I'm seeing this misunderstanding a lot recently. There's TWO components to putting together a viable machine learning rig: - Fitting models in memory - Infere

I'm seeing this misunderstanding a lot recently. There's TWO components to putting together a viable machine learning rig: - Fitting models in memory - Inference / Training speed 8 x RTX 3090s will absolutely CRUSH a single Mac Studio in raw performance.