Apple’s M3 Max is a testament to how far the company has come with its custom silicon, and with the help of TSMC’s 3nm technology, the bar is raised even further, at least for this high-end part. We have been thoroughly impressed that the newest Apple Silicon is slightly faster than the M2 Ultra in Geekbench 6’s multi-core result, and now we have some more GPU comparisons for you. This time, the M3 Max’s 40-core graphics processor takes on NVIDIA’s high-end RTX 3080 and laptop RTX 4080 in GFXBench, and the negligible performance differences between all three will take you off-guard.
GFXBench results do not provide the entire story, as M3 Max can be significantly slower in other tests
The GFXBench Aztex Ruins benchmark was posted by SJV, showing that the M3 Max can achieve a total of 28,517 frames at an average of 443.5FPS at the 1440p resolution. In comparison, the desktop RTX 3080 is barely faster than Apple’s 3nm SoC, posting a total of 28,889 frames at an average of 449.3FPS. There is also another version of the RTX 3080 that achieves a slightly higher score, but the short version is that all of these differences are extremely minute. Even NVIDIA’s laptop RTX 4080 GPU, which is said to be faster than the desktop RTX 3080, achieves a total of 29,759 frames at 462.8FPS.
These results reveal that the M3 Max GPU is 7 percent slower than the laptop RTX 4080, but before readers get too excited about seeing the capabilities of an ARM-based chipset, there is a ton of ‘behind the scenes’ information that we need to relay to you first. Firstly, we believe that GFXBench provides some extremely skewed results, and we say this because previously, the M3’s 10-core was shown to be just 7.4 percent faster than the M2, featuring the same number of GPU cores.
It was later revealed that GFXBench has not been updated in over a year, so it is highly possible that certain optimizations that are meant to show the true capabilities of next-generation GPUs are not visible in these benchmarks. We have also seen GFXBench showing that the A16 Bionic is faster than the A17 Pro, so undoubtedly, there are potential issues that need to be addressed. Last but certainly not least, GFXBench is run through different APIs, depending on which GPU is running the benchmark.
The Apple M3 Max variant with integrated 40-core GPU is just 7% slower than Nvidia’s discrete RTX 4080 mobile GPU in the GFXBench Aztex Ruins benchmark pic.twitter.com/45GQT8AFfC
— SJV (@t3mporarybl1p) November 4, 2023
NVIDIA’s RTX 3080 and laptop RTX 4080 rely on DirectX, whereas the M3 Max focuses on Metal, Apple’s own graphics API, so the lack of differences displayed in GFXBench could also be due to the use of different APIs. We are pointing this out because we have seen in the past that even the M2 Max fails to impress in pure rasterization scenarios, as it struggles to render playable framerates when running older games such as Shadow of the Tomb Raider at higher resolutions.
At the end of the day, dedicated GPUs exist because of their unmatched capabilities, and while the M3 Max is no slouch in a myriad of tasks, at the end of the day, the chipset relies on an internal GPU, not an external one in the form of a dedicated die on the logic board. Still, if you feel enticed enough to make the fully unlocked M3 Max your daily driver, you have to pay a $500 premium when configuring it with either the 14-inch or 16-inch MacBook Pro.