Does anyone know a good empirical way to compare RAM vs processor speed and chip architecture?

Does anyone have an empirical method of comparing the tradeoffs regarding RAM vs processor speed and architecture?  I'd like to be able to go to my management and give them hard numbers rather than a feeling.

There is roughly a 1GB - 1.5GB effective RAM limit on 32 bit. That is, addressable memory. For more on this, search the web for 32 bit ram limit. On 64 bit, your application may use as much of the 6GB RAM as the system allows.In fact, you may raise the RAM even higher on 64-bit.

Furthermore, CF 8 is 32-bit. Sixty-four-bit Coldfusion starts from CF8.0.1.

It is a good plan, but in my experience the numbers are never quite so cut and dry. I am a newbie with regard to virtual machines, so I am sure someone will correct me if I am wrong, but the shared nature of a virtual machine means that the resources at your disposal will be constantly fluctuating as the other virtual machines on the shared server demand and release resources.

What you do have going for you is that administration is often much more willing to put a ton of memory/processor power into one machine running virtual environments rather than upgrading a bunch of separate machines. What this might mean to you is the ability to run on a server with 64 gb of ram (shared among three or four virtual boxes) rather than 8 gb on one single server.

There really is no single right answer. The unknown factors not revealed in your post make it quite challenging to formulate an assessment.