Industry
Most researchers agree that autoregression is best when memory bandwidth is cheap and diffusion is best when FLOPS are cheap. They also admi…
Most researchers agree that autoregression is best when memory bandwidth is cheap and diffusion is best when FLOPS are cheap. They also admit the future of compute is all FLOPS because memory scaling
Most researchers agree that autoregression is best when memory bandwidth is cheap and diffusion is best when FLOPS are cheap. They also admit the future of compute is all FLOPS because memory scaling is hard and scaling FLOPS is easy. So why not go all in on diffusion????
Source: David Holz (X) | 2026-05-27