Which would give the greatest / best performace for use in Mathematical modelling? Speed is paramount here. An i7-3930K 3.20Ghz with six cores - LGA2011 & 12MB L3 cache, or a Xeon E5-2630 2.30 GHz, with six core Tray , LGA2011, 7.20GT/s & 15MB? Thanks for any advice on this.
The i7. It has an unlocked multiplier meaning with sufficient cooling you can really clock it up, which should help with your maths.
Thanks for your help. Without using water cooling, what sort of speed (Ghz) would you reckon I could get? I have read that the Xeon has been used to build large configurations/machines (e.g. NASA) the implication being that it is inherently a faster CPU despite it's apparant lower Ghx than the i7. Is that in fact the case, though?
They are the same CPU basically. They come from the same wafer. Higher clock speed equals higher performance.
The reason xeons are used is because they can be linked together and they support error correcting memory.
Only certain Xeon models support multisocket configurations. The E5-16xx models only support single processor systems while the E5-26xx models support dual processor systems. The E5-46xx models support four or more processors. Unfortunately price goes up quite a bit when you want a multisocket system (even though all the chips come from the same wafer).
To add to the above, it all depends on how many CPU cores you want (or can use with your chosen application software). Anything below 6 cores and you may as well buy an i7. Above that, you are looking at high prices and esoteric components: not just the Xeons, but also server-grade multi-socket motherboards which may not even have an X16 PCI-E slot for a decent graphic card, and ECC registered RAM which comes at a considerable premium over the regular version. I know of which I speak. I have a dual Opteron system from the days when CPU's came with only one core. Although the dual-core buttery smoothness was definitely worth it, and it can still hold its own with current systems, the esoteric motherboard and memory made it a pricey affair (not to mention the demands on cooling: 178W for the CPUs alone!). Now I can build a comparably performing system for a quarter of the price with a vanilla components and an Intel Core Duo running at 65W tops --35W if you use a Mobile Core Duo. My passively cooled HTPC (Mobile Core Duo on a second-hand AOpen mobo) runs as fast as my liquid-cooled Opteron system, for a fraction of the price and at a sixth of the Wattage CPU-wise.
You will need H2O cooling for a 2011 cpu, expecially if you are overclocking. How many cores does your software use? You might be better off with a quad core with hyperthreading at 5.0ghz on the z77 platform which would be much cheaper and probably faster in more tasks than a stock clock 6 core Xeon.
Thanks to all of you with your advice iI shall study it closely, there is a lot to take in and then answer any of your questions. Thank you again.
How many cores does your software use? There is no limit within present M/B constraints, in theory it could handle over 30 probably more and multiple CPU's as well. As you make the point, cost is a factor. Thanks for your suggestion, I will very much ponder that one.