Analysis of Apple's A14 chip shows an increase in single-thread performance of nearly 3x in 5 years, suggesting M1 will be a formidable rival to x86 incumbents
What makes this analysis matter rather than merely interesting is that the M1 it predicts landed days later: [[a:960052|Geekbench uploads show the M1 outscoring every previous Mac in single core even while emulating x86]], and the Mac mini deep dive rates its fan-cooled performance as outstandingly good, besting Intel's chips and running alongside AMD's Zen 3.
First-order effects
Intel and AMD's laptop lines immediately face a competitor whose mobile-derived cores match their best single-thread throughput at a fraction of the power envelope, with Apple shipping it across the Mac lineup at launch pricing.
Apple's own transition risk collapses faster than expected: because the M1 leads every prior Mac even under x86 emulation, the Rosetta-era performance penalty buyers feared largely fails to materialize at the single-core level.
Second-order effects
Intel and AMD are forced to compete on performance-per-watt rather than peak clocks, since the Mac mini results show Apple winning efficiency while only matching — not exceeding — Zen 3 raw speed.
Benchmark cadence itself becomes competitive pressure: each Apple release (M2 scaling to 20B transistors from 16B in the M1) gives reviewers a fresh yardstick that x86 vendors must answer within their own product cycles.
Third-order effects
If the A14's five-year slope holds through the M-series roadmap, the PC market structurally splits between vertically integrated custom silicon and merchant x86 vendors, with laptop design wins migrating toward whoever controls both chip and OS.
The deeper precedent is that smartphone-scale R&D volumes now fund desktop-class cores — meaning the incumbent advantage in PC processors rests on ecosystem lock-in rather than architectural lead, a much weaker position to defend.
The trend: Apple Silicon marks the point where mobile-first custom ARM cores close the single-thread gap with x86 incumbents, turning processor leadership into a systems-integration contest rather than a foundry-clock race.
Full chip performance, Intel vs Apple. ➡ M1 should be up a bit higher ➡ Come on @nuvia_inc, pin the tail on the chart https://www.anandtech.com/... https://twitter.com/...
Apple's M1 chip has 16,000,000,000 transistors!! I still remember picking up transistors one by one during my Engineering lab long ago. How technology has shrunk the size of circuits.. https://twitter.com/...
The Anandtech breakdown of the Apple M1 shows an incredible CPU. The craziest thing is: M1 is likely the slowest ARM Mac chip there will ever be, and it's already faster than a 10900K. Imagine what they'll be able to do with 100 watts instead of 18 watts. https://www.anandtech.co…
I couldn't be more unexcited about M1. It's probably a good SoC for thin and light computers, but without some real-world benchmarks it's really hard to believe Apple's lofty (and vague) claims. And, of course, that's one of the final steps in ruining macOS completely for devs.
Is it more impressive that the M1 (the chip, not the tank) delivers 1/3 the tensor power of the XSS in an ultrabook or that the XSS delivers 3x the tensor power of Apple's best chip yet in a $300 box the size of a Wii U? I can't decide. Maybe it's both?
Apple's new M1-powered Macs are promising. It's a new era for Mac desktops and laptops. Still sitting this one out, though. I typically don't buy first-generation units. Besides, my newly upgraded, secondhand 2012 MacBook Pro is still more than enough for me.
Under appreciated: Apple's M1 chip makes the MacBook the first PC SoC with a dedicated AI accelerator. At 11 trillion op/s, the M1 has ~2x performance of the DLA in Nvidia's Xavier SoC. Intel could offer nothing like this. This is the power of vertical integration. https://twitte…
This is how disruption looks like: Apple CPUs performance vs Intel CPUs (and this doesn't include the new M1) https://www.anandtech.com/... https://twitter.com/...
Apple's Shifting Differentiation Apple is about the integration of hardware and software, but the balance between the two has shifted over time. https://stratechery.com/...
AMD will likely beat Apple M1 at raw, synthetic, integer performance, hence debunking the “Apple has the highest performance CPU”. Synthetic, integer PPW is a different story. Real apps performance and compatibility matter in the end. #AppleEvent https://twitter.com/...