Apparatus and computer-implemented method for determining a memory plan for executing operations

A computer-implemented method for determining a memory plan for executing operations of an artificial neural network, the method comprising the following steps:

• creating a list of memory areas required for executing the operations;
• depending on the list, determining for the operations which memory areas must be present in a first memory for executing a respective operation and which memory areas may be present in a second memory during the execution of the respective operation;
• determining the memory plan depending on whether a memory area must be present in the first memory for execution or may be present in the second memory;
• for each operation, determining an assignment of the memory areas one of the the first memory and to the second memory;
• for each operation, marking the memory areas in the list that do not have to be in the first memory are marked;
• and
• determining the memory plan depending on the list with the marked memory areas, and determining the assignment that minimizes a number of transfers of memory areas between the first memory and the second memory.

An apparatus for determining a memory plan for executing operations of an artificial neural network, the apparatus comprising:

• at least one processor;
• and
• at least one memory, wherein the at least one memory includes instructions which can be executed by the at least one processor, and, upon execution of which by the at least one processor, the following steps are executed:
• creating a list of memory areas required for executing the operations,
• depending on the list, determining for the operations which memory areas must be present in a first memory for executing a respective operation and which memory areas may be present in a second memory during the execution of the respective operation,
• determining the memory plan depending on whether a memory area must be present in the first memory for execution or may be present in the second memory,
• for each operation, determining an assignment of the memory areas one of the the first memory and to the second memory,
• for each operation, marking the memory areas in the list that do not have to be in the first memory are marked, and
• determining the memory plan depending on the list with the marked memory areas, and determining the assignment that minimizes a number of transfers of memory areas between the first memory and the second memory.

A non-transitory storage medium on which is stored a computer program including computer-readable instructions for determining a memory plan for executing operations of an artificial neural network, the instructions, when executed by at least one processor, causing the at least one comprising the following steps:

• creating a list of memory areas required for executing the operations;
• depending on the list, determining for the operations which memory areas must be present in a first memory for executing a respective operation and which memory areas may be present in a second memory during the execution of the respective operation;
• determining the memory plan depending on whether a memory area must be present in the first memory for execution or may be present in the second memory;
• for each operation, determining an assignment of the memory areas one of the the first memory and to the second memory;
• for each operation, marking the memory areas in the list that do not have to be in the first memory are marked;
• and
• determining the memory plan depending on the list with the marked memory areas, and determining the assignment that minimizes a number of transfers of memory areas between the first memory and the second memory.