Apple’s spooky October 30th announcement revealed the first M3 generation Macs, including the MacBook Pro and iMac. But how fast can we expect them to be? Here’s an analysis of what you should and should not believe from their performance charts.
Apple also made a big deal about Dynamic Caching, Mesh shading, and hardware-accelerated raytracing for gaming, but how does that actually translate to performance?
Let’s take a closer look as we review each of the 3 chips’ configurations.
The third generation Apple Silicon chip launches in base M3, M3 Pro, and M3 Max flavors simultaneously, all built with new 3nm technology. The base M3 chip is quite similar in core counts to the M2 chip, with 8 CPU cores and 10 GPU cores, and supporting 8, 16, or 24GB of unified memory. However, be careful as the cheapest new M3 iMac that costs $1299 actually only has an 8-core GPU.
Apple says the M3 CPU performance cores are 15% faster than M2, while the efficiency cores are 30% faster. For the M3 chip that has 4 of each, that means up to 20% performance increase over the M2 chip. GPU performance is also supposedly up to 20% faster as well. Since the core counts are the same between M2 and M3, this seems like a standard generational improvement.
The M3 Pro’s core count goes up to 12 CPU cores and 18 GPU cores, but again, the base version found in the starting $2000 14″ MacBook Pro has only 11 CPU cores and 14 GPU cores. That’s compared with a 10/16 config on the base M2 Pro, or a 12/19 config with the upgraded M2 Pro chip. Notably, the M3 Pro has fewer GPU cores than the M2 Pro, and on the CPU side, the M3 Pro actually has 6 performance and 6 efficiency cores, which is worse than the M2 Pro’s 8 performance and 4 efficiency cores.
The M3 Pro supports either 18 or 36GB of unified memory compared to the M2 Pro’s 16 or 32GB. However, the M2 Pro had 200GB/s of memory bandwidth, while the M3 Pro only has 150GB/s, another spec reduction. That means for the M3 Pro to outperform the M2 Pro, it will need to rely on the per-core performance improvements.
However, it is quite telling that Apple purposefully did NOT INCLUDE any CPU performance for the M3 Pro compared to M2 Pro, and only said it is up to 20% faster compared to M1 Pro. The GPU performance is also only up to 10% faster. The hidden meaning here is that the M3 Pro is NOT a significant upgrade to the M2 Pro.
Finally, the M3 Max can go up to 16 CPU cores and 40 GPU cores, but that’s a $300 upgrade over the base version with 14 CPU and 30 GPU cores. Unlike the M3 Pro, the M3 Max is actually a substantial spec upgrade to the M2 Max, which only had 12 CPU cores and either 30 or 38 GPU cores. Unlike the M3 Pro that went in the opposite direction, all of the extra CPU cores on the M3 Max are performance cores, keeping the efficiency cores at 4.
The M3 Max can now be configured with 36, 48, 64, 96, or 128 GB of unified memory, although with a weird mixture of allowed combinations depending on the base or upgraded chip. This is a higher lower and upper limit from the M2 Max’s 32, 64, or 96GB configurations.
Unlike the M3 Pro’s paltry performance, the M3 Max CPU is said to be up to 50% faster than the M2 Max, which makes sense since the M3 Max can have 16 CPU cores compared to just 12 on the M2 Max. The GPU core count is just slightly more at the highest config than the M2 Max, so being 20% faster is again the standard generational increase, likely due to the new tech like dynamic caching and mesh shading that Apple made a big deal about.
Finally, across all M3 chips, neural engine performance is said to be 15% faster than M2’s, which is not as large as the 40% leap from M1 to M2.
So in summary, the M3 launch sees the base M3 chip as a standard generational improvement, the M3 Pro as potentially a dud in the water with no or very little difference from the M2 Pro, and the M3 Max being the only potentially worthy upgrade from the M2 Max, especially if you are looking for more CPU performance in the upgraded chip.