On 9/28/18 12:56 PM, Wilco Dijkstra wrote: > We've seen too many instances where not keeping a well defined boundary > between int and fp/simd leads to bad code, so not defining all possible legal > combinations is intended. I'll check whether 32-bit w->r and w->w > zero-extension > could ever trigger.
As a guess, I would expect them to happen after some sort of vector reduction operation. r~
