The IRBuilder doesn't know that two floating point to integer instructions have constrained equivalents. This patch adds the support by building on the strict FP mode now present in the IRBuilder.
Details
Details
Diff Detail
Diff Detail
- Repository
- rL LLVM