System and method for fusing instructions

A compiler, comprising:

• a fused instruction generator that is to
• find a first instruction and a second instruction in a program file that each perform a single operation and the second instruction depends on a result of the first instruction;
• form a fused instruction that includes
• a fused opcode that represents both a first operation of the first instruction and a second a operation of the second instruction, wherein the fused opcode swaps the order of applying a temporary operand, that stores the result of the first instruction, to the at least one source operand of the second instruction that is not the destination operand of the first instruction, if the second instruction is non-commutative and swaps the order of applying a first one of the at least one source operand of the second instruction to a second one of the at least one source operand of the second instruction;
• and
• store said fused instruction in a memory device.

A method for executing a fused instruction within one clock cycle, comprising:

• finding a first instruction and a second instruction in the program file that each perform a single operation and the second instruction depends on a result of the first instruction;
• forming a fused instruction that includes
• determining a fused opcode that represents both a first operation of the first instruction and a second operation of the second instruction, wherein the fused opcode swaps the order of applying a temporary operand, that stores the result of the first instruction, to the at least one source operand of the second instruction that is not the destination operand of the first instruction, if the second instruction is non-commutative and swaps the order of applying a first one of the at least one source operand of the second instruction to a second one of the at least one source operand of the second instruction;
• performing a first operation in a first arithmetic logic unit (ALT) of a fused instruction execution unit and a second operation in a second ALU of the fused instruction execution unit,
• wherein a result of the first ALU is input into the second ALU and within one clock cycle, the first ALU performs the first operation and the second ALU performs the second operation.

A machine-readable medium having stored thereon data representing sequences of instructions, the sequences of instructions including sequence of instructions which, when executed by a processor, cause the processor to perform the steps of:

• finding a first instruction and a second instruction in the program file that each perform a single operation and the second instruction depends on a result of the first instruction;
• and
• forming a fused instruction that includes
• determining a fused opcode that represents both a first operation of the first instruction and a second operation of the second instruction, wherein the fused opcode swaps the order of applying a temporary operand, that stores the result of the first instruction, to the at least one source operand of the second instruction that is not the destination operand of the first instruction, if the second instruction is non-commutative and swaps the order of applying a first one of the at least one source operand of the second instruction to a second one of the at least one source operand of the second instruction.