Apple M2 Extremely SoC Is not Sooner Than AMD & Intel Final Yr Desktop CPUs, 50% Slower Than NVIDIA RTX 4080
Apple’s M2 Extremely SoC is being benchmarked proper and left and we are able to evaluate these numbers to AMD, Intel, and NVIDIA chips to see if the brand new SoC actually delivers the claimed efficiency figures.
Apple M2 Extremely Is Gradual Than Final-Yr CPUs From Intel & AMD, 50% Slower Than NVIDIA RTX 4080
The Apple M2 Ultra SoC is the flagship design that the corporate has on supply. It incorporates a whole of 134 Billion transistors, 24 CPU cores operating in a 16 (Efficiency) and eight (Effectivity) configuration, as much as 76 GPU cores, & 192 GB of reminiscence that’s unified and accessible throughout the entire chip at a most bandwidth of 800 GB/s. There’s additionally the brand new Neural Engine which has been upgraded with 32 cores and the corporate is claiming a 20% speedup on the CPU & 30% speedup on the GPU facet.
The efficiency benchmarks of the Apple M2 Extremely SoC had been leaked inside Geekbench 6 and the older Geekbench 5. We get a style of each the CPU and GPU efficiency since compute outcomes have additionally been leaked. So let’s begin with the CPU facet, we first have the clock speeds which vary from sub-3 GHz as much as 3.7 GHz. We clearly cannot say for positive that are the Efficiency and Effectivity cores because the benchmark does not separate the cores based mostly on their structure or states.
The CPU managed to put up a single-core rating of as much as 2809 factors in single-core and 21,531 factors within the multi-core checks. For comparability, the Intel Core i9-13900KS scores 3083 factors whereas AMD’s Ryzen 9 7950X scores 2875 factors. In multi-threaded benchmarks, the identical chips rating 21665 and 19342 factors, respectively. In order you may see, the workstation-grade Apple M2 Extremely SoC is not quicker than the mainstream CPU choices from Intel and AMD.
0
5000
10000
15000
20000
25000
30000
Should you evaluate the chip to one thing like an AMD Threadripper and Intel Xeon W chip, then these would completely crush the M2 Extremely within the multi-threaded checks however the single-threaded lead shall be on Apple’s M2 Extremely since AMD & Intel Professional CPUs are designed for multi-threaded heavy workloads and that is the place they prolong their multi-core management to.
0
10000
20000
30000
40000
50000
60000
Shifting on to the Compute benchmarks, here is the place I noticed comparisons being made in opposition to the NVIDIA RTX 4080 and AMD GPUs throughout the Metallic API which is predominantly optimized for Apple SoC’s. NVIDIA and AMD GPUs aren’t identified for his or her Metallic particular optimizations because the market share of these GPUs operating Apple OS could be very small and non-existent. As such, in OpenCL, the M2 Extremely SoC finally ends up 50% slower than NVIDIA’s RTX 4080 and that is not even the flagship GPU.
Within the extra well-liked OpenCL API, the Apple M2 Extremely solely finally ends up near the NVIDIA GeForce RTX 4060 Ti which is a mainstream graphics card and sure, that is the total 76 GPU core configuration of the M2 Extremely that we’re evaluating.
As soon as once more, these benchmarks show that it is wiser to take the numbers and efficiency introduced by Apple (and all different corporations for that matter) with a grain of salt and take a look at impartial outcomes to see if the numbers maintain as much as their claims. We have solely used mainstream elements for comparability in these benchmarks and utilizing workstation-oriented options will tilt the outcomes extra in favor of Intel, NVIDIA, and AMD who’ve far more highly effective CPU and GPU designs being provided to Professional and Workstation customers on the market.
Moreover, Apple’s M2 Extremely SOC-based Mac Pro offerings will come at a base price of $7000 US and that is not even utilizing the total SoC specs or the 192 GB unified reminiscence.