Apple A8 and A8X
- Thread starter ltcommander.data
- Start date
I'm not getting my hopes up for progress any further beyond a 500 MHz GX6650 with a 1.5 GHz version of the A8's CPU/Cyclone.
4 GBs of RAM would indeed indicate that they're moving to distinguish the functionality of the iPad from even the large screened phones with more of a focus on productivity, multiple on-screen apps, etc. Though beta software has given at least some indication they've been experimenting along those lines, a complete, polished OS build for that purpose would still come as a big surprise.
4 GBs of RAM would indeed indicate that they're moving to distinguish the functionality of the iPad from even the large screened phones with more of a focus on productivity, multiple on-screen apps, etc. Though beta software has given at least some indication they've been experimenting along those lines, a complete, polished OS build for that purpose would still come as a big surprise.
Last edited by a moderator:
iMacmatician
Regular
If Apple is going to release split-screen multitasking for the iPad, then I could see the iPad mini having 2 GB (1 x 16 Gb) RAM and the iPad Air having 4 GB (2 x 16 Gb) RAM, if they decide that 2 GB is the minimum RAM required for the split-screen feature.
I haven't looked closely at the board, but assuming there is 4Gb spread over 2 chips, who is to say that this is not a logic board for a 12.x iPad Pro with full support for split screen apps ? Most are saying that a bigger ipad won't be seen until next year, but that doesn't mean a production logic board wouldn't be around at this time.
And perhaps lose one of the chips for a regular ipad Air ?
And perhaps lose one of the chips for a regular ipad Air ?
looks like apple "accidently" leaked details of mini3 and air2.
http://www.dailytech.com/Oops+Apple+Outs+iPad+mini+3+iPad+Air+2+via+iTunes+Store/article36725.htm
http://www.dailytech.com/Oops+Apple+Outs+iPad+mini+3+iPad+Air+2+via+iTunes+Store/article36725.htm
So does anyone have any performance expectations for A8X? I would expect a 6 cluster GX6650 to have similar performance to mobile Kepler (albeit with lower render precision in GFXBench) given that FP16 and FP32 ALU's increase by 50% compared to GX6450 in A8.
Last edited by a moderator:
It told us that they didn't include an ipad pro on that page.
Could you stop harping on the precision nonsense? Neither IMG, QCOM or Vivante GPUs deliver in those tests less than needed precision.
Just because NV choses to apply high precision in those quality tests (if it would be about performance and/or power consumption rest assured they wouldn't...) it doesn't mean anything.
But if you need something to harp on try sustained perf/W; it might help.
It is what it is. Even the HP (high precision) render quality is significantly lower in comparison for some reason.
And if I ask you define what it actually is I'll most likely get a bunch of copied and pasted whatever from the net because you don't even understand yourself what exactly you are pondering on. PSNR values are roughly the same for the simple and high quality test in Gfxbench for Adreno, Vivante and PowerVR GPUs which represent the wides majority of the market; a wee bit difficult that they are taking any "shortcuts" of any sort.
There's no secret sauce hidden anywhere they've even old dedicated blogposts with detailed explanations for the topic:
http://blog.imgtec.com/powervr/powervr-gpu-the-mobile-architecture-for-compute
http://blog.imgtec.com/powervr/mobile-gpu-compute-must-be-practical-gpu-compute
In sickness and in health you could still spare us with that kind of useless nonsense in every other sentence; yes GK20A PSNR scores are higher but again ONLY because it's a pure quality test and not a quality/performance test where they could/would result to half float values just out of pure "coincidence".