In trying to justify a new system I ran a test to compare hyperthreading - on my system disabling hyperthreading only netted me a 2.4% improvement in solve time on a 1.6M cell thermal flow model with radiation and gravity (66,985sec vs 68634sec).
That being said, it would be easier to justify new hardware (and software) with a benckmark to compare. I know that Anna Wood had previously volunteered to host the results alongside her other benchmark data. Now we as Flow users just need to formulate a meaningful benchmark.
If someone wih more forum experience is willing to create a survey to tally the features most commonly used by intense Flow Simulation users or features that seem to be especially time intensive, I'll volunteer to help setup and configure a benchmark model.