System Benchmark Oddities

I often refer folks building up machines to the latest benchmarks posted by PassMark Software. For whatever reason I happen to get the wild idea to download their software (Performance Test 8.0) to benchmark my machine (which I have never done). When the software is first opened it examines the system and my graphics card starts squealing... I notice in the lower right of that screen FPS: 5000! No clue what that was about, but anyway I tested my system and everything in it benchmarked significantly higher than they have on their benchmark page.
 
The two really odd ones were CPU and GPU:
             Their benchmark     My test                Comparison to fastest reference (they chose)
i7 2600K     8527 (OC to 3.8GHz) 10919 (3.4GHZ no OC)   4% faster than the average i7-3770K
GTX 580 (3D) 4688                5553                   4% faster than the average GTX 680
2D Graphics  not posted          923                    17% faster than the average GTX 680
 
Ironically, everything benchmarked significantly higher except my SSD (my Patriot Wildfire benchmarked 20% slower than the 240GB Intel model).
 
Anyway, I just found this weird enough to post.
 
 

I've been benchmarking computers for 30 years - performance profiles used to be part of my job - and I've never given much weight to absolute numbers. There are so many variables, and small variances don't tell you much about what you can practically do to improve things. Mainly I'm looking to identify the weakest component (singular), the biggest single bang for the buck upgrade, the most significant bottleneck. And, of course, identify things that are plain broken.
 
A disk drive that returns unexpectedly low numbers can be addressed a number of ways, from load balancing to replacement. But if it's 4% slower than the manufacturer says it should be I wouldn't even consider that statistically significant as long as it's not taking excessive errors that might suggest a future failure. The important thing is whether replacing it would improve overall system performance more than any other action. 
 
At the end of the day, the system can either do the required job or it can't. If you can record and play back what you want, having a faster video card or disk drive won't change your experience.