Exercise 8
================

.. admonition:: Exercise
   :class: todo

   **Time: 45 min**

      1. Inspect and then run the code `exercises/exercise_8/9_vector_add_async.cu` to see you CUDA graphs work.
      2. Inspect and then run the code `exercises/exercise_8/10_comparisson.cu` to comare the performance of the CUDA graphs with normal kernel laucnhes.
      3. Inspect and then run the code `exercises/exercise_8/11_custom_graph.cu` to see how to create a custom graph.
      4. Inspect and then run the code `exercises/exercise_8/12_update_node_args.cu` to see how to update the arguments of a node in a graph.