[llvm] [AMDGPU] Generic lowering for rint and nearbyint (PR #69596)

[Matt Arsenault via llvm-commits]

================
@@ -1888,6 +1888,14 @@ class MachineIRBuilder {
     return buildInstr(TargetOpcode::G_FRINT, {Dst}, {Src0}, Flags);
   }
 
+  /// Build and insert \p Dst = G_INTRINSIC_ROUNDEVEN \p Src0, \p Src1
+  MachineInstrBuilder
+  buildintrinsicRoundeven(const DstOp &Dst, const SrcOp &Src0,
----------------
arsenm wrote:

 Capitalization is off. I'd also probably just drop the intrinsic part from the name (we should also drop it from the opcode name)

https://github.com/llvm/llvm-project/pull/69596