[PATCH] D54719: [Intrinsic] Signed Fixed Point Multiplication Intrinsic

[Craig Topper via Phabricator via llvm-commits]

craig.topper added inline comments.


================
Comment at: llvm/lib/CodeGen/SelectionDAG/TargetLowering.cpp:5101
+  if (Scale) {
+    SDValue Hi = DAG.getNode(ISD::MULHS, dl, VT, LHS, RHS);
+    EVT ShiftTy = getShiftAmountTy(VT, DAG.getDataLayout());
----------------
ebevhan wrote:
> Could use a couple comments explaining what we're doing with the values/SRL/SHL.
> 
> Does this work if MULHS in VT is of dubious legality?
Ideally we'd use MUL_LOHI if the target supports it. That should allow X86 to use a single multiply instruction in the test cases.


Repository:
  rL LLVM

https://reviews.llvm.org/D54719