Comment by IshKebab
2 days ago
I think that's still highly debatable. Intel and AMD claim the instruction set makes no difference... but of course they would. And if that's really the case where are the power efficient x86 chips?
Possibly the truth is that everyone is talking past each other. Certainly in the Moore's Law days "marginal impact" would have meant maybe less then 20%, because differences smaller than that pretty much didn't matter. And there's no way the ISA makes 20% difference.
But today I'd say "marginal impact" is less than 5% which is way more debatable.
> And if that's really the case where are the power efficient x86 chips?
Where are the power inefficient x86 chips? If you normalize for production process and put the chips under synthetic load, ARM and x86 usually end up in a similar ballpark of efficiency. ARM is typically less efficient for wide SIMD/vector workloads, but more efficient at idle.
AMD and Intel aren't smartphone manufacturers. Their cash cows aren't in manufacturing mobile chipsets, and neither of them have sweetheart deals on ARM IP with Softbank like Apple does. For the markets they address, it's not unlikely that ARM would be both unprofitable and more power-hungry.
Jim Keller goes into some detail about what difference the ISA makes in general in this clip https://youtu.be/yTMRGERZrQE?si=u-dEXwxp0MWPQumy
Spoiler, it's not much because most of the actual execution time is spent in a handful of basic OPs.
Branch prediction is where the magic happens today.
Jim Keller has to say that.
1 reply →
>Spoiler, it's not much because most of the actual execution time is spent in a handful of basic OPs.
Yet, on a CISC ISA, you still have to support everything else, which is essentially cruft.
1 reply →
Intel spent years trying to get manufacturers to use their x86 chips in phones, but manufacturers turned them down, because the power efficiency was never good enough.
Well, they were targeting Android, and the apps were emulating ARM on x86, and they were going against a strong incumbent. Accounts on the web of this failure seem to bring up other failings as the main problems.
3 replies →
You're basically reiterating exactly what I just said. Intel had no interest in licensing ARM's IP, they'd have made more money selling their fab space for Cortex designs at that point.
Yes, it cost Intel their smartphone contracts, but those weren't high-margin sales in the first place. Conversely, ARM's capricious licensing meant that we wouldn't see truly high-performance ARM cores until M1 and Neoverse hit the market.
2 replies →