5820k Dynamic OC Issues

[OG_Mega]

8 minutes ago, done12many2 said:

 

Nice.  RealBench is a great stress test because it pushes your CPU and GPU to 100% load at the same time.  Total system stress and a lot more realistic to typical workloads than most others.

 

Not sure why your results varied so much between the two benchmark runs like that, but a lot of factors could have impacted the score.  Make sure that you run it with Windows Power Options set to High Performance every time.  Make sure that background tasks are running or at least make sure their the same each time.

 

4.5 GHz is a very respectable overclock especially when you consider that it's stable in RealBench for a couple of hours.  That's more than most folks do here on LTT.  

 

Oh, and to answer your question.  Performance is all that matters.  The clockspeed means nothing if it doesn't produce results.

 

If you feel like it, go to our LTT Cinebench thread and download/run the Cinebench benchmark and share your results with others in the thread.  People will be willing to help you out as much as you need.

 

Good luck man. 

 

Awesome, thanks so much for all your help!

 

To clarify - I only hit 111k at 4.5ghz on benchmark, the 120k at 4.4ghz was hypothetical.

 

When you say "performance is all that matters," you are basically saying that a higher stable benchmark score is better, even if the clock speed is lower, right?

[done12many2]

1 minute ago, OG_Mega said:

Awesome, thanks so much for all your help!

 

To clarify - I only hit 111k at 4.5ghz on benchmark, the 120k at 4.4ghz was hypothetical.

 

When you say "performance is all that matters," you are basically saying that a higher stable benchmark score is better, even if the clock speed is lower, right?

 

Performance trumps everything, but if you are scoring higher at a lower clock speed, that's an indicator that something else is going on.  Higher clock speeds should generally return higher scores unless other issues such as thermal or power throttling are occurring at the higher clock speed preventing the CPU from achieving higher performance.  

[OG_Mega]

16 minutes ago, done12many2 said:

 

Performance trumps everything, but if you are scoring higher at a lower clock speed, that's an indicator that something else is going on.  Higher clock speeds should generally return higher scores unless other issues such as thermal or power throttling are occurring at the higher clock speed preventing the CPU from achieving higher performance.  

I've attached a screenshot of my latest benchmark... I think changing the power settings was a game changer. 

 

I changed noting except changed Power Options to High Performance like you suggested and went from 109k benchmark straight to 144,830... insane.

 

I should mention that this was after 1hour of Real Bench stress testing.

[done12many2]

Very nice. Congrats on the overclock man. 

[NuclearKing]

13 hours ago, done12many2 said:

Not sure why your results varied so much between the two benchmark runs like that, but a lot of factors could have impacted the score.  Make sure that you run it with Windows Power Options set to High Performance every time.  Make sure that background tasks are running or at least make sure their the same each time.

I've noticed that with the Realbench test, running it the first time results in the image editing test to take forever to load up, and Realbench considers this extra delay as part of the score for some reason, so the first run will score significantly lower. Running it a second time without closing the program fixes this issue. I don't know if this may be a contributing factor, but it is always the best idea to test a second time after doing one run, at least in my case (benchmark is stored on HDD).