Not enough info. What are you going to be using it for?

Marketing gimmick? AMD had to cut prices by over 1/3 to tempt buyers because they suck.

The person in your youtube video was comparing the integrated graphics performance, which is not what a gamer would use. Seriously think about it, 36fps (8350) vs 21fps (3750) in Metro 2033 at 1080? He couldn't have been using a video card.

The Bulldozer/Piledriver models have been a big disappointment from AMD. And the way things are going AMD's cpu division is in serious jeopardy. And I'm saying this as a fan of AMD.

Ok. Which makes it even more puzzling.
This review is probably the only one which shows metro/warhead being faster with the Piledriver. He made a mistake with the 3570-1080 in Warhead xsplit by saying 20-something when it was clearly showing 30-something (which is faster than the 8350 in his benchmarks). That and the abnormally low framerates leads me to believe that he either broke something in the setup or accidently benched the Intel cpus at higher resolutions.

Fake or not, the AM3+ is a dead end and its highly unlikely that an 8 core steamroller will ever come out. AMD has slashed headcount twice around last year and so its going to be highly improbable that AMD can continue on the existing AM3+ roadmap.

The 7790 and the r9-290 weren't the 8 series which people were expecting but rebadged 7 series Tahiti era cards. The r9-260 will be one of the first new gcn2 cards.

Still unknown? AMD is trying hard not to say that the product line is dead.The context of your earlier post makes it sound like the FX is going to continue post-Vishera which is highly unlikely. AMD is in a worse position several years ago with less resources to devote to continuing their big core consumer line.

I made a typo. What I was getting at was that the 2xx series are very small tweaks from the 7 series, not even a real tick (using the tock tock analogy).

You sound as if all the problems that AMD is having was intentionally directed for the last 7 years to get to HSA which doesn't make sense. The path AMD is taking now is more unplanned and remedial.

Around 7 years ago, AMD had better performing solutions vs Intel. They were the first to write the spec for x64 and had X2s and could charge premium prices for them which was a turnaround from their usual low margin price slashing strategy. Then AMD messed up big time by delaying the development of the K10 which also pushed Bulldozer ahead by a year or two. AMD paid way too much for ATI, abandoned gpgpu (for HSA?) and seemed to ignore ATI.It was only after Rory Read came in and figured out that BD/PD/SD disaster wasn't going anywhere, decided to cut AMDs losses by slashing headcount and focusing on smaller FM2 APUs. The the console deal (which Nvidia rejected for being too low margin) allowed AMD to leverage its mantle api which otherwise wasn't going to happen. And its still a open question on whether the market will be willing to take on a new proprietary api.

But how was Bulldozer going to be an APU when it was already running hot at at a large chip size at what 1.1B transistors? Slapping on more transistors would make the whole project infeasible. And Bulldozer also had relatively poor ipc for its integer cores so it wasn't just a problem of having less fpus. Llano wasn't competitive with i3 in integer heavy benchmarks. Llanos fusion style gpu wasn't that useful outside of gaming either.

Bulldozer doesn't have an APU architecture. Bulldozer was always going to be competing with Intel's big core cpus.
The first APUs the Fusion/HSA line starting with the Llano which has an integrated gpu. Trinity and the upcoming Kaveri have more hardware which provides better integration of the cpu/gpu.

You're correct about the Llano using K10 cores. I got mixed up because AMD skipped over K9.

Ok I don't disagree with your there.

What I was trying to say from my first reply to your post is that they couldn't have planned for Bulldozer to be a flop with lower ipc compared with the K10 with such low integer workload performance. AMD certainly did not plan to shed 1/3 of its headcount from 2007 vs today. AMD couldn't have planned to for its market share of the high margin server segment to crash from 2006. AMD didn't plan to have to shell out big bucks to GF to reduce their purchase commitment in Dec 2012 because there was low demand for their BD/PD line. This all happened because AMD executives were making out big time back then, their employee morale was low and AMD was bleeding talent so its no surprise that their subsequent products were subpar.

The review did state that the new Kaveri was alot more efficient per clock so I don't get your point. Their previous reviews (and reviews elsewhere) normally include a low end discrete card for comparison, so its not like that website tried to paint Kaveri in a bad light.

OK I had a good look at the review and the statement you were talking about refers to the iris pro beating out the Kaveri. While it seems unfair, it doesn't spoil the rest of the review. That review website has a special section for AMD and takes money from that company. They are trying to look even handed especially after getting criticised for their method of reviewing Trinity.