Originally Posted by majjo
No, I'm saying "up to 7x faster" without specifics is meaningless.
Benchmarks can be used to lead he gullible around by the nose. Much like Apple did in the G5 days. They posted all sorts of Alt-Vec benchmarks while ignoring the terrible integer performance of the G5.
Short answer is no. Irregardless of what you display, at the end of the day, you still have to calculate color values for all of the additional pixels.
Most of the time yes you do. It will be interesting to see how Apple approaches this. However if they can get to a sub 32nm node they might be able to actually increase performance while lowering power.
In the end though all of those pixels will require a new architecture. I would be surprised to find out that they will go with one iPod the video memory technologies used on the latest video cards for main memory. Bandwidth will be a problem. They could also address rather bandwidth issue by putting a frame buffer right on the SoC. A simple run of the mill SoC implementation won't cut it.
Oh, I'm not saying it doesn't blow the iPad1 out of the water. I just meant that the term "up to 7x faster" is not enough information to draw any useful conclusions from.
CPU wise A5 isn't that much faster. Having two Cores does make a difference as does Apples attention to developer tools to support access to that computational power. In the end though Apple focused on the GPU which is frankly refreshing for Apple.