Calculation of the parallel capacity of computer systems based on iteratively executing according to step variable

A method comprising:

• providing input features to a machine learning model, the input features comprising one or more system parameters of a parallel processing system;
• obtaining, as output from the machine learning model, an initial starting value for a step variable, the machine learning model having been trained on training data comprising prior executions of capacity determination operations for parallel processing systems other than the parallel processing system;
• iteratively executing increasing numbers of execution units in the parallel processing system, the increasing numbers being determined based on the step variable, until an execution time of at least one of the executing execution units meets or exceeds a threshold derived from a baseline execution time;
• determining a maximum capacity of the parallel processing system based on the iteratively executing;
• and
• operating the parallel processing system with no more than the maximum capacity of execution units.

A non-transitory machine-readable medium storing instructions which, when executed by one or more processors, cause the one or more processors to perform operations comprising:

• providing input features to a machine learning model, the input features comprising one or more system parameters of a parallel processing system;
• obtaining, as output from the machine learning model, an initial starting value for a step variable, the machine learning model having been trained on training data comprising prior executions of capacity determination operations for parallel processing systems other than the parallel processing system;
• iteratively executing increasing numbers of execution units in the parallel processing system, the increasing numbers being determined based on the step variable, until an execution time of at least one of the executing execution units meets or exceeds a threshold derived from a baseline execution time;
• determining a maximum capacity of the parallel processing system based on the iteratively executing;
• and
• operating the parallel processing system with no more than the maximum capacity of execution units.