// detection graph, disconnecting each loop from the enclosing graph.
  // Specifically, we:
  // * add a new "frame" node for each loop.
  // * replace edges to "Enter" nodes, and edges from "Exit" nodes with edges
  //   to/from the corresponding frame node. In essence, we collapse the loop
  //   into a single node for the purpose of cycle detection in the enclosing
  //   graph.

      /*parallel_iterations=*/10, false,
      /*shape_invariant=*/false);

  // `_lower_using_switch_merge` is the default for While ops created
  // in TensorFlow and allows lowering to V1 control flow for loop
  // parallelization.
  dataset_while->setAttr("_lower_using_switch_merge",
                         builder.getBoolAttr(true));

  return dataset_while;
}

	HotPgoInitial          = HotPgo | HotInitial                // special case; single block loop, initial block is header block has a flow-in entry, but PGO says it is hot
	HotPgoInitialNotFLowIn = HotPgo | HotInitial | HotNotFlowIn // PGO says it is hot, and the loop is rotated so flow enters loop with a branch

            }
          }
        }
      }
    }

    // Merge node causes a loop so we remove NextIteration->Merge edge before
    // performing shape inference. But removing those edges also prevents us
    // from inferring output shape for Merge node (we need shapes for all its
    // inputs).
    // For loop invariant resource input's Merge node, we set output resource
    // shape as Enter node's resource shape.

                                TF::WhileOp* while_op) {
  // Given a while loop body function 'body_func', find the tf.While Op that
  // uses it.
  auto uses_optional = body_func.getSymbolUses(module);
  if (!uses_optional.has_value()) {
    body_func.emitOpError() << "no use of while loop body";
    return LogicalResult::failure();
  }
  *while_op = nullptr;

      %zero = arith.constant dense<0> : tensor<i32>
      %ne = "tf.NotEqual"(%carg1, %zero) : (tensor<i32>, tensor<i32>) -> tensor<i1>
      "tf.Yield"(%ne) : (tensor<i1>) -> ()
    },
    {
      // loop body
      ^bb0(%barg0: tensor<*xf32>, %barg1: tensor<i32>):
      %add = "tf.Add"(%barg0, %barg0) : (tensor<*xf32>, tensor<*xf32>) -> tensor<*xf32>
      %one = arith.constant dense<1> : tensor<i32>

// Test to verify loop outlining.

// RUN: tf-opt --split-input-file --tfl-while-loop-outline %s | FileCheck %s
// Check that while loop outlining is nop if re-ran.
// RUN: tf-opt --tfl-while-loop-outline %s -o %t1
// RUN: tf-opt --tfl-while-loop-outline %t1 -o %t2
// RUN: diff %t1 %t2

// CHECK-LABEL: func @while
func.func @while() -> tensor<1xf32>
    attributes {tf.entry_function = {outputs = "result"}} {

  // Collect external values used.
  llvm::SetVector<Value> extern_values;

  // The basic block arguments correspond to values that are loop carried, while
  // all those post are loop independent. Initialize extern_values with while_op
  // not loop carried operands.
  auto num_loop_carried = while_op.getCond().getNumArguments();
  auto not_carried_operands =
      while_op.getOperands().drop_front(num_loop_carried);

	// Eliminate bounds checks in the loop.
	size := len(x.limbs)
	xLimbs := x.limbs[:size]
	yLimbs := y.limbs[:size]

	equal := yes
	for i := 0; i < size; i++ {
		equal &= ctEq(xLimbs[i], yLimbs[i])
	}
	return equal
}

// IsZero returns 1 if x == 0, and 0 otherwise.
func (x *Nat) IsZero() choice {
	// Eliminate bounds checks in the loop.
	size := len(x.limbs)
	xLimbs := x.limbs[:size]

		}
	}
	m.cache.RUnlock()

	// Loop through all inactive pods and call UnprepareResources on them.
	for _, podClaims := range inactivePodClaims {
		if err := m.unprepareResources(podClaims.uid, podClaims.namespace, podClaims.claimNames); err != nil {
			klog.ErrorS(err, "Unpreparing pod resources in reconcile loop", "podUID", podClaims.uid)
		}
	}
}