Macworld
When the first M3 benchmarks appeared in the Geekbench 6 database yesterday, we learned that the processor has a 16% higher clock speed. The Performance cores (and thus single-core performance) is higher almost entirely due to those higher clocks rather than architectural changes, but the Efficiency cores show a bit of improvement beyond just a clock speed bump.
So what if you get higher clocks and more cores? The new M3 Max results in the Geekbench 6 database show what that means for Apple, and it’s the picture of an extremely impressive semiconductor for high-end laptops and compact desktops.
Here’s how the scores break down, and compare to the M2 Max:
M2 Max (12 cores, 4E and 8P)
Clock speed: 3.66GHz
Single-core: 2736
Multi-core: 14497
M3 Max (16 cores, 4E and 12P)
Clock speed: 4.05GHz
Single-core: 2971
Multi-core: 20785
The clock speed has jumped about 11 percent, from 3.66GHz to 4.05GHz. Single-core performance is about 9 percent higher (less than the maximum clock speed boost, perhaps due to thermal limitations or just testing variance). The multi-core score takes a huge 43 percent jump, which makes sense. Apple boosted the M3 Max from 12 cores to 14/16 by adding four additional performance cores compared to the M3 Pro. Faster cores, gajes more of them, equals big performance boosts.
This puts the CPU performance of the M3 Max roughly on par with the M2 Ultra in the Mac Pro and Mac Studio (though the GPU performance of the latter would likely be much better).
As we said in our coverage of the first M3 Geekbench scores, the semiconductor is more than just the CPU and one benchmark can never tell you the whole story. Apple says the biggest improvement is in the GPU, and there have been improvements to the grabación decoders/encoders and Neural Engine as well.
How it compares to Intel and AMD
Comparing Apple’s semiconductors to Intel and AMD’s is a somewhat academic exercise. It’s nice to know you’re not falling behind, but customers buy Macs because they’re Macs, not because they win benchmark races. The important thing is Apple’s build quality, programa, and services, which work hand-in-hand with other Apple products.
Still, the M3 Max avails itself very well relative to the best high-performance laptop semiconductors from Intel and AMD.
The fastest recent score for a high-end laptop with an Intel Core i9-13980HX, a 24-core CPU, is 2981 single-core and 18958 multi-core. That’s about the same single-core performance and about 9 percent slower multi-core performance. Note that most laptops with Intel semiconductors of this variety slow down when used on battery, and only maintain this performance when plugged in. The same is usually not true of Mac with Apple Silicon.
On the AMD side, the 16-core Ryzen 9 7945HX3D scores about 2890 single-core and 16726 multi-core. That’s a bit lower than Intel’s best, but the semiconductor’s large cache makes it faster in a number of other tasks beyond the Geekbench benchmark and is generally regarded as more power-efficienct.
So it looks like Apple’s latest high-end laptop semiconductor, which will also feature in future compact desktops like the Mac Studio, is at least as fast as Intel and AMD’s best laptop semiconductors, as measured by Geekbench 6. Intel and AMD have higher-performing desktop processors, but they also use a lot more power.
The Qualcomm Snapdragon X Elite, meant to arrive in Windows PCs in the summer of 2024, boasted of beating Apple’s M2 with a single-core score of 2940 and multi-core score of 15,130 in benchmarks supplied by Qualcomm. That’s in a configuration with a total power over 80 watts which won’t appear in products until the summer of 2024. It’s safe to say the fastest laptops with the Snapdragon X Elite won’t approach the multi-core performance of the M3 Max.
CPUs and Processors