The problem is you keep polarizing it. People need good performance from both. you can get great CPU performance from SB, and terrible GPU. Or you can get good performance from Llano, and good GPU. OEM's and business love that. They will lap it up. I'd not be at all surprised if Apple drops SB for Llano for their next generation.

The problem you have is you have determined, without using or seeing independent results, that Llano CPU sucks. It doesn't. Why? Because Phenom II's don't. What the difference is, your definition of the word SUCKS means 'not the fastest in the world', where everybody else uses 'offers good performance and value for money'. If you're uncompromising, SB is your CPU, and you should be complaining you had to buy a waste of transistor budget iGPU when you have perfectly good dGPU's. If you're looking for bang/buck in a budget, which basically means everyone else especially business, then Llano is a no brainer.

Not everybody who games, or even primarily games, is an ultra enthusiast. The market shows that. Buy a dGPU? Why, you get a competent one in the llano, perfect for today's popular screen resolutions like 1366x768, 1680x1050, even 1920x1080. At ultra enthusiast settings? No, of course not. Just because someone doesn't spend $1K on dGPU's doesn't make them any less of a gamer. It makes them less spendy.

As I said, nice "Glass is half full" position, I salute your positive spin on the state of AMD processing power.

The focus is shifting from cpu to platform capability. The means CPU + GPU. If it means something that Llano isn't faster than SB in CPU specific benchmarks, why doesn't it matter than Llano is faster in GPU specific benchmarks? Why isn't total platform cost a consideration? Because you don't swim in this end of the pool, you don't care about it and it changes what you think is important. Does that mean it's important for everyone? No. That's my opposition to your statements, you don't consider the people on smaller budgets and looking for a different definition of value. The platform as a whole is the argument for APU's.

The essence of the argument is no different from single core at super high clocks vs. multi-cores at lower clocks and using multi-threaded software. We saw how that worked out.

So how long have you been benching the new Llano's/BD's before everyone else to come too such negative conclusions? Just exactly what games/programs can't Athlon's run according to your high standards? :hmm:

It's public knowledge Llano's CPU cores perform on par with today's current, marketed architecture. Zero advance. And AMD is famous for trying to compete in an arena, finding the arena too hot for it's tastes, and then trying to change the rules to better suit it's inability to compete in that arena anymore.

Some call it flexibility. I call it giving up and throwing in the towel.

I'm not debating from the joe-six pack perspective here, you're absolutely right, but it was never my point. Right now, and for the foreseeable future, AMD is putting too much emphasis on relying on their iGPU, without enough tasks to make that reliance reasonable.

Sorry, I don't see AMD's platform as balanced at all. I see it as a econo-gaming GPU being shoehorned into a CPU to make up for the CPUs lack of horsepower, and a desperate hope AMD can push enough GPGPU development in areas it may find quite difficult to penetrate.

I see too much riding on hope with this APU's future, and not enough riding on real workloads.

I think you're missing the big picture of all their products, existing and predicted to launch this year, and lack perspective of what actual workloads and use cases exist.

And that is a battle that AMD have to fight, perception of what is important in PC's. Intel market CPU as being be all, end all. NVIDIA markets to GPU being the savior of computing, and are ditching x86 for ARM as the platform under their GPU's. AMD are promoting balanced approach of both, because they have both technologies.

You can call it throwing in the the towel all you want, but two new CPU architectures, a new APU concept, and the best discrete GPU line-up doesn't sound like quitting to me. If it is, Intel and NVIDIA better be glad AMD are throwing it in, they'd be dead if otherwise.

BTW, it is not public knowledge that AMD's Llano cores perform on par with today's CPU's, it is speculated and all AMD have said is it uses existing x86 technology. Let's not forget that Llano features improved turbo core, too - something that only the Thuban x86 cores have in the current x86 lineup.

no one buys an i7 and uses the onchip GPU....
And who the hell buys a high end i7 for HTPC use......?

And when has discrete GPU ever mean a bad thing....?

And its clear where the CPU performance of those chips lie in that graph....cut out the GPU part of the graph on ALL the products and BD compares favorably with SB......

I don't see that balance. I see AMD leaning on GPUs to make up for CPU weakness. I see too much focus on the GPU going into it while letting the CPU execution capabilities stagnate. I also see AMD trying yet again to try to change the rules in their favor, but by over-pushing the GPU into places it won't necessarily be well suited.

that only applies to the APU's since AMD knows it has no chance for its PII CPU cores to compete with intels SB cores.

In the case of Intel you get great CPU performance and crappy GPU performance. AMD you get good CPU performance and good GPu performance, a better balanced scenario. Its not like you're getting celeron levels of crap CPU performance.

Now if BD's CPU performance is on par with SB CPU performance then thats a good sign. Of course they could be late to the game like the PII (The barcelona that should have been) since Ivy Bridge is just around the corner.

Good CPU performance? Yeah, good for circa 2010. Good enough is NOT an area I have any faith in AMD's execution of anymore. Good enough was what kept them riding on their laurels while Conroe was kicking every dual core AMD had in the butt. And that's not the only time.

I'm sorry, I see a repeat of history that hurt AMD last time they pulled it. Whenever AMD tries to ride on CPU performance laurels, they get screwed.

I don't see any evidence of riding on laurels, I think you are borrowing trouble to support your point. You seem unaware of the prevalence of software that already uses GPGPU capabilities, which is going to explode once APU's get mainstream, which is this year.

So that would mean your i7 920 isnt good enough for anything anymore right?:nuts:

As previously stated, if 'just around the corner' here has the meaning of '2H 2012'. AMD will have product in 2012 as well.

Well, fair enough question, lets review it! It's now been relegated to server duty, doing little more then a Celeron would find easy pickings. Replacing it with my i7 2600K is a major upcoming planned move when socket 2011 launches.

Wanna buy it? :evil:

Kinda sounds like it doesn't matter what AMD is planning or releasing for you, you've made your decision sight unseen. Unless you meant you would decide as to whether to buy an AMD platform or an Intel one, at the time of socket 2011 launch?

In fact, unfortunately and as much as I hate to say it, but I think you're right (not that I hate saying your right normally, just this time specifically). I kinda have made up my mind the moment I saw "similar per core performance to Phenom II" coming out of AMD's own leaked slides. I have never, and will never, buy the previous year's performance in any upgrade.

This is NOT to say I'm not leaving room for a surprise upset in my purchasing plans, so you would be right on that point as well. But I'm NOT expecting one, at all. I do NOT foresee AMD having anything remotely close to competing with socket 2011 based CPUs unless they've hidden some totally badass, and UNLEAKED as yet super-monster-beasty-chip. I call Sandy Bridge-E a "super-monster-beasty-chip" on paper, and am quite comfortable that the reality will be very close to expectation based on my current experience with Sandy Bridge.