Apple A6X

[MrX8503]

It would also explain why Apple decided to move to a custom core this generation. They knew Cortex A15 wouldn't be a viable solution for their thinner/lighter philosophy, and so they had to create an alternative that would not be thermal throttled that easily.

That was my theory as well. It also makes sense as to why we didn't see an A15 smartphone this year.

 

[runawayprisoner]

That was my theory as well. It also makes sense as to why we didn't see an A15 smartphone this year.

Once 32nm becomes the norm and we start looking at 28nm process, then I think we'll start seeing A15 smartphones.

At the same time, Apple themselves would have moved on to PowerVR G6 series (fly like a G6!?) and also increased frequency for Swift even higher. 1.3GHz currently isn't so shabby at all. It really does make my iPhone 5 fly, and I can't even begin to imagine how much faster it can be.

In fact, I think the reason they didn't clock the CPU higher in the A6X is purely because the GPU needs a lot of headroom to breath. At 32nm, the size of the A6X die is nothing short of massive...

 

[dagamer34]

Also since the Surface hits 900's on Sunspider with a pretty weak CPU compared to what's out there by Samsung and Qualcomm.

The CPUs in the Tegra 3 (Cortex-A9 x 4 @ 1.3Ghz) and Exynos 4 Quad (Cortex-A9 x 4 @ 1.4Ghz) aren't that different. It's the GPU where the Exynos is stronger, but that bears no influence on a browser benchmark.

Besides, browser benchmarks are so software dependent that I feel it's useless to compare small differences in numbers that are hard to tease in actual web browser sessions where network performance makes a much larger difference.

 

[MrX8503]

Once 32nm becomes the norm and we start looking at 28nm process, then I think we'll start seeing A15 smartphones.

At the same time, Apple themselves would have moved on to PowerVR G6 series (fly like a G6!?) and also increased frequency for Swift even higher. 1.3GHz currently isn't so shabby at all. It really does make my iPhone 5 fly, and I can't even begin to imagine how much faster it can be.

In fact, I think the reason they didn't clock the CPU higher in the A6X is purely because the GPU needs a lot of headroom to breath. At 32nm, the size of the A6X die is nothing short of massive...

Yes, the GS4 is expected to get a die shrink, the Exynos 5400. I think we'll see Apple become more and more aggressive with their SoC as they gain experience designing their own chip.

Competition won't be in short supply.

 

[runawayprisoner]

Yes, the GS4 is expected to get a die shrink, the Exynos 5400. I think we'll see Apple become more and more aggressive with their SoC as they gain experience designing their own chip.

Competition won't be in short supply.

Not sure how much more aggressive they can be. Their die size is almost twice that of the closest competitor... and they are using GPUs that no one else does.

As an aside, reading comments in the hardware preview, I can see a lot of people are disappointed with the way Mali T-604 turned out. Adreno 320 seems like the closest contender now. But it looks like Adreno 320 gets thermal throttled pretty easily.

 

[Mopetar]

Not sure how much more aggressive they can be. Their die size is almost twice that of the closest competitor... and they are using GPUs that no one else does.

As long as the process is mature, the die size really doesn't matter. It certainly adds to their cost, but unless they're getting loads of defective parts, they probably don't care too much if they need to spend an extra few dollars per SoC since their margins are so high.

At any given time, most of the SoC isn't actually doing anything so it's not as though making a bigger die is going to make it drain the battery like crazy either. Given that adding some highly specialized hardware can actually save energy by allowing the CPU cores to stay off, building bigger might have some tangible power benefits.

The GPU power is getting a little egregious, and it will be interesting to see how much of a performance bump comes from switching over the 600 series GPUs when they become available. If anything, that will probably allow Apple to shrink their die back down while still maintaining similar performance levels. Either that or they'll just use the extra space to cram in even more stuff than before.

Since they bought up all of those 4G patents, I also wonder if they're planning on making their own cellular radio and baseband parts in the future and adding some of that in to the SoC like Qualcomm has done.

 

[Aikouka]

As long as the process is mature, the die size really doesn't matter. It certainly adds to their cost, but unless they're getting loads of defective parts, they probably don't care too much if they need to spend an extra few dollars per SoC since their margins are so high.

It seems that Apple has been rather keen on using what could be defective silicon by using pared down processing in other devices such as the AppleTV (it currently uses a single core A5).