Maynard Handley (name99.delete@this.name99.org) on November 20, 2020 11:54 am wrote:
> I don't understand your reasoning.
> You agree that Apple found it valuable to create X versions of their SoCs for the higher end?
> And that these SoCs were based on the base-line SoC, with only very specific
> modifications (like cache size, number of cores, number of GPU cores)?
> And that on at least one occasion when the timing made sense, they delayed
> the release of such an X core to catch the upcoming TSMC process?
>
> What in this chain of logic do you believe is flawed?
> Of course Apple COULD delay the iMac and larger MBPs till the A15. Is that likely?
> Of course Apple COULD have changed schedules so that the A15 is such a small
> design change that it ships six months after the A14. Is that likely?
>
> Of course Apple COULD release the M1X on N5 rather than N5P? Is that likely? -- That's really the alternative
> hypothesis that makes sense, and it's not implausible. But I think the win from catching N5P is worth
> delaying the release of the M1X machines by a month or two if that's the price of doing so.


Who is talking about delaying them "until the A15". The A15 is the iPhone chip, it isn't used in Macs. I said Apple could have planned several years ago to have the basic A15 design (specifically the cores and GPU) done earlier in the year than they typically do so it could be used in the "8+4" chip in high end laptops.

I wasn't suggesting the A15 is a small design change from the A14. I'm fitting this in your existing claim that the A14 is a small design change from the A13. Why would they do that, unless they had a bigger engineering challenge they needed to expend effort on.

Their chip designers were clearly not challenged by creating M1, which all evidence indicates is to A14 as A12X is to A12. They will also not be challenged by creating M1X, if all it is is 4 more big cores, 8 more GPU cores, and N5P. But if they "rushed" the A14 design (resulting in a "small design change" from A13) because they wanted to change their cadence and start the A15 design earlier in the cycle than they already did.

I'm not saying my theory is the only possible theory, but the idea that waiting for N5P is a big win because it buys them 5% - or 3.36 GHz instead of 3.2 GHz - is kind of silly. That's small potatoes. If your position is that Apple is going to try to wow people, they need more than a 160 MHz clock boost.