use(p100k[i32])

	use(s[ui32])
	use(a1[ui32])
	use(a1k[ui32])
	use(a100k[ui32])
	use(p1[ui32])
	use(p1k[ui32])
	use(p100k[ui32])

	use(s[i64])
	use(a1[i64])
	use(a1k[i64])
	use(a100k[i64])
	use(p1[i64])
	use(p1k[i64])
	use(p100k[i64])

	use(s[ui64])
	use(a1[ui64])
	use(a1k[ui64])
	use(a100k[ui64])
	use(p1[ui64])
	use(p1k[ui64])
	use(p100k[ui64])

                       %n: i64 {tfr.name="N"}) -> !tfr.tensor {
  %index = arith.constant 0 : index
  %cst = arith.constant 1 : i64
  %eq = arith.cmpi "eq", %n, %cst : i64
  %v1 = tfr.get_element %values[%index] : (!tfr.tensor_list, index) -> !tfr.tensor
  %res = scf.if %eq -> !tfr.tensor {
    scf.yield %v1 : !tfr.tensor
  } else {
    %step = arith.index_cast %cst : i64 to index
    %end = arith.index_cast %n : i64 to index

                       %n: i64 {tfr.name="N"}) -> !tfr.tensor {
  %index = arith.constant 0 : index
  %cst = arith.constant 1 : i64
  %eq = arith.cmpi eq, %n, %cst : i64
  %v1 = tfr.get_element %values[%index] : (!tfr.tensor_list, index) -> !tfr.tensor
  %res = scf.if %eq -> !tfr.tensor {
    scf.yield %v1 : !tfr.tensor
  } else {
    %step = arith.index_cast %cst : i64 to index
    %end = arith.index_cast %n : i64 to index

    %0 = "tf.FakeQuantWithMinMaxArgs"(%arg0) {device = "", max = 2.000000e-01 : f32, min = -1.000000e-01 : f32, narrow_range = false, num_bits = 8 : i64} : (tensor<1x3x4x3xf32>) -> tensor<1x3x4x3xf32>

"builtin.module"() ({
  "func.func"() ({
  ^bb0(%arg0: tensor<i32>):
    %1617 = "tf.While"(%arg0) {_lower_using_switch_merge = true, _num_original_outputs = 7 : i64, _read_only_resource_inputs = [], body = @main_while_body_4225150, cond = @main_while_cond_4225140, device = "", is_stateless = false, parallel_iterations = 10 : i64, shape_invariant} : (tensor<i32>) -> (tensor<i32>)
    "func.return"() : () -> ()

      %0 = "tf.opA"(%arg0) {_xla_compile_device_type = "TPU", _replication_info = "cluster"} : (tensor<i1>) -> tensor<i32>
      %1 = "tf.While"(%0) {body = @while_body_with_cluster_attr, cond = @while_cond_with_cluster_attr, is_stateless = false, name = "A", parallel_iterations = 10 : i64} : (tensor<i32>) -> tensor<i32>

  // CHECK: RemoteCall
  // CHECK-SAME: (tensor<!tf_type.string>, tensor<!tf_type.string>, tensor<i64>) -> tensor<!tf_type.string>
  %0 = "tf.RemoteCall"(%arg1, %arg0, %arg2) <{f = @__inference__next_func_3760}> {device = ""} : (tensor<!tf_type.string>, tensor<!tf_type.string>, tensor<i64>) -> tensor<!tf_type.string>
  return
}

    %0 = stablehlo.convolution(%arg0, %arg1) dim_numbers = [b, 0, 1, f]x[0, 1, i, o]->[b, 0, 1, f], window = {pad = [[0, 1], [1, 1]]} {batch_group_count = 1 : i64, feature_group_count = 1 : i64} : (tensor<1x3x4x3xf32>, tensor<2x3x3x2xf32>) -> tensor<1x3x4x2xf32>
    return %0 : tensor<1x3x4x2xf32>
  }
}

// CHECK-LABEL: quantize_conv_per_tensor
// CHECK-SAME: %[[ARG0:.+]]: tensor<1x3x4x3xf32>

        %3 = "tf.While"(%1) {body = @while_body_without_cluster_attr, cond = @while_cond_with_cluster_attr, is_stateless = false, name = "C", parallel_iterations = 10 : i64} : (tensor<i32>) -> tensor<i32>

		go i64sender(c64, sync)
		if buffer > 0 {
			<-sync
		}
		try = 0
	Recv64:
		for {
			select {
			case i64 = <-c64:
				break Recv64
			default:
				try++
				if try > maxTries {
					panic("i64sender")
				}
				sleep()
			}
		}
		if i64 != 234567 {
			panic("i64sender value")
		}
		if buffer == 0 {
			<-sync
		}

		go breceiver(cb, sync)
		try = 0