i will give you the whole "hyper threading" thing in plain terms.
2 lanes of traffic (1 traffic cop) only 1 lane can go at a time (any given time)
Windows sends two "lanes of data or "traffic" out onto the highway. The processing point or "merging area" is managed by a hyper threading or the "police officer" he says stop to 1 lane of traffic and go to the other, he can stop and go the traffic lanes at ANY GIVEN TIME. BUT ONLY 1 LANE AT A TIME...
Now you may be wondering "this sound cool and all but it does not sound like a MAJOR performance increasing feature......"
Well you are right, on an average benchmarking matrix of non-ht vs. ht setups, sometimes the HT setup would be 2-4% maybe 5-6% faster than non-ht setups. But hey, any extra performance jumps or increases are welcomed! Back when HT when being marketed intel claimed up to 30% performance increases..... Those numbers were VERY OS and SOFTWARE dependant and rare to even come close to. Also in some areas of computing, HT would sometimes deal a blow to maximum computer efficiency and optimization.
And as peter mentioned above, to REALLY see the 5% increase (if you could notice it at all) would be in SMP (symmetrical multi-processing) programs.