[CORE] Fixed several issues that arose in the development of the

torch-blocksparse package:

* Now using warp shuffle in reductions when possible
* Various bugfixes in layout inference
* Added INFINITY, exponential and select
* Better error messages for unimplemented constructs

lib/lang/ast.cc

@@ -655,6 +655,9 @@ void UnaryOp::TypeChecking() {
   case Token::REDUCE:
     return ReduceOpTypeChecking();
 
+  case Token::EXP:
+    return IntrinsicOpTypeChecking();
+
   default:
     assert(false);
   }
@@ -769,6 +772,10 @@ void UnaryOp::CastOpTypeChecking() {
   }
 }
 
+void UnaryOp::IntrinsicOpTypeChecking() {
+  type_ = ScalarOrLikeTile(operand_, ArithmType::New(T_FLOAT));
+}
+
 /*
  * Transposition Operator
  */