I'm curious as to why these Firestrike scores deviate so much. I have linked my two examples and a third for fun.
 
Score of 17,771. This is my personal high that have not been able to replicate with the past three newest drivers. Even with higher clocks around 2100mhz on my 1080. That's the highest I can get this dud. (lost the silicon lottery) http://www.3dmark.com/fs/9412758
 
The second score is MUCH lower at 15,926. As you can see the ONLY difference is the GPU core is at 2038mhz instead of 2055mhz. A mere 17mhz can drop my score over 1000 points? http://www.3dmark.com/fs/9415009

These are both on the same drivers. 368.81 WHQL.

My most recent run on newer (not newest) drivers still can't beat my 17,771 score even at a higher core clock of 2088mhz. http://www.3dmark.com/fs/10575745
 
Can someone explain these huge discrepancies and why they occur? 

I have never seen a Benchmark Return the Same Score run after run, CPU or GPU.
 
I have a Batch File that I run between Benchmarks.
 
Del "C:\Users\%USERNAME%\AppData\Local\Temp\NVIDIA Corporation\NV_Cache\*.*" /Q
Del "C:\Users\%USERNAME%\AppData\Local\Temp\NvidiaLogging\*.*" /Q
Del "C:\Users\%USERNAME%\AppData\Local\Temp\Low\NVIDIA Corporation\NV_Cache\*.*" /Q
Del "C:\Users\%USERNAME%\AppData\Local\Temp\*.*" /Q
Del "C:\Windows\Temp\NVIDIA Corporation\NV_Cache\*.*" /Q
Del "C:\Temp\NVIDIA\ComputeCache\*.*" /Q
Del "C:\Windows\Temp\NvidiaLogging\*.*" /Q
 

bcavnaugh
I have never seen a Benchmark Return the Same Score run after run, CPU or GPU.
 
I have a Batch File that I run between Benchmarks.
 
Del "C:\Users\%USERNAME%\AppData\Local\Temp\NVIDIA Corporation\NV_Cache\*.*" /Q
Del "C:\Users\%USERNAME%\AppData\Local\Temp\NvidiaLogging\*.*" /Q
Del "C:\Users\%USERNAME%\AppData\Local\Temp\Low\NVIDIA Corporation\NV_Cache\*.*" /Q
Del "C:\Users\%USERNAME%\AppData\Local\Temp\*.*" /Q
Del "C:\Windows\Temp\NVIDIA Corporation\NV_Cache\*.*" /Q
Del "C:\Temp\NVIDIA\ComputeCache\*.*" /Q
Del "C:\Windows\Temp\NvidiaLogging\*.*" /Q
 

Oh definitely, that is to be expected. I'm just curious why the difference in scoring is so vast between each run.

Before you start

"In general, you should benchmark every device you test under the same conditions. For example, you should test every system in the same location, at room temperature, and away from direct sunlight and other heat sources.
The precision of Futuremark benchmarks scores is usually better than 3%. This means that running a benchmark repeatedly on a consistently performing system in a well-controlled environment will produce scores that fall within a 3% range.
Individual scores may occasionally fall outside the margin of error since the factors that influence the score cannot be completely controlled in a modern, multitasking operating system. There are also devices that simply do not offer consistent performance due to their design. In these cases, it is necessary to run the benchmark multiple times, and then take either an average or a mode of the results."
 
- https://www.futuremark.com/support/guides
 
Other than that maybe your overclocks aren't 100% stable.