System for store forwarding assigning load and store instructions to groups and reorder queues to keep track of program order

• US 6,349,382 B1
• Filed: 03/05/1999
• Issued: 02/19/2002
• Est. Priority Date: 03/05/1999
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Feiste, Kurt Alan, Ronchetti, Bruce Joseph, Shippy, David James
• Primary Examiner(s): Maung, Zarni
• Assistant Examiner(s): EL HADY, NABIL M
• Application Number:
  US09/263,665
• Time in Patent Office: 1,082 Days
• Field of Search: 712/23, 712/216, 712/220, 712/225
• US Class Current: 712/216
• CPC Class Codes: G06F 9/3824 Operand accessing
• CPC Class Codes: G06F 9/3834 Maintaining memory consistency

Abstract

In a load/store unit within a microprocessor, load and store instructions are executed out of order. The load and store instructions are assigned tags in a predetermined manner, and then assigned to load and store reorder queues for keeping track of the program order of the load and store instructions. When a load instruction is issued for execution, a determination is made whether the load instruction is attempting to load data to a memory location that is the same as a previously executed store instruction is waiting to complete. If so, then the data waiting to be stored within the cache by the store instruction is directly forwarded to the load instruction.

Claims

1. In a processor, a method comprising the steps of:

issuing a load instruction to a load/store unit for execution, wherein the load instruction is assigned to an instruction group;

comparing a memory address location of data to be loaded by the load instruction with memory address locations in a store reorder queue of data to be stored by older and previously executed store instructions;

if there is a match of the memory address location of data to be loaded by the load instruction with one or more of the memory address locations in the store reorder queue of data to be stored by older and previously executed store instructions, then forwarding the data to be stored to a register designated by the load instruction;

wherein if the data is not forwardable to the load instruction, then determining if the load instruction and the store instruction are in the instruction group; and

if the load instruction and the store instruction are in the instruction group, then flushing the load instruction; and

if the load instruction and the store instruction are not in the instruction group, then rejecting the load instruction.

2. The method as recited in claim 1, wherein the forwarding step forwards the data corresponding to a youngest one of the one or more store instructions.

3. The method as recited in claim 1, wherein if there is not a match of the memory address location of data to be loaded by the load instruction with any memory address locations in the store reorder queue of data to be stored by older and previously executed store instructions, then the load instruction is provided with the data from the cache/memory subsystem.

4. The method as recited in claim 1, further comprising the steps of:

before the step of comparing a memory address location of data to be loaded by the load instruction with memory address locations in a store reorder queue of data to be stored by older and previously executed store instructions, comparing the memory address location of the data to be loaded by the load instruction with memory address locations of older store instructions in execute; and

rejecting the load instruction if there is a match of the memory address location of the data to be loaded by the load instruction with any memory address locations of older store instructions in execute.

5. A processor comprising:

a load/store unit;

a store reorder queue resident in the load/store unit;

a dispatch unit for issuing a load instruction to the load/store unit for execution, wherein the load instruction is assigned to an instruct group;

circuitry for comparing a memory address location of data to be loaded by the load instruction with memory address locations in the store reorder queue of data to be stored by older and previously executed store instructions;

circuitry for forwarding the data to be stored to a register designated by the load instruction if there is a match of the memory address location of data to be loaded by the load instruction with one or more of the memory address locations in the store reorder queue;

wherein if the data is not forwardable to the load instruction, circuitry for determining if the load instruction and the store instruction are in the instruction group;

if the load instruction and the store instruction are in the instruction group, circuitry for flushing the load instruction; and

if the load instruction and the store instruction are not in the instruction group, circuitry for rejecting the load instruction.

6. The processor as recited in claim 5, wherein the forwarding circuitry forwards the data corresponding to a youngest one of the one or more store instructions.

7. The processor as recited in claim 5, wherein if there is not a match of the memory address location of data to be loaded by the load instruction with any memory address locations in the store reorder queue of data to be stored by older and previously executed store instructions, then the load instruction is provided with the data from a cache/memory subsystem.

8. The processor as recited in claim 5, further comprising:

before comparing the memory address location of data to be loaded by the load instruction with memory address locations in a store reorder queue of data to be stored by older and previously executed store instructions, circuitry for comparing the memory address location of the data to be loaded by the load instruction with memory address locations of older store instructions in execute; and

circuitry for rejecting the load instruction if there is a match of the memory address location of the data to be loaded by the load instruction with any memory address locations of older store instructions in execute.

9. In a processor, a method for store forwarding comprising the steps of:

issuing a load instruction to a load/store unit for execution, wherein the load instruction is assigned to an instruction group;

comparing an effective memory address location of data to be loaded by the load instruction with effective memory address locations of older store instructions in execute;

rejecting the load instruction if there is a match of the effective address location of the data to be loaded by the load instruction with any effective memory address locations of older store instructions in execute;

comparing a memory address location of data to be loaded by the load instruction with memory address locations in a store reorder queue of data to be stored by older and previously executed store instructions;

if the load instruction is contained within an older and previously executed store instruction, then forwarding the data to be stored to a register designated by the load instruction, wherein the forwarding step forwards the data corresponding to a youngest one of the one or more store instructions, wherein if there is not a match of the memory address location of data to be loaded by the load instruction with any memory address locations in the store reorder queue of data to be stored by older and previously executed store instructions, then the load instruction is provided with the data from the cache/memory subsystem;

wherein if the data is not forwardable to the load instruction, then determining if the load instruction and the store instruction are in the instruction group;

if the load instruction and the store instruction are in the instruction group, then flushing the load instruction; and

if the load instruction and the store instruction are not in the instruction group, then rejecting the load instruction.