Chaotic GPU benchmark results

[Aeneas]

Looking at the Benchmark Charts for Graphics cards, there is massive variability in the results for the same GPU, e.g. GTX 580 or 560, in the same benchmark test list. What is causing the chaotic results ?
How much predictability is there that the buyer of one of these GPU cards would obtain the top performance.
And are any of the listed results actually performance of an SLI or Crossfire linkage of 2 or more cards instead of the 1 expected by the reader ?

 

[Guest]

The system setup plays a part in benchmark results, significant or not. Old CPU and RAM with bad timings can limit the performance to some extent.
While most graphic related tasks are assigned to graphics card only, there're exceptions.

Any respectable benchmark chart would indicate if SLI/Crossfire setup was used, if it doesn't then it's single card setup.

True there's some variability within cards with identical GPU, mostly about its ability to overclock. I personally own a card that is massively behind the average results for that GPU and runs 100C on manufacturer setting, probably it's defective, I'll be contacting dealer whether it's covered by warranty or can be replaced.

 

[Aeneas]

Are these results generated by Tom's Hardware personnel or sent in by random forum members ? There seem to be very little detail about the hardware, software, operating system configuration or the exact video card product being used in the benchmarks reported.
Also, where would I find a benchmark test of my own Diamond ATI 4550, for comparison ?