Processor performance realistically has little weight on the overall performance past throughput of instructions and delegation of tasks.
So while sure, the more money you spend to get more cores will improve performance to a degree simply given you'll have new threads run in a serialised form across all of the cores; despite parralisation being a better option as it make better use of the multi-core aspect this has to be done application specific.
Same can be said about the raw performance.
I've done this analogy before, but think of the processor as the engine of a car. Sure a bigger block, more cylinders, etc... will mean a better torque. The trouble is it does you no good if all that performance means is your tyres will keep popping.
Putting together a "fast" computer is realistically as easy as putting the best possible components in it your money can afford. That doesn't mean it will be particularly good at doing tasks properly.
For example a Drag Car is a damn fast car, but try turning a corner and you find you need to slow everything to a crawl in order to do this.
It's finding a decent balance, and parts that work together that you'll be able to have something that specification wise seems quite modest to actually out-perform someone with more money than sense.
Simplist thing you can do is basically making sure you have a processor, graphics card and memory that all can play well together. You delve deeper in to this tuning though, and getting those parts that can sync their bus speeds to the same multiple ranges; and you'll notice a huge performance boost from eliminating the usual suspect bottlenecks.
While multi-cores for example are cool, right now they're still fairly young as a technology and still rely heavily on the developer taking advantage. This isn't to say they're a pointless technology, but 2-3-4-8 Cores... realistically makes very very little difference unless you're encoding audio or visual feeds.
I'll use my own computer as an example, given it is quite finely tuned anyway but more so because the hardware was bought specifically to work together.
The system I'm on right now might seem modest:
AMD Athlon 64 X2 Dual Core 3800+ 2.0GHz 1MB Cache
AMD Series-7 750FX Motherboard
Crucial 2GB (2x1GB) 800MHz PC2-6400 CL5 Memory
AMD ATI Radeon HD 2600 XT CrossFireX 256MB 800MHz (Dual Card)
Western Digital 250GB SATA-2 16MB Cache
This system gets a nice respectable 4.8 Vista SPI, limited only by the Processor; which it believes is a weak-point.
Now a beautiful part of the setup, firstly is that everything inside works on multiples of 200MHz FSB. This is quite an important fact as this sets the system to run syncronised from the start, meaning there is very little downtime waiting for cycles of each component to finish and sync data again.
Another important thing is that memory bandwidth is not stretched to close to the limitation allowing it to really utilise what is available.
Something else to note is the Memory setup. I could've bought a single 2GB stick for the same prices as the two identically sized sticks; it might not seem important but memory is designed to work in pairs, just the way it's always been. Although sure now you do have Dual Inline Memory Modules no longer requiring you to pair, fact is that the majority of memory going in computers now is actually SoDIMM; meaning while designed to run as a single inline, you will see performance from adding a second identical board.
I could also go in to how it can also help the CL performance by allowing memory passes on both incoming and outgoing calls; but meh.
Another thing to really note is that having more than one provides more than one memory area that can be accessed at once. This leads to quite dramatic performance increases.. and possibly one of the more important aspects as well that's what gets you closer to that theoretical performance the manufacturers' claim.
Now another point to note is the Hard Disk size and cache size. Again these are quite important (same with Optical Media Drives) is that you want more Cache even if it is at the sacrifice of raw performance or space. In-fact for Hard Disks, more space per disk = poorer performance.
Think of the Hard Disk as a library, and the cache as the trolly you have to take the books to the front desk. No matter how quickly you can get those books at, if you've got a pathetic trolly then you're going to spend ALOT of time taking the small collection you find back'n'forth.
Also if you have a much larger collection, again no matter how well organised which honestly with Windows as the librarian often means books aren't always where you expect them or in any order that makes sense over time... the reality again is more books = more search times. As such a smaller drive frankly performances better as the data is closer together and easier to keep in check (with defraggers).
Realistically the static media of optical media, ends up honestly being a quicker way to keep data as it never fragments and is a fairly small space to search. As such actual data search and retrieval is extremely small. I mean there is often a difference in the "number of employees" so to speak between the two meaning they both are better are different things... plus performance/bandwidth is greater on one over the other, but when accessing constantly hard disks end up overworked thus fragmentation thus eventual slowing.
This might seem like some stupid over thinking just for building a computer, but you consider it is running technology that realistically is now completely out-dated; yet by simply making sure parts work together providing the correct bandwidth and syncronisation, there are very noticeable performance results in and outside of games.
One I'm quite proud of is despite having several start-up programs:
Messenger, Media Center, Autodesk Backburner, Steam, Live, Live Mesh, Visual Studio Development Environment, XNA
All on boot... yet being able to go from a cold boot to a useable machine in under 20 seconds. Compared to adverage systems even from fresh installs (this has had Vista since release installed) can take up to 60seconds (or more).
That is a noticeable performance increase. The same spills over in to games as well, with games that are often requiring more performance for recommended; running at acceptable playable performance rates with graphics maxed.
You'd be surprised how much of a difference keeping in mind the small easily overlooked aspects of the system can be.
