Geek discretion advised: Apple M1 Max trails behind RTX 3080 Laptop by nearly 30% in GFXBench but is about 8% faster than RTX 3060 Laptop, mileage varies depending on considered benchmarks

[Redaktion]

GFXBench 5 results of the Apple M1 Max indicate that raw performance may really not match up to an RTX 3080 Laptop GPU, though there are likely benefits when power consumption is factored. The early results show that the M1 Max is probably more on the lines of an RTX 3060 Laptop GPU than an RTX 3080. However, it all depends on the type of benchmark considered and there's no definitive conclusion just yet.

https://www.notebookcheck.net/Geek-discretion-advised-Apple-M1-Max-trails-behind-RTX-3080-Laptop-by-nearly-30-in-GFXBench-but-is-about-8-faster-than-RTX-3060-Laptop-mileage-varies-depending-on-considered-benchmarks.574178.0.html

[[email protected]]

Who cares about benchmark ? They often don't relate to real world performance.

Also, even if it had 3080 deskptop performance, great... there are almost no games available on MacOS, so what's the point

[Jeff78]

Apple has been remarkable at how good they can generate FPS in GFXBenchmark.  But gaming performance and compatibility are still major issues to overcome.  I would expect the GPU to be a 3050 or 3060 class, or perhaps somewhat better, given the class of laptop it's being designed for.  It also has the advantage of TSMC 5nm process and a very lean API layer.  But the 3080 comparisons are very optimistic. We won't see that performance on real workloads.

[123]

using Paralles 17.10 no? You have your tons games, using RPCS3 others, not enought? Install Linux and RPCS4 all PS4 games.

Goodbye x86, in 2025 russian cpu based on ARM will come out in market, Qualcomm iiniPC market, Apple with their macmini and so on...

AMD is over with their trash apus with +10% per year

[Ariliquin]

Nothing changes the fact the Apple architecture is truely revolutionary for a laptop. These new chips will encourage both customers and developers to push for better gaming support on Mac. The days of Mac not being a gaming option are over.

Also with TB4 ports you can run an external RTX 3089 if you really need to.

[Kaulana]

The fact that a laptop will be able to maintain sustained 3060+ performance unplugged - for hours - is pretty mind boggling.  And this is just their 1.5 gen chip.  Apple M-series chips are another iPod/iPhone-level game changer.

[Tridents]

We need to wait for much better benchmarking.
Not only GFXBench doesn't inspire much confidence, but there are actually 2 versions of the M1 Max with different number of GPU cores enabled.

[LL]

Why people don't use Redshift benchmarks? .
it is a renderer with x86 and Metal versions

Blender when 3.1 version arrives probably in first quarter 2022 will also have metal version.

[Timmiers]

Who games with mac anyway$? Without proper gaming population, gaming studios will treat macOS as mistress OS. As long as the big companies like EA, Ubisoft, 2K will commit for ground up support then we could say it will truly revolutionize laptop space.

[Cat Dog]

So, is this the 24-core variant of M1 Max or the 32-core variant???
If it was the former then it would make a lot more sense.

[Ryan Weiss]

How is this correct? The 3060 fps numbers are way higher than the 3080?? 🤔

[r.j.a. schmitz]

Who uses a laptop for benchmarks anyway...

[Rico Mico]

Two aspects here:
1) most mac clients use macs for productivity, so on "everything" except games, these chips excel! I have a Mac Mini M1 (my first Mac ever) and it's mind blowing

2) being these chips even better on parallels and Crossover, even more x86 windows apps and games are possible;

3) if the PRO versions (the ones that will sell the most) are near an RTX 3050 / RTX 2060, 3D is already fabulous and enough for most people. Which will make game producers even consider Mac OS as an attractive platform

4) Apple should commit themselves with Vulkan too, that way it would be even more attractive for producers.

[danwat1234]

I wonder if M1Max GPU could be used for folding@home or other distributed computing projects that use CUDA? Some efficient translator could be developed?
M1 could be quite an efficient number cruncher for science and engineering.

*Ya or use openGL, Vulcan, etc