Skip to content

microsoft/clc: Only apply float scaling to 32bit fdiv

Jesse Natalie requested to merge jenatali/mesa:scale-fdiv-32bit-only into master

Currently the fdiv scaling pass is run early on, before constant folding. We don't support doubles, so if we have any fdiv ops on doubles, they have to get constant-folded away or else the compile will fail, so don't bother trying to introduce precision-increasing scaling.

Merge request reports