[PATCH] D28696: DAG: Constant fold fp16_to_fp/fp16_to_fp

[Matt Arsenault via Phabricator via llvm-commits]

arsenm added a comment.

In https://reviews.llvm.org/D28696#657465, @RKSimon wrote:

> LGTM - How come we haven't been able to remove these intrinsics and just use (f32 fpext(f16 bitcast(i16 src))) style patterns?


Not sure. I think we could even autoupgrade these, though the nodes are still handy for targets without legal f16


https://reviews.llvm.org/D28696