M1 Pro and M1 Max Geekbench and Cinebench benchmarks show how fast they are [U: GFXbench 5]

Even though they are not out yet in the hands of users, Apple’s M1 Pro and M1 Max have made appearances in Geekbench and Cinebench benchmarks and show us what they feature and what are they capable of. As expected, they destroy most of the competition in terms of performance per watt usage.

The new details posted on Geekbench and Cinebench (on CPU-Monkey) show where it stands against the competition, and how it scores in these popular benchmarks. Here is what we have found so far, including what we already know.

M1 Pro and M1 Max specifications and benchmarks

M1 Pro technical specifications

Process: 5nm
CPU Frequency: 3.20GHz
Cores: 10 – 8x Firestorm (high performance) + 2x Icestorm (high efficiency)
GPU frequency: 3.20GHz
Execution units: 2048
Maximum Memory: Up to 32GB
Memory type: LPDDR5-5500
Memory Channels: 6
Memory bandwidth: 200GB/s
PCIe version: 4.0
TDP: 30W
Max TDP: 60W
L2 cache: 28.00 MB
FP32 performance: 5200 GFLOPS

M1 Pro Geekbench scores

Single-core: 1768
Multi-core: 11680

M1 Pro Cinebench R23 scores

Single-core: 1562
Multi-core: 14970

M1 Max technical specifications

Process: 5nm
CPU Frequency: 3.20GHz
Cores: 10 – 8x Firestorm (high performance) + 2x Icestorm (high efficiency)
GPU frequency: 3.20GHz
Execution units: 4096
Maximum Memory: Up to 64GB
Memory type: LPDDR5-5500
Memory Channels: 12
Memory bandwidth: 400GB/s
PCIe version: 4.0
TDP: 30W
Max TDP: 60W
L2 cache: 28.00 MB
FP32 performance: 10400 GFLOPS

M1 Max Geekbench scores

Single-core: 1768
Multi-core: 11968
Metal: 68870

M1 Max Cinebench scores

Single-core: 1562
Multi-core: 14970

Both of these chips rank behind other Intel and AMD that run at 95W – 140W and have more cores.

In terms of where they stack against actual GPUs from Nvidia and AMD, we have some calculations from NotebookCheck They have estimated the teraflops values of the new chips as follows:

M1 8-core = 2.6 TF

M1 Pro 14-core = 4.5 TF

M1 Pro 16-core = 5.2 TF

M1 Max 24-core = 7.8 TF

M1 Max 32-core = 10.4 TF

This puts the M1 Max 32-core at the same level as a desktop Nvidia RTX 2080 and Sony PlayStation 5. Keep in mind that Apple is doing this on battery-powered devices, that provide the same performance whether plugged in or not, and also at a maximum TDP of 60W.

When we look at Apple’s own numbers, it is important to look to remember that they said that the A15 was 50 percent faster than Snapdragon 888, but it turned out to be 62 percent faster as per reviews. We expect such M1 Pro and M1 Max comparisons to come out as soon as the new MacBook Pro models are available to reviewers and consumers next week.

GXBench

New benchmarks have surfaced on GFXBench for M1 Max, as discovered by a Redditor. The numbers show that it competes against Nvidia RTX 3080 Mobile, and AMD Radeon RX 6800M. In most of these benchmarks, the M1 Max GPU outperforms its competitors.

M1 Max (at 58W)

Aztec Ruins Normal Tier (offscreen): 703.2 fps

Aztec Ruins High Tier (offscreen): 292.8 fps

Car Chase: 425.1 fps

Manhattan 3.1: 946.4 Fps

Texturing (offscreen): 57664 MTexels/s

Nvidia RTX 3080 Mobile (80-150W)

Aztec Ruins Normal Tier (offscreen): 455.1 fps

Aztec Ruins High Tier (offscreen): 220.9 fps

Car Chase: 438.9 fps

Manhattan 3.1: 528.1 fps

Texturing (offscreen): 226365 MTexels/s

AMD Radeon RX 6800M

Aztec Ruins Normal Tier (offscreen): 390.9 fps

Aztec Ruins High Tier (offscreen): 242.0 fps

Car Chase: 298.7 fps

Manhattan 3.1: 389.5 fps

Texturing (offscreen): 19196 MTexels/s