Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Apple is falling behind, M4 Max nowhere close to RTX 5090.

Z

Zest28

macrumors 68030
Original poster
Jul 11, 2022
2,688
4,255
When the M1 Max launched, Apple said that it rivaled the flagship NVIDIA GPU at the time, the RTX 3080.

However, in 2025, when comparing the M4 Max MacBook Pro to PC laptops equipped with the flagship NVIDIA GPU, the RTX 5090, which costs the same as a M4 Max MacBook Pro, the MacBook Pro gets destroyed, it is not even close.

And this is true even on battery power.

1743682456662.png


1743682479610.png
 
Last edited:
  • Haha
  • Like
  • Angry
Reactions: rafark, spyguy10709, maxoakland and 16 others
Zest28 said:
When the M1 Max launched, Apple said that it rivaled the flagship NVIDIA GPU at the time, the RTX 3080.

However, in 2025, when comparing the M4 Max MacBook Pro to PC laptops equipped with the flagship NVIDIA GPU, the RTX 5090, which costs the same as a M4 Max MacBook Pro, the MacBook Pro gets destroyed, it is not even close.

And this is true even on battery power.

View attachment 2498773

View attachment 2498774
Click to expand...
Bad review. Why use Opencl when metal provides more legitimate scores, as would cuda.
 
  • Like
  • Haha
Reactions: Project Alice, ikir, souko and 14 others
OptimusGrime said:
Bad review. Why use Opencl when metal provides more legitimate scores, as would cuda.
Click to expand...
Sadly no more CUDA in GB 6. :(

Zest28 said:
When the M1 Max launched, Apple said that it rivaled the flagship NVIDIA GPU at the time, the RTX 3080.

However, in 2025, when comparing the M4 Max MacBook Pro to PC laptops equipped with the flagship NVIDIA GPU, the RTX 5090, which costs the same as a M4 Max MacBook Pro, the MacBook Pro gets destroyed, it is not even close.

And this is true even on battery power.

View attachment 2498773

View attachment 2498774
Click to expand...
Yeah I watched this review and there's a lot to unpack beyond his consistent error throughout the video that the M4 Max has 20 cores. Let's start with price. That the overall device costs the same shouldn't entail an expectation that a particular component of the device, in this case the GPU, should have the same performance. Each device has its strengths and weaknesses and focus - i.e. if we were to assume otherwise (that two devices that cost the same should have the same performance in every aspect), one could ask why the PC's CPU is beaten so handily by the Mac's? Why doesn't the PC have 64GB of VRAM? Why does the PC only have a 1600p display? or flip that around: why doesn't the Mac have a 240Hz display? etc ... The raw GPU performance is the area where the PC excels (while one stays below its 24GB VRAM capacity) as is befitting a 5.8lbs gaming laptop.

In terms of generation upon generation raw performance, there was no 3090 mobile variant back in the day though on desktop the 3090 did exist. Meanwhile, the 3080 mobile still handily beat the Max at its max power and the Apple graphs were almost certainly on Wild Life Extreme or Aztec Ruins, benchmarks which Apple does exceedingly well in. As you can see from the first chart, it's the 3070 mobile (fine print is hard to read, but pretty sure the compact laptop they name had a 3070 mobile) that best resembled the M1 Max's performance and indeed Apple's M4 Max performance tends to be around the 4070 mobile - 4080 mobile depending on the benchmark (and beyond purpose-built benchmarks actual game performance can vary wildly and generally not in the Mac's favor) and on the configuration of the mobile Nvidia card (many are factory overclocked but also stuffed into enclosures that can't properly cool them). 5000-series mobile chips are only just coming out so we'll see how Apple's current compare to the full lineup of Nvidia's upcoming but yes one expects the M4 Max to drop to maybe the 5060-5070 mobile range depending though the 8GB of VRAM will be a limiting factor for the Nvidia GPUs, again depending on the circumstance.

Short version of all the above: @galad is correct :)
 
Last edited:
  • Like
Reactions: meneush, clam zero, OldMike and 18 others
Why would those results be surprising to anyone? Mobile 5090 alone has a TGP of 175W. MBP 16” M4 Max 40c has a total max load of about 140W, for the whole laptop. The GPU alone draws about 50W during max load. Also OpenCL was deprecated in 2018 after Mojave. macOS supports only up to version 1.2. The latest Windows version is 3.0.17. So it’s totally misleading but you can’t expect more from such sloppy youtubers with different errors in their tests as always.
 
