The Azec Ruins (GFXBench) benchmark, which is one of the popular graphics performance tests among ARM SoC users, showed a result of 52.2 FPS for the Apple M3, which, compared to the predecessor M2 with 48.6 FPS, means an intergenerational shift in performance of only 7.4%. The fact that the performance measured on the M3 was the first to reach the world can still be considered a success, because the M3 Pro, which has fewer graphics blocks than its predecessor, could have done even worse.
Many media presented the integrated graphics architecture as the most significant intergenerational difference between the M3 generation and the M2, but at least this result does not indicate a greater difference than the slightly higher clocks of TSMC’s 3nm process compared to the original 5nm.
The published architectural differences talk about new memory management, hardware support for RayTracing and Mesh Shaders. However, it seems that applications that do not use these two technologies (which are the majority in the Apple world) will not benefit significantly from the new architecture.
The Azec Ruins benchmark is one of those. And so passionate debates started in the discussions about this report, whether the low difference – rather than Apple’s hardware – is to blame for an inappropriately designed benchmark, which has been lacking an update for a long time. In the end, it seems that a very popular test until recently, which, despite its simplicity, does not reflect the reality of current games, will quite possibly start to fall out of favor among ARM SoC fans after many years.