Yes, I believe Cook recognizes this operational threat to Apple's success. Furthermore, I think it would be useful for companies like Apple to help grow domestic semiconductor manufacturing capacity. What I assume will happen is Apple grows more Asiatic semi-conductor partners - due to the labor cost advantages - but in locations more solidly protected by the US sphere of influence (South Korea) or capable of defending themselves (India).
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Geekbench score for the Intel i9-13900K is out and it’s not looking pretty for Apple …
- Thread starter Retskrad
- Start date
- Sort by reaction score
true that. and i like that apple doesnt lock down the M chips in the Macs so u can install whatever u want (Needs to be ported first to m series tho)Look at the charts. Every Apple Silicon processor made is within the top 30 in single threaded performance. ALL of them. And the HIGHEST single threaded performing i9 is still within just a few hundred points of the worst performing Apple Silicon chip. That is quite remarkable.
Then, as you go down the charts you see ALLLLLLLL the i3’s, i7’s and i5’s that pad Intel’s low end so they can charge dearly for the high end.It’s really like Apple just makes and puts i9’s in everything.
Well, the important thing is that the machinery TSMC uses is sourced from ASML, a Dutch company.
So in case China attacks Taiwan (doubt it. For now) this would of course be baaad, but not a total catastrophe and/or the end of chip manufacturing in the west. The crucial tech is Dutch, not Taiwanese
These are scores from real benchmarked hardware. So that's what people run, right? I suppose the units with scores ~1000-1200 are the lower TDP models and the units with ~1600 scores are the higher TDP units.
We don't really know much about the new MBA chassis. Still, the GB5 scores I have linked are of passively cooled M2 MacBook Air, TDP notwithstanding.
The point that I am trying to make is that the passively cooled M2 MBA is likely to be faster in single-core than a high-end actively cooled ADL-P ultraportable, while offering comparable multi-core performance. And much faster graphics of course. I would be curious to see in-depth comparisons between the M2 and the new 6800U, especially the new AMD GPU.
The way how some software uses the GPU (or at least at the moment), actually means Nvidia is even more efficient then the Apple GPU or more similar then some would expect.
On the CPU Apple is very efficient one the GPU not so much.
Still can’t wait to see what 3nm brings 🤞for the rest of the M2 line using that process.
Last edited:
I’d still say the Apple GPU is very efficient when it’s used like an Apple GPU. If it’s used like an Nvidia GPU (spending the time calculating a lot of vertices then shoving them over a fast bus to an off-chip GPU… instead of intelligently interleaving the work having the CPU and GPU accessing the same pool of memory (kind of like running down a zipper) then yes, any developer doing that will end up using Apple GPU’s very inefficiently.
It’s going to take awhile for developers used to working one way are able to effectively shift gears and be as efficient on Apple Silicon.
From what I remember Intel has the latest EUV machines *on order* from ASML, but touted that one of their new processors is *going* to their first on a full EUV process.
So from that, I don’t think they have been doing EUV as long as TSMC.
That said, just like a node isn’t the end all and be all in performance, neither is the etching technology…but you’re definitely at a disadvantage without the smallest possible etching.
Considering the GPU in the strict sense — the compute engine — Apple GPUs are crazy efficient, M1 series offers 1.3TFLOPS of peak general-purpose compute per 5 watts. That's not something other GPUs can reach. But of course, if you are looking at certain specialised applications — like ones involving ML acceleration or raytracing, M1 GPU loses efficiency since it doesn't have the dedicated hardware unlike Nvidia.
Should that happen, having the latest MacBook will be one of the things of least concern given that WW III will be commenced.
A better node would play a role, but obviously there’s a lot more than the node in a design. There are loads of design choices like using asynchronous logic, clock gating, transistor stacking, etc. Then there are special purpose in-house design tools leveraging machine learning, language theory (say, for proving correctness or short-cutting unnecessary logic), fancy simulators, or who knows what.
But in the context of specifically comparing Apple to AMD or NVIDIA… No idea 🙂
There have been some recent articles about new techniques for transistor stacking, though IIRC those were referring to something Intel was cooking up.
No idea. Better node definitely plays a role, but as @altaic wrote above, the node advantage alone is not enough to explain the efficiency lead. It has been speculated that Apple's GPU ALUs are more streamlined than the competition and possibly that Apple uses a larger hardware thread execution width (but I am not convinced about that last bit).
The Pro and Max are identical CPU-wise, the difference is purely GPU.
Also considering power level I’d rank it
A15 -> i3
M2 -> i5
M2 Pro/Max -> i7
M2 Ultra -> i9
Ultra is a workstation-level CPU. I'd put it in the same category as Threadrippers and Xeons. Looking strictly at the product segmentation, the M2 Pro should compete with the i7/i9 from Intel — and it does so quite successfully in the laptop space, on desktop — not so much.
In which applications has Apple's GPU proven to be more efficient than Nvidia's CUDA Cores?
And yet as a CEO, Tim Cook has to plan for it. And I'm sure that he has.
You could say the same thing about Ukraine but we're all still around.
Real applications you mean? Games and photo/video editing comes to mind. I am not aware of any popular software that uses GPU compute on a large scale — except renderers of course, where Nvidia commands a strong lead thanks to hardware raytracing and more mature implementation.
This gen of Xeons are not that great (they don't have a high core count option like AMD or Apple does).
Seems like healthy competition. I would love to experience Windows with an efficient and competent processor from Intel that levels up to Apples Silicon.
Like AMD Ryzen 7 6800U?
Apple M2 SOC Beats AMD Ryzen 7 6800U's RDNA 2 iGPU In Shadow of The Tomb Raider But Loses In CPU Benchmarks
Apple's M2 SOC has been tested against a vast variety of AMD & Intel CPUs, showcasing its gaming & application performance potential.
wccftech.com