dear me, Bulldozer does well in some things, poor in others, how about we leave it at that...? also be careful where you read reviews, some of them are shocking bad, Toms Hardware being an example where the review is designed to make Bulldozer look as bad as possible by not adding any benchmarks that might potentially favour the modular, many core, many threaded architecture.
below listed from the OCUK Forum holy grail site (Anandtech):
CB11.5 single threaded test = beaten by 2500K.
CB11.5 multi-threaded test = defeats 2500K
x264 first pass performance = beaten by 2500K.
x264 second pass performance = defeats 2500K.
7-Zip benchmark = defeats 2500K.
AES-128 benchmark = defeats 2500K.
faster than 2500K four times out of six, pretty relevant considering the sort of test it performs better than the 2500K in, those that are multi-threaded.
in-fact so far as I have noticed, most things that aren't games tend to do quite well on Bulldozer whenever the thread count heads upwards, also let us not forget the competitor to the 8120/8150 is the 2500/2500K, not the 2600K, that was never the intention, look at their slides for where Bulldozer is positioned and the 2600K is up top, all by its lonesome with no direct competitor. the competition = one Bulldozer module vs. one Intel core.
in the grand scheme of things, for a processor that is fabricated on a less than perfect, immature process, with cache latency troubles, inefficient programming with 33% less resources in each 'part-core' it performs reasonable, based on those sorts of comments, Bulldozer as an architecture = winner, as a product not so much, not at the moment at least.
know immediately that someone will say 'those benchmarks favour Bulldozer so it isn't valid...' but then 99% of the anti-Bulldozer arguments feature swathes of gaming benchmarks, an area that it is not designed to perform 'well' in. given the fact most games are bottlenecked through graphics, and the world is becoming ever more threaded suggests that the principle of 'Bulldozer will be slow, out-dated before X Intel processor' might not be so true, I for one look forward to seeing how it goes, many cores, great parallelism vs. less, strong cores, am 100% sure the trend will continue with one winning some benchmarks and the other winning the others, its more competitive than K10.5 ever was just so happens the area where it lacks, single-threaded, games in particular is the area where K10.5 is consistently stronger.
