Internal pipeline for run_model and run_script commands #43
Description
Description
When executing run_model and run_script with a Redis cluster, data movement is needed to make sure that the input and output tensors conform to the hash slot data locality requirements for Redis. As a result, the run_model() and run_script() commands actually consist of multiple Redis commands. Pipelining should make these multiple commands more efficient. Note that this is not a feature to expose pipelining the user.
Justification
Pipelining the run_model and run_script API calls should increase performance.
Implementation Strategy
This feature can be implemented with the pipelining features in underlying redis libraries.