How to find index of maximum value of a 2D tensor?

Question

I have three questions:

1. Is there any way to extract maximum values and its indices from a 2D tensor?

Suppose we have a tensor x = tf.constant([[0, 2, 1], [0, 0, 8], [2, 9, 0]]), the desired output would be max = 9, index = [2,1].

I've tried the function tf.argmax, however the argument axis of tf.argmax doesn't accept tuple type, for exmple: it doesn't works when axis = (0, 1), only when axis = 0 or axis = 1.

I have referred to this question but they did not answer my question above.

2. Also, how can I find a number of the maximum value (let's say the first 5 maximum value) of a 2D tensor, since tf.reduce_max only returns 1?

3. And after having the indices, how can I use these indices to index the value from another tensor y that has the same size of x above?

Update 1: As Aldream pointed out that the first two of my question is a duplicate of this question How to find the top k values in a 2-D tensor in tensorflow. Still, the third question remains.

Update 2: For the third question, if anyone has the same problem, we can use tf.gather.