Vishera is only 10 months old
Where's the announcement of a successor? How long would it take for a successor to arrive? Historically, FX processors have followed APU launches, so that leaves us with an awfully delayed successor to Vishera, even if it were to exist.
And er, how's Kaveri coming along? That one will be close to 18 months by the time we see it, if not longer. Unless you're counting Richland, that is.
ATOM is in the same segment, i dont see BayTrail blowing Kabini out of the water.
Can you show me an example of Kabini running in a smartphone?
Bay Trail doesn't beat Kabini in performance bacause it isn't supposed to. Sub 10W processors aren't supposed to beat 25W processors. The fact that Intel will be coming so close is not good news at all for AMD
And what this has to do with Kabini vs Baytrail ???
I really wish my valid comparisons wouldn't go over your head...
I hear this since Llano was first released, next year Intel will blow AMDs iGPU out of the water.
And I heard Bulldozer would blow Intel out of the water.
What others have said don't apply to what I have said.
Three generations after that and at the same segment Intel still falls behind with HD4600.
It's all a matter of perspective. Intel's HD 5000 blows Kabini out of the water within the same power limit. That's pretty amazing, considering how much larger Haswell is.
ps: Can you now explain why do you thing that Kabini is a disappointment ??
Sure.
__________________________
The Jaguar architecture is a disappointment because it is a very uncompetitive architecture for tablets.
It is a two-wide OoOE architecture, however it only clocks in at 1GHz. Now, we all know that GHz aren't everything, however Silvermont has almost twice the clock speed and is also a two-wide OoOE architecture. At similar TDPs, Silvermont will easily beat Kabini's CPU performance, and I'm curious to see how the graphics side of things look when AMD has their hands tied behind their back with 300MHz or less GPU clocks.
If we cut Intel out of the equation entirely, AMD still doesn't look good. GCN in a tablet does sound interesting, but its compute abilities are of questionable usefulness. I can't really comment on GPU performance, but I don't expect them to come out on top. They should be pretty good for what they're going to be used for, though. However, their CPU is not going to be much to look at when it's hamstrung at 1GHz. Their one redeeming factor is their instruction support -- AVX is pretty nice, however I believe that is the only instruction set they'll have on Silvermont.
AMD, for now, is tied to Windows tablets. I know Android tablets and such are coming, but how far away is that? Windows tablets aren't necessarily the cool place to be right now, and they'll have to go up against Intel's (likely) better Atoms.
A piece that frustrates me about AMD, across their entire x86 and praphics portfolios: turbo boost is not ubiquitous. It needs to be, desperately. It's a tremendous advantage, and they've missed countless opportunities to capitalize on it. They need it on their GPUs. They need it on their CPUs. To their credit, they offer it on pretty much all of their Piledriver APUs and their FX processors. However, in the APU that they need it most in, they only have Turbo enabled on one SKU.
Now they did launch a few more Jaguar-based (I think?) APUs with turbo enabled recently, which is good; but why is it not enabled everywhere?
_________________________
Other than their turbo issues, and my opinion that Temash is absolutely not cut out for the tablet world, I do like Kabini. Now, as you can tell by my sudden change in heart, I've only developed this opinion very recently -- probably because I was looking at the wrong reviews.
The really good: Console wins are an obvious thing to put here. The best thing, far and away, is that AMD should find a cute little niche between Atom and Haswell. Unlike Brazos, Kabini "craptops" will be actually usable, and there's some potential for AMD to steal a good amount of market share from Intel in the craptop segment (sorry, I don't like budget notebooks). Kabini is simply in a very good position (contrasted with Temash, which is not).
The good: Wide instruction support. This should fare nicely against Celerons and Pentiums. I really hope this pressures Intel to not be so stingy on their instruction sets. Now that AVX is sure to have gathered some decent market adoption, it should be a pretty big advantage for AMD. One last bit: Kabini is a vast improvement over its successor, and it manages to get pretty close to Ivy Bridge i3s for a lot less.
The bad: Kabini's graphics still aren't quite there for notebooks. I sort of see HD 4000 as being the bare minimum for relevance in the PC space. I've been playing on HD 4000 for the past ~8 months, and while they work, they do that and nothing more. Kabini comes close to HD 4000 in many cases, and I'd imagine the A6-5200 wouldn't be that bad, but gaming on AMD's "Cat" processors won't be a pleasant experience in a notebook until their 20nm offerings come around. AMD will need to look out for stiff competition from Haswell here, though. On the bright side, this is one area that AMD is completely safe from Atom, until Airmont drops.
The ugly: Haswell Pentiums are about to drop, and eventually Celerons will work their way in. Battery life will be Intel's game, and AMD won't be able to do anything about it this is true for both their Atom line and their Core line.
Also, AMD's CPU advantages largely are due to Intel disabling so many instructions on their Pentium and Celeron lines. All it would take for Intel to shrug AMD off would be to re-enable those instructions. Obviously they can't do this ex post facto, but it's a precarious situation for AMD.
The really ugly: Still no turbo. What the hell, AMD.