M1 Pro, 16GB.
My Cinebench R23 testing is as follows:
- Tech Preview Update 2, 8 cores, 8GB RAM: 2500 Multicore
- Release client, 8 cores 8GB RAM: 2700 Multicore
- MacOS Native performance: 7700 Multicore
Other machines I have at hand (most to show how impressive the M1 is given it's wattage, lack of hyperthreading and only 4 performance cores):
- Ryzen 7 5800X, 4.75Ghz Allcore, 140w draw: 16200 Multicore
- Core i9 10850k, 4.8Ghz Allcore, mild undervolt, 250w draw: 15800 Multicore
Other observations:
I use a piece of software called S-Frame for structural design. The calculations it runs are essentially massive matrix operations (it utilizes finite element methods to determine structural displacement/reactions). Running a 2D mesh comprised of about 1000 nodes under 10 condition analysis takes my Ryzen 5800x machine longer than my M1 on Parallels.
My Cinebench R23 testing is as follows:
- Tech Preview Update 2, 8 cores, 8GB RAM: 2500 Multicore
- Release client, 8 cores 8GB RAM: 2700 Multicore
- MacOS Native performance: 7700 Multicore
Other machines I have at hand (most to show how impressive the M1 is given it's wattage, lack of hyperthreading and only 4 performance cores):
- Ryzen 7 5800X, 4.75Ghz Allcore, 140w draw: 16200 Multicore
- Core i9 10850k, 4.8Ghz Allcore, mild undervolt, 250w draw: 15800 Multicore
Other observations:
I use a piece of software called S-Frame for structural design. The calculations it runs are essentially massive matrix operations (it utilizes finite element methods to determine structural displacement/reactions). Running a 2D mesh comprised of about 1000 nodes under 10 condition analysis takes my Ryzen 5800x machine longer than my M1 on Parallels.