System and method for retiring approximately simultaneously a group of instructions in a superscalar microprocessor

• US 7,934,078 B2
• Filed: 09/17/2008
• Issued: 04/26/2011
• Est. Priority Date: 05/01/1992
• Status: Expired due to Fees

• Current Assignee: Samsung Electronics Co. Ltd.
• Original Assignee: Seiko Epson Corporation (Seiko Group)
• Inventors: Wang, Johannes, Deosaran, Trevor, Garg, Sanjiv
• Primary Examiner(s): ELLIS, RICHARD L
• Application Number:
  US12/212,361
• Publication Number:
  US 20090013155A1
• Time in Patent Office: 951 Days
• Field of Search: 712/218, 712/217, 712/206
• US Class Current: 712/217
• CPC Class Codes: G06F 9/3012 Organisation of register space, e.g. banked or distributed register file
• CPC Class Codes: G06F 9/3013 according to data content, e.g. floating-point registers, address registers
• CPC Class Codes: G06F 9/3836 Instruction issuing, e.g. dynamic instruction scheduling or out of order instruction execution
• CPC Class Codes: G06F 9/3838 Dependency mechanisms, e.g. register scoreboarding
• CPC Class Codes: G06F 9/384 Register renaming
• CPC Class Codes: G06F 9/3853 of compound instructions
• CPC Class Codes: G06F 9/3856 Reordering of instructions, e.g. using queues or age tags
• CPC Class Codes: G06F 9/3858 Result writeback, i.e. updating the architectural state or memory
• CPC Class Codes: G06F 9/3863 using multiple copies of the architectural state, e.g. shadow registers
• CPC Class Codes: G06F 9/3885 using a plurality of independent parallel functional units

Abstract

An system and method for retiring instructions in a superscalar microprocessor which executes a program comprising a set of instructions having a predetermined program order, the retirement system for simultaneously retiring groups of instructions executed in or out of order by the microprocessor. The retirement system comprises a done block for monitoring the status of the instructions to determine which instruction or group of instructions have been executed, a retirement control block for determining whether each executed instruction is retirable, a temporary buffer for storing results of instructions executed out of program order, and a register array for storing retirable-instruction results. In addition, the retirement control block further controls the retiring of a group of instructions determined to be retirable, by simultaneously transferring their results from the temporary buffer to the register array, and retires instructions executed in order by storing their results directly in the register array. The method comprises the steps of monitoring the status of the instructions to determine which group of instructions have been executed, determining whether each executed instruction is retirable, storing results of instructions executed out of program order in a temporary buffer, storing retirable-instruction results in a register array and retiring a group of retirable instructions by simultaneously transferring their results from the temporary buffer to the register array, and retiring instructions executed in order by storing their results directly in the register array.

Claims

1. A superscalar processor adapted to execute at least one instruction out of program order, said superscalar processor comprising:

an instruction window that has a first storage location and a second storage location for storing instructions, an instruction stored in the second storage location being stored in the first storage location when an instruction stored in the first storage location is retired;

a plurality of functional units that execute an instruction out of program order;

a buffer that has storage locations at which an execution result of each instruction is stored;

register renaming circuitry that associates uniquely an address indicating a fixed storage location in the buffer with each instruction included at each storage location in the instruction window, regardless of a change in the storage location of the instruction in the instruction window;

a register array that includes a plurality of array locations referenced so that an execution result of a retired instruction can be provided to the referenced array location;

a retirement control block that determines whether an executed instruction can be retired or not; and

an instruction retirement unit that retires an instruction that can be retired by associating an execution result of each instruction that can be retired with an array location within the register array.

2. The superscalar processor according to claim 1, wherein the instruction window includes a first instruction and a second instruction, the first instruction appears earlier than the second instruction in program order, the second instruction is executed prior to execution of the first instruction, and an execution result of the second instruction is stored in the buffer before an execution result of the first instruction is stored in the buffer.

3. The superscalar processor according to claim 1, wherein the register renaming circuitry associates a first unique address with the first instruction, and associates a second unique address with the second instruction, the buffer has a first storage location and a second storage location, the first unique address indicates the first storage location of the buffer at which a first instruction execution result is stored, and the second unique address indicates the second storage location of the buffer at which a second instruction execution result is stored.

4. The superscalar processor according to claim 3, wherein the first instruction and the second instruction are retired approximately simultaneously.

5. The superscalar processor according to claim 3, wherein the execution result of the first instruction and the execution result of the second instruction are associated with respective register locations of the register array approximately simultaneously.

6. The superscalar processor according to claim 2, wherein the second instruction is not retired before the execution result of the first instruction is stored in the buffer.

7. The superscalar processor according to claim 2, wherein a first plurality of instructions includes the first instruction, a second plurality of instructions includes the second instruction, and the first plurality of instructions appear earlier than the second plurality of instructions in program order.

8. The superscalar processor according to claim 1, wherein instructions are retired in a group that includes at least two instructions.

