I've been a dual CPU nut since before it was cool with the dual cores.
My first dual board was a Tyan Tomcat III, an old dual Pentium board in which I had two Pentium 133MHz chips.
"Old Reliable" here is an ABIT VP-6 running twin 1GHz PIIIs. And it still cuts the mustard, although my other box, a single CPU AMD-64 2GHz is actually faster.
Old Reliable will soon be getting another board, and I'm going to go with some sort of dual AMD-64 (or Intel EM64T, sometimes jokingly called iAMD64) mo/bo. WHat I would like to get is a dual, dual-core, for a total of 4CPU.
MS sort of played funny. NT Workstation (which evolved in Win2K, XP, and Vista) was limited to two CPUs -- only the server line would run more, and now with a pricing tier on that. Cheapest versions are limited to 4, next price will run 8, etc. Same kernel, just a limit on how many CPUs it would use. (Linux will run on up to 32 or 64).
But the question came up about hyperthreading, and dual cores, and unless MS changed its mind, they relented that one CPU is one physical chip that plugs in a socket. So a dual core counts as only "one physical CPU chip", and so XP or Vista will use all 4 equivalent CPUs of a dual socket, dual core board.
Hyperthreading is a mere "logical CPU". THe physical CPU processes two instruction streams, using the same resources. Dual core is truly two physical instruction processing units on the same chip, although they do share some other stuff.
Because of that, there are some subtle differences in how best to take advantage of hyperthreading.
Two physical CPUs, such as my ABIT board here, will do best with each one having its own separate task, where the caches of each don't interfere with each other. A hyperthreader, using the same physical caches is different. It's best there to have them just splitting up the same task.
The Win2K kernel, while it would run on a hyperthreader as two CPUs, was not aware of this difference, and would actually bog down a bit in some situations becase of this, running much slower than with hyperthreading disabled (cache "thrashing", I believe one could call it). They made the XP and later kernels aware of the difference, and wouldn't make that mistake.
-Richard