Method and apparatus for splitting packets in multithreaded VLIW processor

• US 7,096,343 B1
• Filed: 03/30/2000
• Issued: 08/22/2006
• Est. Priority Date: 03/30/2000
• Status: Expired due to Fees

• Current Assignee: Avago Technologies General IP PTE Limited (Broadcom, Inc.)
• Original Assignee: Agere Systems Incorporated (Broadcom, Inc.)
• Inventors: Jeremiassen, Tor E., Kaxiras, Stefanos, Heintze, Nevin, Berenbaum, Alan David
• Primary Examiner(s): Donaghue, Larry D.
• Application Number:
  US09/538,755
• Time in Patent Office: 2,336 Days
• Field of Search: 712/23, 712/24, 712/215, 712/231, 717/149, 718/102, 718/104, 718/106
• US Class Current: 712/24
• CPC Class Codes: G06F 9/3851 from multiple instruction streams, e.g. multistreaming
• CPC Class Codes: G06F 9/3853 of compound instructions
• CPC Class Codes: G06F 9/3885 using a plurality of independent parallel functional units

Abstract

A method and apparatus are disclosed for allocating functional units in a multithreaded very large instruction word (VLIW) processor. The present invention combines the techniques of conventional very long instruction word architectures and conventional multithreaded architectures to reduce execution time within an individual program, as well as across a workload. The present invention utilizes instruction packet splitting to recover some efficiency lost with conventional multithreaded architectures. Instruction packet splitting allows an instruction bundle to be partially issued in one cycle, with the remainder of the bundle issued during a subsequent cycle. The allocation hardware assigns as many instructions from each packet as will fit on the available functional units, rather than allocating all instructions in an instruction packet at one time. Those instructions that cannot be allocated to a functional unit are retained in a ready-to-run register. On subsequent cycles, instruction packets in which all instructions have been issued to functional units are updated from their thread'"'"'s instruction stream, while instruction packets with instructions that have been held are retained. The functional unit allocation logic can then assign instructions from the newly-loaded instruction packets as well as instructions that were not issued from the retained instruction packets.

Claims

1. A multithreaded very large instruction word processor, comprising:

a plurality of functional units for executing instructions from a multithreaded instruction stream, said instructions being grouped into instruction packets by a compiler; and

an allocator that selects instructions from said instruction stream and forwards said instructions to said plurality of functional units, said allocator assigning instructions from at least one of said instruction packets to a plurality of said functional units, wherein said functional units can be allocated independently to any thread in said multithreaded instruction stream.

2. The multithreaded very large instruction word processor of claim 1, wherein said allocator assigns as many instructions from a given instruction packet as permitted by an availability of said functional units.

3. The multithreaded very large instruction word processor of claim 1, further comprising a register for storing for execution in a later cycle an indication of those instructions from a given instruction packet that cannot be allocated to a functional unit in a given cycle.

4. The multithreaded very large instruction word processor of claim 3, wherein instruction packets in which all instructions have been issued to functional units are updated from the instruction stream of said thread.

5. The multithreaded very large instruction word processor of claim 3, wherein instruction packets with instructions indicated in said register are retained.

6. The multithreaded very large instruction word processor of claim 1, wherein said allocator can split an instruction packet provided the semantics of the instruction packet assembled by the compiler are not violated.

7. The multithreaded very large instruction word processor of claim 1, wherein said allocator can split an instruction packet provided a source register for one of the instructions in a first part of said packet is not modified by one of the instructions in a second part of said packet.

8. A method for processing instructions in a multithreaded very large instruction word processor, comprising the steps of:

executing said instructions using a plurality of functional units, wherein said instructions are grouped into instruction packets by a compiler;

assigning instructions from said instruction stream to said plurality of functional units, wherein instructions from at least one of said instruction packets are assigned to a plurality of said functional units, wherein said functional units can be allocated independently to any thread in said multithreaded instruction stream; and

forwarding said selected instructions to said corresponding functional units.

9. The method of claim 8, wherein said assigning step assigns as many instructions from a given instruction packet as permitted by an availability of said functional units.

10. The method of claim 8, further comprising the step of storing for execution in a later cycle an indication of those instructions from a given instruction packet that cannot be allocated to a functional unit in a given cycle.

11. The method of claim 10, wherein instruction packets in which all instructions have been issued to functional units are updated from the instruction stream of said thread.

12. The method of claim 10, wherein instruction packets with instructions indicated in said register are retained.

13. The method of claim 8, wherein said assigning step can split an instruction packet provided the semantics of the instruction packet assembled by the compiler are not violated.

14. The method of claim 8, wherein said assigning step can split an instruction packet provided a source register for one of the instructions in a first part of said packet is not modified by one of the instructions in a second part of said packet.

15. An article of manufacture for processing instructions from an instruction stream having a plurality of threads in a multithreaded very large instruction word processor, comprising:

a computer readable medium having computer readable program code means embodied thereon, said computer readable program code means comprising program code means for causing a computer to;

execute said instructions using a plurality of functional units, wherein said instructions are grouped into instruction packets by a compiler;

assign instructions from said instruction stream to said plurality of functional units, wherein instructions from at least one of said instruction packets are assigned to a plurality of said functional units wherein said functional units can be allocated independently to any thread in said multithreaded instruction stream; and

forward said selected instructions to said corresponding functional units.

16. A multithreaded very large instruction word processor, comprising:

a plurality of functional units for executing instructions from a multithreaded instruction stream, said instructions being grouped into instruction packets by a compiler; and

an allocator that selects instructions from said instruction stream and forwards said instructions to said plurality of functional units, said allocator assigning instructions from at least one of said instruction packets to a plurality of said functional units, wherein said functional units can be allocated independently to any thread in said multithreaded instruction stream, provided the semantics of said instruction packet assembled by said compiler are not violated.