From what I understood in this article,
In this article he is claiming that Apple got over this by limiting the resolution to 1024x768 even though the app "requests" retina resolution. If that is actually what is happening in ipad3, then his analysis does make sense.
Interesting. The brand new google nexus 10 tablet running a 32nm Soc with the latest mali graphics processor has barely 20% more performance @ native resolution than the 10 month old 45nm Soc-ed Apple ipad3 @ native resolution, under Glbenchmark 2.5.
Must be unbalanced too....no ?
"The Apple iPad 3 is an unbalanced design with an underpowered graphics processor for its high-end display"
Such a ridiculous statement. These analysts will cook up any silly comments for a few more clicks. Have he done any technical analysis before commenting that the graphics was underpowered? There could be many other reasons for a quick update other than this. And it was the CPU performance than the GPU performance that improved drastically between A5x and A6x.
I have used an iPad 3 for around 6 months before replacing it with a mini. I have extensively used it for 3D gaming and I can confirm, that the 3D graphics performance in games were very smooth.
Good point. I think Apple feeds the analysts a fair amount of Kool-Aid so it's hard for them to be soberly critical.
Kanter is not one of the usual Apple hangers on because he has more of a focus on semiconductors, something Apple rarely says anything about in public except to say it has a new chip the name is AXX and it is better than the last chip.
Nice to see that all of these analysts are coming out NOW and admitting iPad 3 was underpowered, when it was something obvious from the beginning, but I didn't see any of them saying anything then. Were they too afraid to ruin Apple's momentum? Why did they just mention it now when Apple already "fixed" the problem? And by fixed, I mean they've barely made it have the same performance as an iPad 2, at its native resolution.
Blog Doing Math in FPGAs Tom Burke 23 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...