[clang] [CIR] X86 vector fcmp-sse vector builtins (PR #167125)

[Andy Kaylor via cfe-commits]

================
@@ -43,7 +43,56 @@ mlir::Value CIRGenFunction::emitX86BuiltinExpr(unsigned builtinID,
   // Find out if any arguments are required to be integer constant expressions.
   assert(!cir::MissingFeatures::handleBuiltinICEArguments());
 
-  switch (builtinID) {
+  llvm::SmallVector<mlir::Value, 4> Ops;
+
+  // Find out if any arguments are required to be integer constant expressions.
+  unsigned ICEArguments = 0;
+  ASTContext::GetBuiltinTypeError Error;
+  getContext().GetBuiltinType(BuiltinID, Error, &ICEArguments);
+  assert(Error == ASTContext::GE_None && "Should not codegen an error");
+  for (unsigned i = 0, e = E->getNumArgs(); i != e; i++) {
+    Ops.push_back(emitScalarOrConstFoldImmArg(ICEArguments, i, E));
+  }
+
+  // OG has unordered comparison as a form of optimization in addition to
+  // ordered comparison, while CIR doesn't.
----------------
andykaylor wrote:

This is fine for the current PR, but we should really revisit this. Both the `llvm` and `arith` dialects have separate `icmp` and `fcmp` operations, each with their own set of predicates. We should follow that model.

https://github.com/llvm/llvm-project/pull/167125