  • Like
  • Haha
Reactions: ikir, histeachn81, wanha and 11 others
OptimusGrime said:
Bad review. Why use Opencl when metal provides more legitimate scores, as would cuda.
Click to expand...

because Nvidia gpu's don't have metal support

there years and older AMD gpu's do though, and they still beat most apple silicon
 
I am sorry but considering Apple only released their first Mac-based Silicon less than 5 years ago, they are doing pretty well. Comparing them to Nvidia who has been in the game for decades is not even a good comparison.

They don't even target the same markets or consumers.
 
  • Like
Reactions: ikir, Ericdjensen and IJustWannaTalk
I think this will be addressed within the next 2 years. The M4 Max is the first Apple Silicon chip that is graphically ‘fast enough’ for a lot of 3d workflows in my use case. I’d like more performance but it’s quite good.

Don’t engage with terrible tests from clickbait Youtubers. You’re never going to fix stupid.
 
  • Like
Reactions: ikir, Levithescienceguy, KeithBN and 3 others
TokyoKiller said:
Comparing them to Nvidia who has been in the game for decades is not even a good comparison.
Click to expand...

Eh, this doesn’t really work when Apple brought the comparisons themselves. Also they’re both GPU designers, of course they’ll be compared just as Apple’s CPUs are compared to Intel/AMD’s.

They don't even target the same markets or consumers.
Click to expand...


Again, eh. I doubt Apple is perfectly fine with where the Mac Pro is right now. When they can compete at the level of Nvidia’s highest end products they’ll shout about it.
 
  • Like
Reactions: -DMN-
Howard2k said:
Wasn't the comparison to the RTX 3080 "at equivalent power draw"?
Click to expand...
Yes. If only people would search a bit before making entire threads about some idiot youtuber who didn’t have their facts straight. The chart shows a comparison with a 3090.

Then again, understanding what “relative performance” means in this case requires some technical literacy.
Apple_M1_Ultra_gpu_performance_01.jpg
 
  • Like
  • Love
Reactions: ikir, blob.DK, wanha and 9 others
nathansz said:
because Nvidia gpu's don't have metal support
Click to expand...
That’s really irrelevant to my point, which was: the most optimized api should be used to determine the top performance of the gpu. Especially given the creator of Geekbench states scores with different apis can be compared. OpenCL is deprecated on macOS.
nathansz said:
there years and older AMD gpu's do though, and they still beat most apple silicon
Click to expand...
The 6900XT and 6950XT beat most, but most of Apple’s gpus are laptop gpus. Even so, the M3 Ultra beats all of those previous AMD gpus on Metal, and strangely it beats the newer ones using OpenCL or Vulkan.

Mac15,14 - Geekbench

Benchmark results for a Mac15,14 with an Apple M3 Ultra processor.
browser.geekbench.com

Even the M2 Ultra is close:

Mac Pro (2023) - Geekbench

Benchmark results for a Mac Pro (2023) with an Apple M2 Ultra processor.
browser.geekbench.com

Careful with the Geekbench charts. Some of the charts don’t distinguish between different core counts (ie 80 vs 60 core Ultra)

Worth remembering that this is just Geekbench Compute. It’s an ok test but not the only one. In Blender for example, the M4 Max (and maybe the M3 Max) easily beats all of AMD’s gpus. Nevermind the Ultras, which are significantly better.
 
  • Like
Reactions: ikir, Homy, Basic75 and 1 other person
OptimusGrime said:
That’s really irrelevant to my point, which was: the most optimized api should be used to determine the top performance of the gpu. Especially given the creator of Geekbench states scores with different apis can be compared. OpenCL is deprecated on macOS.
Click to expand...

that's fair, so what would be a good way to go about comparing apple silicon to Nvidia gpu?

I suppose it depends on what you want to do with your computer, my m4 pro is not as performant playing games as my 1400k/rx6800 hackintosh was, but I was willing to make that trade off for the many other advantages (though I am bummed that cities: skylines 2 stopped working in whiskey even if it was choppier)

also a well done price/performance comparison would be interesting to see

high end discrete gpu's are expensive for sure, but so is adding an 80 core gpu to a Mac Studio...
 
Last edited:
OptimusGrime said:
Especially given the creator of Geekbench states scores with different apis can be compared. OpenCL is deprecated on macOS.
Click to expand...

fair enough, that's why I linked specifically to the Metal benchmarks page, which is dominated by AMD

like I said above, I don't know how to go about comparing Apple to Nvidia
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back