The i5 isnt going to happen anytime soon - it is way over my starting budget. Listen to me here... I am currently running a Pentium D 925 which as you know is the definition of ancient, it will go under full load just from loading up chrome. I think that might be the reason I am looking for more threads than I actually need... Could it be that my low end computer history is telling me to get more threads instead of single thread performance? But on the other hand i wanted to try some overclocking as i have been missing out on all the fun... Is it true the i3 will OC to about 200-300MHz more than stock? Am i being really stupid for not getting an i3 instead?
In the simplest terms, a processor's performance is based on how fast it executes code.
For one single core, the speed at executing code is measured in IPS(instructions per second), which is the product of clockspeed and instructions per cycle(IPC). In equation form, IPS=IPC*clockspeed. Clockspeed is arbitrarily set and varies little. A whole host of factors affects the instructions per cycle, both hardware and software affect IPC.
When another core are added to a processor, the total IPS of the whole processor is no longer the equation above. Now, I'm not sure about the precise details, so the following explanation is not 100% accurate. IPS(total)= one single core IPS + proportion of the second core*(one single core IPS)
However, utilizing multiple cores is dependent on the developer specifically writing code that takes advantage of multiple cores. If the programmer does not do so, the 2nd core contributes almost nothing* for the application itself. Now, if a program does take advantage of more than one thread, that does not necessarily mean that it will scale perfectly and take full advantage of the other core. It might, like with video encoding, but other programs will might only partially take advantage of extra cores.
Obviously, adding another core can increase IPS for the whole processor, but the total IPS can be improved by increasing the IPS of each single core. Clockspeed is one component on a single core's IPS.
So, when overclocking a CPU, what you are doing is increasing the IPS of each single core on the processor.
http://www.overclockersclub.com/reviews/amd_fx8350/7.htm
The overclock in that link had the FX-8350 overclocked to a crazy 5.2 Ghz with water cooling. At that clockspeed, its overclocked FX-8350's Cinebench Single Core result is similar to a 3960X at stock.
Increasing IPC can improve IPS of a single core. Hence, even a Sandy Bridge Celeron G530 outright crushes a CPU like your Pentium D, even if one core on the Celeron is disabled.
Interestingly, as I look at Anand's Bench results between the i3-3220 and FX-6300, the benchmarks like Cinebench clearly show the i3 has superior single core performance. However, lightly threaded benchmarks like the Windows Media 9 encoder or the x264 1st pass benchmark, the gap is barely existent. This might be Turbo Core, CPU cache size, platform differences, and/or something else.
* almost nothing because a single core can be "stalled" at times. This is why my computing experience on a Prescott Pentium 4 is smoother when I enable HyperThreading on it. This is because sometimes multiple separate programs are waiting for threads to be executed and the OS task scheduler can distribute threads from different apps to different threads/cores. If there were only one program not coded for multiple cores running, the the 2nd core contributes totally nothing.