All of this is interesting. I'm currently using - as many do - a 2080TI for ML training. With this my training is <30min. I would not use a laptop for trainin

All of this is interesting. I'm currently using - as many do - a 2080TI for ML training. With this my training is <30min. I would not use a laptop for training as I no longer use "developer laptops" that are more expensive with lower performance than my Linux desktop, especially now with homeoffice I no longer need a laptop - I understand others have a need for laptops YMMV. But the benchmarks could give a glimpse for a desktop M1. What I want to know, is it 0.5x or 2x the speed of a 2080TI.