Questions about CPU Benchmark values and data validation

[gannster]

I've been comparing the difference between the price of systems, vs the change in the benchmarks. With last months scores, an IBM x3650 with the E5-2665 had a %12 increase in the overall price and a 25% increase in your CPU benchmark compared to the x3650 with the default E5-2650. (21,661 last month, 21,460 today)

Can somebody explain why the benchmarks have dropped in the past 30 days? Which benchmark is more accurate -- the one shown today, or the one I copied a month ago?


Last month (around Sept 27) I made these notes:

[Dual CPU] Intel Xeon E5-2665 @ 2.40GHz
benchmark 27,150

Today, the site shows

[Dual CPU] Intel Xeon E5-2665 @ 2.40GHz
benchmark 21,074

[David (PassMark)]

The older charts were based on PerformanceTest V7 results.

The newer charts are based on V8 results.

You can find more detail of the changes here,
http://www.passmark.com/forum/showth...-go-live-today

These multi-CPU systems get their high scores as a result of there large number of cores / threads. 32 threads in this case.
But in V8, one of the nine CPU tests is single threaded. This will pull down the results a bit of these systems as for this single test only 1 of the available 16 cores (32 threads) will be used.