Apple M3 Pro & M3 Max analysis - Apple has significantly upgraded its Max CPU

[Redaktion]

Apple has made big changes to its new M3 Pro and M3 Max chips and for the first time, the the M3 Max not only offers considerably more GPU performance compared to the M3 Pro, but it also features a much faster CPU. This is in part due to its new core configuration. But what does the new SoC's efficiency look like?

https://www.notebookcheck.net/Apple-M3-Pro-M3-Max-analysis-Apple-has-significantly-upgraded-its-Max-CPU.782974.0.html

[IvanK]

Why do articles like this continue to use Cinebench R23? I know it is inconvenient (and maybe not practical) to rerun every system on Cinebench 2024, but Maxon themselves state that Cinebench 2024 is more appropriately optimized for Apple Silicon and therefore a better platform for comparison of Apple / Intel / AMD. In the multicore section you note AMD and Intel as the clear performance winners, as they are significantly faster in R23 (while being, though it is not noted in the text, a good bit slower in Geekbench). But if you look around online there are numbers out there for Cinebench 2024 for all of these systems, and they tend to show only about a 4% lead for the 13900hx and 7945hx over the M3 Max in multicore, and the M3 Max as the clear leader in single core. Which is a very different story, especially once the differences in efficiency are factored in.

[MQ.Yang]

chipsandcheese com/2023/10/22/cinebench-2024-reviewing-the-benchmark

I wouldn't recommend this benchmark, it makes poor use of the capabilities of the x86 architecture

[Julian M]

The M3 Max is indeed a beast, it's too bad however the M3 Pro didn't follow - Apple figured the Pro was too good in past iterations, so instead of making it a lite version of the Max, it has become a plus version of the base processor.

[IvanK]

chipsandcheese com/2023/10/22/cinebench-2024-reviewing-the-benchmark

I wouldn't recommend this benchmark, it makes poor use of the capabilities of the x86 architecture

Yes I've read their overview quite thoroughly. The analysis says nothing of that sort?

[Neenyah]

chipsandcheese com/2023/10/22/cinebench-2024-reviewing-the-benchmark

I wouldn't recommend this benchmark, it makes poor use of the capabilities of the x86 architecture

I will quote myself:

What even is the point of running Cinebench? One weaker CPU can get much better score than another much stronger CPU there and what does that mean? Only that one CPU will better run Maxon's apps ( https://www.maxon.net/en/ ).

Cinebench is not a general purpose benchmark yet people and reviewers keep using it for exactly that. The idea behind it is to evaluate how well will Maxon's apps and software run on your PC, literally nothing more than that and I can assure you that Maxon's team probably couldn't care less if their super-optimized professional-grade 3D rendering engine (Redshift now in Cinebench 2024) exercises all CPU elements fairly for general everyday computing tasks and usage or for gaming.

If one has no plans to ever touch Cinema 4D or ZBrush or anything else made by Maxon then there really is no reason to run Cinebench at all.
 

The performance jump is minimal therefore:
The 12gen, 13gen and 14gen are the same and all at 10nm++++ and consuming more energy and higher temperatures.
AMD performance increased Zen 2, Zen 3 = Zen 3+, Zen 4
Likewise increase in iGPU RNDA, RDNA 2, RDNA 3 all at 4nm
Honestly, if you are interested in buying a laptop, think about a Zen 4 7040 Phoenix

Stop trolling with this spam. i5 14600K is literally a desktop CPU.

[Yeshy]

Could you make a power scaling graph for gaming with both Witcher 3 and either a Rosetta or native game also available on Windows?

Since it's not easy to power limit on macOS, you can use Low Power mode or simply running the game with a frame cap with lower settings.