> The point everyone conveniently seems to miss is a new SoC for mobile devices running at 10-20 watts shouldn't be in the same universe when it comes to performance with any Ryzen desktop processor—but here we are.
10-20 watts is how much desktop processors use for a single thread, so they're not claiming an unfair advantage there. It would be truly impressive if the M1 at that power level was competitive with desktop processors on multi-threaded workloads where they actually use their full TDP, but it isn't even close.
> The 16-inch MacBook Pro ($2,799.00) comes with a 2.3GHz 8‑core 9th‑generation Intel Core i9 processor, Turbo Boost up to 4.8GHz and for an additional $200, you can get it with a 2.4GHz 8‑core 9th‑generation Intel Core i9 processor, Turbo Boost up to 5.0GHz.
Which are old slower Intel CPUs. Intel's latest is 10th-generation and even those are slower than Ryzen, much less the older 9th-generation ones. And even those older Intel mobile CPUs defeat the M1 on multi-threaded workloads like Cinebench.
Core i9-9880H scores 8311 compared to 7833 for the M1 (native):
Which are old slower Intel CPUs. Intel's latest is 10th-generation and even those are slower than Ryzen, much less the older 9th-generation ones. And even those older Intel mobile CPUs defeat the M1 on multi-threaded workloads like Cinebench.
You seem to be ignoring the benchmarks where the M1 beats the Ryzen 9 5950X on several--but not all--benchmarks. [1]
From the article:
In SPECint2006, we’re now seeing the M1 close the gap to AMD’s Zen3, beating it in several workloads now, which increasing the gap to Intel’s new Tiger Lake design as well as their top-performing desktop CPU, which the M1 now beats in the majority of workloads.
And there's this:
The situation doesn’t change too much with the newer SPECint2017 suite. Apple’s Firestorm core here remains extremely impressive, at worst matching up Intel’s new Tiger Lake CPU in single-threaded performance, and at best, keeping up and sometimes beating AMD’s new Zen3 CPU in the new Ryzen 5000 chips.
And regarding power consumption:
While AMD’s Zen3 still holds the leads in several workloads, we need to remind ourselves that this comes at a great cost in power consumption in the +49W range while the Apple M1 here is using 7-8W total device active power.
That's the point I previously made—Apple’s first go at making their own processors for their entry level laptops and the Mac mini is super competitive with AMD's newest and best processors and beats it in several benchmarks, at a fraction of the cost and power consumption.
That's pretty impressive.
It has to make one wonder what the results will be when Apple ships a processor made for higher-end desktops, where they won't have the thermal and power restrictions of a laptop processor?
I'm sure when they were contemplating a transition away from Intel, they had to do their due diligence and consider if AMD could be an option. But clearly they knew they have a much higher ceiling with their architecture.
You know there are prototypes in Apple's lab now that far surpass Ryzen that'll be in machines spring/summer 2021. Apple's been relentless with shipping faster A series chips every year for the iPhone; I suspect we'll see the same thing with the M series.
> You seem to be ignoring the benchmarks where the M1 beats the Ryzen 9 5950X on several--but not all--benchmarks.
These are all single-threaded benchmarks. They don't even list the 5950X for the multi-threaded benchmarks because it's not even a comparison -- CineBench R23 MT for the M1 is 7833, the 5950X is 28641.
> And regarding power consumption: While AMD’s Zen3 still holds the leads in several workloads, we need to remind ourselves that this comes at a great cost in power consumption in the +49W range while the Apple M1 here is using 7-8W total device active power.
Which is, again, not surprising, because of the way CPU power curves work. The mobile Ryzen U-series processors have a similar power consumption to the M1. The desktop versions of the same processors are only slightly faster for a single thread but use a lot more power, because that last few hundred MHz is extremely inefficient and has only a marginal performance gain, so you only do that on desktop where power consumption isn't a priority.
A mobile processor being more power efficient than a desktop processor isn't a novelty. A mobile processor having about the same single thread performance as a desktop processor isn't a novelty. These things are par for the course. If they weren't true you aren't even in the game. Desktop processors primarily use their TDP to hold higher clocks even when running many threads.
> at a fraction of the cost
But you're again comparing a 16-core desktop processor to a 4+4 core mobile processor when the desktop processor would embarrass it on anything threaded, which is why it costs more. The lower core count processors have nearly the same single thread performance and cost less than half as much, and the "budget" Zen 3 processors, which should again have similar single thread performance, weren't the first to be released.
> It has to make one wonder what the results will be when Apple ships a processor made for higher-end desktops, where they won't have the thermal and power restrictions of a laptop processor?
It's not hard to infer what would happen. It would have marginally better single thread performance and much better multi-thread performance. Which is what the competing desktop processors already have.
> You know there are prototypes in Apple's lab now that far surpass Ryzen that'll be in machines spring/summer 2021.
But then it'll be competing with Zen 4 on 5nm and Socket AM5 with DDR5 memory. And 5nm and faster memory are two of the obvious reasons that the M1 is competitive as it is.
10-20 watts is how much desktop processors use for a single thread, so they're not claiming an unfair advantage there. It would be truly impressive if the M1 at that power level was competitive with desktop processors on multi-threaded workloads where they actually use their full TDP, but it isn't even close.
> The 16-inch MacBook Pro ($2,799.00) comes with a 2.3GHz 8‑core 9th‑generation Intel Core i9 processor, Turbo Boost up to 4.8GHz and for an additional $200, you can get it with a 2.4GHz 8‑core 9th‑generation Intel Core i9 processor, Turbo Boost up to 5.0GHz.
Which are old slower Intel CPUs. Intel's latest is 10th-generation and even those are slower than Ryzen, much less the older 9th-generation ones. And even those older Intel mobile CPUs defeat the M1 on multi-threaded workloads like Cinebench.
Core i9-9880H scores 8311 compared to 7833 for the M1 (native):
https://www.cpu-monkey.com/en/cpu-intel_core_i9_9880h-958
https://www.anandtech.com/show/16252/mac-mini-apple-m1-teste...