There are many kinds of instances in Amazon Web Services (AWS) (nearly 200-300 instance types). These instances are decided into classes based on the task they are specialized in. Here we need more computing power. So, we are going to use C-class for the worker nodes. And for the mater nodes, as we are not performing any computationally intense process, we can continue with General-Purpose instances. As listed in Table 1, in the C-class after comparing many instances by taking the rendering time, cost per hour, etc into consideration, we came up with a conclusion that c5a.4xlarge is the best kind of instance as it gives better output for lesser price and that is the main motive of this render farm. Having multiple nodes working together on the same blender project would have the rendering time of the whole project go down drastically. And in cloud computing, using one instance for 10 hours or using 10 instances for 1hour costs the same. So, the concept of cluster computing is used to get the output even faster. Table 2 shows the comparison of the rendering times while the number of worker nodes differs. Assuming,
-
Each frame takes 5 min to render
-
Configuration of nodes and laptop are the same
It can be clearly seen from Table 2 that the proposed rendering process takes lesser time that conventional method which leads to better computational efficiency. Table 3 compares the costs of one of the existing render farms for blender software and how this optimized render farm is cheaper than the existing solutions.