This again proves that realistically we're hardware bound (waiting for nvidia to release bigger, more efficient and better hardware. We won't see any improveme

This again proves that realistically we're hardware bound (waiting for nvidia to release bigger, more efficient and better hardware. We won't see any improvements until the next generation of hardware comes out and allows us to do things that were computationally impossible before. Yes, of course we can develop better transformers and improve reasoning capabilities, but that's not what I am talking about. I am mostly referencing the ability for newest "AI" hardware to multiply 256 matrices in a single clock cycle per core.