InvestorsHub Logo
Followers 29
Posts 25865
Boards Moderated 0
Alias Born 09/11/2002

Re: None

Wednesday, 10/01/2014 3:57:48 AM

Wednesday, October 01, 2014 3:57:48 AM

Post# of 151791
iPhone6 review

http://www.anandtech.com/print/8554/the-iphone-6-review

There's a ton of additional data here to reevaluate the density claims on Apple's new A8. At an SoC level, it's a smaller die with a lot more transistors, but without precise transistor data or information on the types of transistors, it has lead to imprecise conclusions. So here's some new data on the CPU and GPU cores that I think can give us data on performance per die area between A7 and A8 designs.

CPU:
17.1 mm2 vs. 12.2 mm2
29% reduction, for about 10-20% better performance

While there are outliers, on average perf*density is about 1.5x.

GPU:
The GX6450 has the same number of GFLOPS as the A7's GX6430, but offers anywhere between 15% performance, up to 50%, averaging about 35%, according to Anandtech.

Meanwhile, die area shrunk from 22.1 mm2, to 19.1 mm2, a 14% reduction. Thus, perf*density gets a similar 1.5x improvement.

Clearly, A8 is a nice SoC, and a general improvement over its predecessor. However, I think earlier claims of 2x density and 1.5x better performance (implying 3x improved perf*density) were grossly overstated, and the A8 only gets half that.

On a usual process transition, I would actually call it underwhelming - but that's consistent with my view of TSMC's 20nm process. Their first real 20nm process is what they call 16FF, and we may see that next year. But at this point, 20nm is looking more like a half-node.
Volume:
Day Range:
Bid:
Ask:
Last Trade Time:
Total Trades:
  • 1D
  • 1M
  • 3M
  • 6M
  • 1Y
  • 5Y
Recent INTC News