It'll all depends on the applications you run, as different programs make use of different aspects of the sysem. The best way to get an idea of what FSB and clock speed changes do, would be to underclock your current CPU/FSB by the same percentage and do some benchmarks.
Tom's is nowhere near as good as they were with their charts. In the good old days you could look up pretty much every CPU model released (and usually before they were on the market too), and they'd say which benchmarks were affected by FSB/cache/clocks.
high bus speed increase peformance in many apps - the bus cares data from memory to cpu. this better performace is by product of overclocking, in a way its like factory authorized overclocking.
matching bus speed, memory and cpu is important so when you overclock the bus you overclock the memory
intel seeing the benfits of overclocking and faster bus speeds and value in decided to push bus speeds with better chip sets faster memory etc.
so the old ddr400 ran at 800fsb or 2:1 ratio, the ddr2 pushed the speeds higher and allowed higher bus speeds
ddr3 now runs 1600+ or 4 times the old ddr400, memory has been redesign with lower voltage to keep it from burning up
so a stock 1333 part should be faster at the cpu speed as a stock 1066 part, now we have 1600 parts.
sticking a lowly 800fsb cpu (such as the 2180) in a fsb 1600 chipset - such as intel p35 makes for fun and low cost high speed peformance.
you still need to match your speeds at 1333fsb part on a 400bus need a memory ratio of 2.5:1 for 1066 memory or a 2:1 ratio for 800 memory.
lower latency ram with highest fsb chip set and cpu producese the best results when you have whole number ratios. regardless the multipler is the same with the 1066 part or 1333 part when the bus is changed. a 2.66 1066 part or 2.66 1333 part have a different multiper but both can be clocked to around the same 3.4-3.6ghz
Q6600 with a 1066 clocked to 3.61ghz or fsb 400 runs quite well with ddr800 4-4--3-10 timings, or you can run 5-5-4-12 timings with 1066 ram. the ddr800 ram creates a latency since the cycles are slightly out of step but the lower latency ram compentensates.
ddr3 meory at fsb 1700 is 1:1 with a 1700fsb, the cpu at 1066 or 1333 requires the same multipler when sync'd at 1700fsb. latencies of 7-7-7-18 with ddr3 give a big boost in performance
i did crappy job of writing this someone will explain it better -!!!! it may help some
would be better 2 quad core processors (total of 8 cores) that have 2GHz or one with 2.6Ghz?
obvisouly for multitasking the 8 cores would perform better, but for single tasking?
I mean, can the 8 cores be utilized for one task?
My first thought was no, each core needs to have it's own task (thread)
That is not what you ment though is it? Many tasks have multiple threads that can be run concurrently.
Those are mostly programs that have a use in servers. Some have been developed for use in pcs.
Photoshop is a good example.
You have to remember though that most computers do more than one thing at a time. While you are surfing the net, your computer is running anti-virus, maintaining a firewall, doing spamblocking, maintaining windows, playing music, listening for IMs and trying to find the sight you want. If each task had it's own core things would go much faster, and M$ would find more things for your chip to do at the same time.
It all comes down to what task? Right now, for most people, they wouldnot be able to notice any difference between a dual core @ 2.6, and 2 quads @ 2ghz.