Yeah I remember a few projects attempting "grid computing", where the idea was distributing application threads across CPUs over freaking ethernet.
Have to say that would be a fun puzzle to try and optimize for, but network latency would always be a hard physical constraint no matter how fast. Maybe some niche use cases, but then multi-core CPUs and GPU processing really took off and I guess it just got even less useful.
Have to say that would be a fun puzzle to try and optimize for, but network latency would always be a hard physical constraint no matter how fast. Maybe some niche use cases, but then multi-core CPUs and GPU processing really took off and I guess it just got even less useful.