Power Draw difference between the 2 Runs even though both have the same exact OC. This is because the 2nd Run is also in High Priority Mode with less services battling for clock cycles, allowing it to push the CPU even further.
This could be the difference between a Clean Windows Install and a bloated One. Or in my case, and a lot of seasoned benchmarkers case, the difference between closing the services before a benchmark run and putting the Benchmark Program in High Priority Mode, Vs just allowing the services to do what they want. Its insane the difference it makes in scores, but it is pretty big.
Edit: The Power Draw Difference between the 2 after going through my notes was actually because HWinfo was closed during the second test, I did a CB15 run after opening HWinfo before the second screenshot to show approximate Full Temps, but obviously CBR20 draws more power than CBR15, so thats why the difference in Power Draw here. Having HWinfo Closed for the run also definitely helps in having a higher score vs the 1st Run where HWinfo Was Open during the CBR20 Run. (When first Posting this I looked at the screen shots too fast and thought that the second, screenshot had the higher power draw. There would be a higher power draw between the 2 scenarios, but its more in the realm of 5 to 10 watts, not 10 to 15w, so my bad there.
Keep in mind I could still
close HWinfo, and
use a Completely Clean Windows Install with Explorer Killed, to get an even higher score, but this goes to show you that you don't have to go through that length just to get a good benchmark score, just close ALL Non Essential Windows Services and Programs.