Apple’s latest M3 Pro chip in the new 14-inch and 16-inch MacBook Pros has 25% less memory bandwidth than the M1 Pro and M2 Pro chips used in the equivalent models of the previous two generations.
Based on the latest 3-nanometer technology and featuring an all-new GPU architecture, the M3 series chips are said to represent the fastest and most power-efficient evolution of Apple silicon yet. For example, Apple says the 14-inch and 16-inch MacBook Pros are 40% faster than the 16-inch model with the M3 Pro chip.
However, according to Apple’s own hardware specifications, the M3 Pro system on a chip (SoC) has a memory bandwidth of 150GB/s, compared to 200GB/s in the previous M1 Pro and M2 Pro. As for the M3 Max, Apple says it’s capable of “up to 400GB/s,” since the scaled M3 Max with a 14-core CPU and 30-core GPU only has 300GB/s of memory bandwidth, whereas the equivalent 12-core CPU and 30- Measured with core GPU, the M2 Max has a bandwidth of 400GB/s, similar to its most powerful 12-core CPU, 38-core GPU iteration.
Notably, Apple has also changed the key ratios of the high-end M3 Pro chip compared to its direct predecessor. With a 12-core CPU the M3 Pro has 6 performance cores (versus 8 performance cores in the 12-core M2 Pro) and 6 performance cores (versus 4 performance cores in the 12-core M2 Pro), while the GPU has 18 cores (versus 19 in the equivalent M2 Pro chip). .
Additionally, while the M3 chip’s 16-core neural engine has the same number of cores as Apple featured in the 3nm-based A17 Pro chip that debuted in the iPhone 15 Pro series in September, it’s relatively weak on paper in terms of maximum performance. Achievable throughput, which is measured in trillions of operations per second (TOPS).
According to Apple, the M3 Neural Engine is capable of 18 taps, whereas the A17 Pro Neural Engine is capable of 35 taps. It’s hard to say for sure, but the iPhone 15 Pro may need a more powerful neural engine for features like computational photography and Face ID, while the M3 can use its extra GPU cores to compensate in other areas like machine learning.
Taken together, it’s unclear what real-world difference these changes make to the M3’s performance against Apple’s equivalent predecessor chips in various usage scenarios, especially since the latest processors have new dynamic caching memory allocation technology. Each task uses as much memory as it needs.
This opacity isn’t helped by Apple promoting the power of the new M3 Pro and M3 Max chips by comparing them to the M1 Pro and M1 Max over the latest M2 variants. Hopefully we’ll know more when the first full third-party benchmarks become available.
The new MacBook Pro models are available to order now and will begin arriving to customers and will be launched in stores on Tuesday, November 7. Check out our MacBook Pro announcement coverage for all the details.