[clang] [CIR] Added fast-math flags to LLVM intrinsic calls (PR #224899)

[Kunal Dubey via cfe-commits]

================
@@ -830,6 +830,15 @@ class CIRGenBuilderTy : public cir::CIRBaseBuilderTy {
                                             std::forward<Operands>(op)...)
         .getResult();
   }
+
+  mlir::Value emitIntrinsicCallOp(mlir::Location loc, const llvm::StringRef str,
+                                  const mlir::Type &resTy,
+                                  mlir::ValueRange operands,
----------------
xakep8 wrote:

Yes that should be better, I was building it with a very narrow use case of __builtin_reduce_assoc_fadd and that already builds a SmallVector so I used the ValueRange but the better shape would be `Operands &&...op`. I'll change this one

https://github.com/llvm/llvm-project/pull/224899