When you consider multi-threaded apps, like kewl munkey mentioned, you'll have to determine performance differently than in simple multitasking terms. When looking at it this way you can add the cores up, in a sort, though not all SMP code is the same, and most SMP code is poor, so the end result tends to be most discordant.
Too bad we can't chart that up, it would require more than calculation, but active testing of specific platforms running specific code.. I regress, the calculation of the c2d chips as 20% more efficient clock for clock seems about right, according to the benchmarks I've seen. I've seen that number grow to as far as 30% as it is highly subjective towards what's being run. The wolfsdale should have a slightly better core architecture as well, elevating it higher than it's older kin.
|