The fluctuation isn't that much and its hard to compare when you run anything but the default test.
First thing is to check what you have running in the back ground. Something simple as indexing service can really wreak havoc on scores.
Shut down AV while bench marking and all other apps, your task tray should be virtually empty except the essentials.
If you are running GPU tweak it has a feature where you can maximize the performance by adjusting the display setting, shutting down unnecessary services and clearing unnecessary junk from your ram.
Have a look at task manager before hand and see what your CPU is and whats running. 3D marks wont stress say a 10 core machine as much as it wont use more than 4 cores but for a 4 core having anything else consuming clock cycles will change your scores more. For your own troubleshooting and to shorten testing custom is great but for reaching out elsewhere you should really run the default values.
You can toss out your second run as its invalid, look like it was interrupted as Graphics and physics are 0s
Then compare the Grahics scores and physics score and they are well within the margin of error of only a few points.
Please run the default full test.
First thing is to check what you have running in the back ground. Something simple as indexing service can really wreak havoc on scores.
Shut down AV while bench marking and all other apps, your task tray should be virtually empty except the essentials.
If you are running GPU tweak it has a feature where you can maximize the performance by adjusting the display setting, shutting down unnecessary services and clearing unnecessary junk from your ram.
Have a look at task manager before hand and see what your CPU is and whats running. 3D marks wont stress say a 10 core machine as much as it wont use more than 4 cores but for a 4 core having anything else consuming clock cycles will change your scores more. For your own troubleshooting and to shorten testing custom is great but for reaching out elsewhere you should really run the default values.
You can toss out your second run as its invalid, look like it was interrupted as Graphics and physics are 0s
Then compare the Grahics scores and physics score and they are well within the margin of error of only a few points.
Please run the default full test.
lpiotrowski211 wrote:
I turn off all background programs during that runs.
Your screenshots are showing differently. The only thing missing from the equation on the shots is what the CPU is doing assuming you have the GPU doing the physics computations. The first is the example that doesnt count. then look at the other two that are spread by 500 points or so on the combined, the graphics and physics are dran near identical which only leaves what your CPU is doing to affect the combined. I can have the same set up going with background apps and services not running and repeat the default results within a margin of error of maybe 2% for a 100 runs.
As suggested please do default runs and a screen shot of task manager for processes and services.
Here is a shot from my 6700K rig
