Methods and apparatus for adjacency network delivery of operands to instruction specified destinations that reduces storage of temporary variables

A method of executing a chain of instructions as part of a program executed by a processor, the method comprising:

• receiving a chain of instructions that includes a first instruction and, in the chain of instructions, a destination instruction having multiple dependencies on the first instruction;
• decoding the first instruction to identify a first function that generates a first result, a first storage unit for storing the first result, that an instruction at an instruction position encoded in the first instruction is the destination instruction, and first adjustment information required to execute the destination instruction;
• decoding the identified destination instruction to identify a second function that generates a second result and a type of execution unit;
• executing first function according to the first instruction to generate the first result and deliver the first result and the first adjustment information through an internal network of execution units and associated storage units (EXstore units) to the identified first storage unit and to a second storage unit, respectively, wherein the identified first storage unit and the second storage unit are both located with a destination execution unit of the identified type and that is part of the internal network and wherein the EXstore units of the internal network are arranged according to a Row by Column (R×C) array and configured with interconnection wiring between the Exstore units according to a specified adjacency of connections for the R×C array, R and C positive integers;
• and
• executing the destination instruction on the destination execution unit by accessing the first result from the identified first storage unit and the first adjustment information from the second storage unit, operating on the accessed first result according to the second function and according to the first adjustment information, and generating the second result for use in the program executed by the processor.

A method of executing a chain of instructions as part of a program by a processor, the method comprising:

• receiving a chained execution packet (CEP) that includes control information for executing the CEP and a chain of instructions that includes a first instruction, and a destination instruction having a dependency on the first instruction, wherein the first instruction identifies a first function that generates a first result, the destination instruction based on an instruction position encoded in the first instruction, and a storage unit for receiving the first result;
• decoding the CEP to determine for each instruction in the chain of instructions, including the first instruction and the destination instruction, an associated execution unit slot position in a prolog instruction code memory (PICM);
• executing the first function according to the first instruction to generate the first result for delivery through an internal network of execution units and associated storage units (EXstore units) to the identified storage unit located with a destination execution unit that is part of the internal network and storing the first instruction in an execution unit slot position associated with the first instruction in the PICM, wherein the EXstore units of the internal network are arranged according to a Row by Column (R×C) array and configured with interconnection wiring between the EXstore units according to a specified adjacency of connections for the R×C array, R and C positive integers;
• executing the identified destination instruction on the destination execution unit by accessing the first result from the identified storage unit, operating on the accessed first result according to a second function specified by the identified destination instruction, generating a second result, and storing the identified destination instruction in an execution unit slot position associated with the identified destination instruction in the PICM;
• and
• continuing executing and storing the instructions from the CEP in the PICM during a prolog of the program until execution slot positions are filled according to a body of the program, wherein in the body of the program includes the first instruction and the identified destination instruction.