I was searching the web for benchmarks (but couldn't find any) that would indicate hardware performance (probably cpu in particular) in different GIS software in time consuming tasks (like hillshade generation, raster reproject, topology buildup), not for the purpose of comparing different GIS software but to compare performance for different cpus, multithreading impact, ssd, etc. If i started with that, I suppose i'd need a standarized set of data, publicly available so anyone could try and post results. But does anyone know if such comparisons already exist anywhere?
I know that in the past our organization has made a custom suite to benchmark different hardware setups for GIS, so I would assume that (back then, at least) there was nothing that fit the bill for comparison.
As SP11 mentions, you should be specific as to what you want to compare. I would assume it is likely just a small number of different hardware configurations. In that case, just adding some benchmark logging to a set of standard sample tasks would probably give you enough information to make a good decision.
And then, once you've done that, you can post the results on the web for everyone!
At FOSS4G they perform a Benchmark test every year against various WMS Servers. Being in the Open Source spirit, they use free datasets, document everything, and release the source code so you can probably pick what you want from it.
You can find out a little more from the wiki page: http://wiki.osgeo.org/wiki/FOSS4G_Benchmark
Unfortunately it looks to be a poorly maintained wiki and many pages don't like to other relevant pages so you'll have to search a bit to find everything (its scattered around). http://wiki.osgeo.org/wiki/Category:Benchmarking might be a good place to start.
They also have results where they discuss things like what was limiting stuff, including hardware, i.e. from slashgeo
Actually its very good question..I think its related to load balancing of various GIS tasks..
I hope above points will give you the direction..