What we know: On the CPU side of things, the A5x has two Arm Cortex A9 cores, and the Tegra 3 has 4 Arm Cortex A9 cores. Both can be clocked up and down depending on the thermal sinking available, so there's no real way to compare clock speeds. On the GPU side, the A5 (non x, in the ipad2 and iphone 4S) contains a PowerVR SGX543MP2. This is a more grunty chip than the Kal-El Geforce used in the Tegra 3. Even if the A5x makes no changes to the GPU, it will still crunch more numbers than the Tegra 3. What makes things fuzzier is the resolution jump: previously, the Tegra 3 had to push 1024000 pixels to the A5 786432 (i.e. 30% more). The A5x has to handle 4x the pixels than the A5 (3145728), but to make things fair the Tegra 3 that will almost certainly be in the Transformer Infinity will also be handling just over double the pixels it was (2304000), meaning the tables have turned and the pixel-pushing workload for the A5x will be 30% more than the Tegra 3. Now, graphics performance may not scale directly with the number of pixels rendered (vertex shader limited? Who knows, I'm not that deep into the nitty-gritty of GPU pipelines), but the A5x will by default contain a more powerful GPU than the Tegra 3. It may even be more powerful at the same resolution, but when the increase in screen real-estate is taken into account, the actual performance difference in comparative FPS may not be so great, certainly not 4x. Remember also, the the GPU has little to do with non-gaming tasks as long as it's set up to handle high resolutions. Even current chips should be able to handle anything 2D at 2048x1536 (or 1920x1200) without breaking a sweat. Unless you intend to play a lot of graphically intensive games, the GPU powering your fondleslab of choice is a non-issue.