9. The superscalar processor according to claim 8, wherein the instruction retirement unit is adapted to retire at least two instructions in one clock cycle.

10. The superscalar processor according to claim 1, wherein the associating an execution result is an updating of a value stored at a location of an array, and the array is referenced to provide an in-order state of the processor.

11. A computer system having a processor and a memory adapted to store instructions having a program order, said processor comprising:

an instruction window that has a first storage location and a second storage location for storing instructions, an instruction stored in the second storage location being stored in the first storage location when an instruction stored in the first storage location is retired;

register renaming circuitry that associates uniquely an address indicating a fixed storage location in the buffer with each instruction included at each storage location in the instruction window, regardless of a change in the storage location of the instruction in the instruction window, said register renaming circuitry associating an address with at least one instruction in one clock cycle;

a buffer coupled to the register renaming circuitry, said buffer storing an execution result of an instruction at a location described by an address associated with each instruction;

a plurality of functional units coupled to the buffer, said plurality of functional units executing an instruction out of program order;

an array having a plurality of locations each adapted to identify an execution result of a retiring instruction;

a control block that determines whether an executed instruction can be retired; and

an instruction retiring section coupled to the control block circuitry and the array, said instruction retiring section retiring an instruction that can be retired by associating an execution result of the instruction that can be retired with a location in the array, and allowing the execution results of the instructions that can be retired to be stored at respective particular locations of the array.

12. The computer system according to claim 11, wherein the instruction window includes a first instruction and a second instruction, the first instruction can be executed earlier than the second instruction in program order, the second instruction is executed prior to execution of the first instruction, and an execution result of the second instruction is stored in the buffer before an execution result of the first instruction is stored in the buffer.

13. The computer system according to claim 12, wherein the register renaming circuitry associates a first address with the first instruction, and associates a second address with the second instruction, the buffer has a first location and a second location, the first address identifies the first location of the buffer at which the execution result of the first instruction is stored, and the second address identifies the second location of the buffer at which the execution result of the second instruction is stored.

14. The computer system according to claim 13, wherein the first instruction and the second instruction are retired approximately simultaneously.

15. The computer system according to claim 13, wherein the execution result of the first instruction and the execution result of the second instruction are associated with respective locations of the array in one clock cycle.

16. The computer system according to claim 15, wherein the second instruction is not retired before the execution result of the first instruction is stored in the first location of the buffer.

17. The computer system according to claim 13, wherein a first plurality of instructions includes the first instruction, a second plurality of instructions includes the second instruction, and the first plurality of instructions appear earlier than the second plurality of instructions in program order.

18. The computer system according to claim 11, wherein the number of instructions that can concurrently be retired is any of two, three or four.

19. The computer system according to claim 18, wherein the execution result of the instruction is stored at the first location of the buffer in response to an address associated with the instruction.

20. The computer system according to claim 11, wherein the association of an execution result is an updating of a value stored at a location of an array, and the array is referenced to provide an in-order state of the processor.

21. A superscalar processor for executing at least one instruction out of order, said superscalar processor comprising:

an instruction window that has a first storage location and a second storage location for storing instructions, an instruction stored in the second storage location being stored in the first storage location when an instruction stored in the first storage location is retired;

a buffer for storing an execution result of a first instruction among an instruction group and an execution result of a second instruction among an instruction group;

a superscalar register renaming circuitry that associates, among a first address and a second address each of which indicates a fixed storage location in the buffer, the first address with the first instruction among the instruction group and the second address with the second instruction among the instruction group, regardless of a change in the storage location of the instruction in the instruction window, the first address associated with the first instruction indicating a first location of the buffer at which the execution result of the first instruction is stored, the second address associated with the second instruction indicating a second location of the buffer at which the execution result of the second instruction is stored;

a plurality of functional units that execute the first instruction and the second instruction out of order among the instruction group;

a register array that has a plurality of register array locations for storing an execution result of a retired instruction;

a retirement control block that determines whether the first instruction can be retired and determines whether the second instruction can be retired; and

an instruction retirement unit that retires the first instruction and the second instruction by storing the execution result of the first instruction stored in the first location of the buffer at a first register array location and storing the execution result of the second instruction stored in the second location of the buffer at a second register array location approximately simultaneously.

22. The superscalar processor according to claim 21, wherein the first address is associated with the first instruction among the instruction group in one clock cycle, and the second address is associated with the second instruction among the instruction group in one clock cycle.

23. The superscalar processor according to claim 21, wherein the execution result of the first instruction stored in the first location of the buffer is stored at the first register array location in one clock cycle, and the execution result of the second instruction stored in the second location of the buffer is stored at the second register array location in one clock cycle.

24. The superscalar processor according to claim 21, wherein the execution result stored in a plurality of the register array locations of the register array indicates an in-order state of a superscalar processor.

25. The superscalar processor according to claim 21, wherein the second instruction is not retired before the execution result of the first instruction is stored in the buffer.