overlooking bus, lithography, cache, power consumption, and other things that might or might not make a difference to you depending on how you use it... and that will be determined by your choice.... the most significant differences to you would be: - bandwidth (Memory, PCI, etc.) - reliability (ECC, etc.) http://ark.intel.com/compare/65719,64621 and you can have multi-processor with the E5 class... more important if you are doing modeling and need maximum number of threads... otherwise, it is just bragging rights and nothing more...
also, when deciding which CPU, if you know what your application workload characteristics are like (single thread vs. multi-thread) you can use this benchmark as a baseline to help you choose.. http://www.cpubenchmark.net/singleThread.html
IMHO this is about 400 bucks worth of hardware your trading application will probably never use Unless you wrote the application yourself
This probably explains why my Acer is proving to be more trouble than it is worth. Ah well, we live and learn. Scataphagos, ofthomas, NoBias - thanks for the patience and for going to all that effort to help me, I'm sure others reading will find the advice useful too. I have a tonne of info to digest and research to do before I can make any decisions, but off the cuff I'm thinking maybe get a ready built box for the main machine, use the Acer temporarily as the 2nd machine, and learn how to build one meantime. That way if I get stuck with the build, everything still moves along fine. I like the idea of learning how to build, because then I can fix them myself and not be dependent on service centres that take forever.
I think its better to keep simulations off the trading machine. (Again, IMHO) The box I spec'd is a very effective trading machine for 400 bucks I have built several on this spec for other traders - so far positive feedback. I dont know much about Excel modeling volativity surfaces so I can's speak to that. With other Excel work it matters alot how you code it. Using tables or CSE functions in Excel will grind any machine to a halt. Pivots are more efficient but you cant do everything with Pivots Monte Carlo simulation is sort of the same thing. I can wear out any box if I select a large enough # of simulations and a large enough # of trades per simulation. just my 2 cents