[citation][nom]gnesterenko[/nom]For full explanation as to why, see a rather in depth article found here: http://techreport.com/articles.x/21865That said, Bulldozer is to AMD as Nehelem was to Intel. Great for server architecture - useless on the desktop. Only difference is that, althought it is uselss on the desktop, you can still put one on there at a reasonable price. In Intels case, you could spend $600-$700 to put a server chip on your motherboard, with about equal effect in terms of performance. All that said, I'm sticking to my original plan and building a Phenom 1100T system. One generation back, mad cheap, equal performance to current gen offerings at a fraction of the cost. Considering my almost 8 year old 939-socket Athlon 64 X2 (2.2GHz/core) is still good enough to play modern games with (paired with a modern GPU of course), I am pretty happy with the fact that I can get an exponential boost to processing for under $350 (mind you, thats for CPU, motherboard AND RAM combined). Intel did a great job with their 2600K and 2700K chips, hands down these are the best peformers out there today. But marginal performance difference? Worth the extra few hundred bucks that the platform will cost? That's a question only your budget and computing needs can answer. For things like Crysis though, much more benefit from spending money on a dual GPU setup then worrying about which CPU performs better in which synthetic benchmark.Posting from work, so need this disclaimer:"The views expressed here are mine and do not reflect the official opinion of my employer or the organization through which the Internet was accessed."[/citation]
That's a decent performance gain for BD gain from assigning affinity to certain cores. The problem is, by the time Windows 8 with fixed BD scheduler hits the market, Piledriver cores should be out. Therefore, buying BD as it is right now makes no sense (except for specific multithreaded applications and Linux which can have a BD friendly scheduler redone much faster than Windows).
That's a decent performance gain for BD gain from assigning affinity to certain cores. The problem is, by the time Windows 8 with fixed BD scheduler hits the market, Piledriver cores should be out. Therefore, buying BD as it is right now makes no sense (except for specific multithreaded applications and Linux which can have a BD friendly scheduler redone much faster than Windows).