Well said.
Computer shopper recently ran their tests, and to summarise, they got best performance running several benchmarks at once, only time they could get bd to beat a 2500k.
Basically they summarised that you would have to have need of encoding two videos at once whilst watching a third regularly to actually find benefit in bd.
Maybe by windows 8 we shall all be doing this, but i doubt it.
hate to tell one this but the 8150 beats the 2500K in a number of tests, and not just ones that involve running version iterations of the benchmark at the same time, does even better in Linux beating the 2500K in the majority of multi-threaded benchmarks, but just like Windows based tests it disappoints somewhat in single-threaded performance, but you can't just say that it is a terrible product because it doesn't win everything. it is exactly what we always knew was going to be the case, Bulldozer simply is not designed with 'legacy style' single-threaded code in mind which is shown by the benchmarks, however in more modern code that makes best use of its attributes it fairs very well. you cannot call it bad because it doesn't do something well which it was never intended to do well!
also love how 'clear cut' this is for everyone, not even considering Intel shady past, including various things they have been sued for in the past. also this low instruction per clock stuff is balls as well, since instruction per clock depends as much as what code is being ran on the processor as it does how the processor is designed, there are situations where AMD processors fare better than Intel processors and vice versa, which is almost entirely based on the fact that they are different at an architectural level and do things in subtle but different ways. look at Super PI as an example, people are always quoting that benchmark even though for an absolute age it has been a very Intel favourable benchmark? the 2500K does 32M in less than half the time it takes a Phenom II 1100T, is it twice as fast, is it balls!
there are tons of applications that show one architecture or the other in a 'favourable' light that doesn't at all, in the slightest way represent actual performance, like in some tests on Linux where the FX-4100 is as fast as a 2500K, does that mean that the FX-4100 is the dogs balls? no it doesn't, it just means that certain batch of code operates well on the FX architecture, again there are some tests where the 8150 is as much as 30% faster than the 2500K, but again does that mean that it is 30% faster? no, again just certain codes, could go on forever but don't think it would make a difference to opinions. its easy to cherry pick a benchmark that would show Bulldozer destroying a 2500K, just as easy as it is to show a one where the opposite happens (i.e. gaming) which is the one everyone brings up, doesn't make a difference that the architecture is generally strong in other tasks.
Although I'm sure they'll end up with their own variation
.
