Apple M6 GPU arrives on Geekbench with massive generational improvements

Just a few days ago, we reported on a Geekbench listing regarding the newly introduced Apple M6 SoC. The listing revealed, perhaps unsurprisingly, impressive multithreaded performance and simply unparalleled single-core performance.

With a score of 4,071 and 22,783 in the single- and multicore departments respectively, the M6 managed to match the massive M3 Max in multicore performance. In single-core, the performance uplift is far more massive, as one can expect from three generations of improvements. In fact, a different result even revealed a single-core score of a whopping 4,600 points. 

Now, it appears that we have concrete data revealing just how powerful the M6 SoC's 12-core GPU is. In Geekbench 7's Metal benchmark, the M6 iGPU seemingly manages to score somewhere in the ballpark of 92,000 points, which puts it roughly 31% ahead of its predecessor, the 10-core M5 iGPU.

Clearly, these are genuinely impressive generational improvements. Given that the Metal API is unique to Apple, we cannot draw any comparisons between Apple's latest iGPU and the discrete offerings from Nvidia or AMD.

That said, OpenCL tests do exist, and the Apple M6 seemingly manages to score around 58,000 points, barely keeping up with an 8-year old Nvidia GeForce RTX 2060 GPU Max-Q GPU. However, it should be noted that OpenCL was deprecated on macOS around 8 years ago, which makes these numbers much less reliable.