Executing Multiple Instructions Multiple Data ('MIMD') Programs on a Single Instruction Multiple Data ('SIMD') Machine

• US 20080288746A1
• Filed: 05/16/2007
• Published: 11/20/2008
• Est. Priority Date: 05/16/2007
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Inglett, Todd A., Peters, Amanda, McCarthy, Patrick J.
• Granted Patent: US 7,979,674 B2
• Time in Patent Office: Days
• US Class Current: 712/20
• CPC Class Codes: G06F 9/5061 Partitioning or combining of resources

Abstract

Executing MIMD programs on a SIMD machine, the SIMD machine including a plurality of compute nodes, each compute node capable of executing only a single thread of execution, the compute nodes initially configured exclusively for SIMD operations, the SIMD machine further comprising a data communications network, the network comprising synchronous data communications links among the compute nodes, including establishing one or more SIMD partitions, booting one or more SIMD partitions in MIMD mode; establishing a MIMD partition; executing by launcher programs a plurality of MIMD programs on two or more of the compute nodes of the MIMD partition; and re-executing a launcher program by an operating system on a compute node in the MIMD partition upon termination of the MIMD program executed by the launcher program.

Claims

1. A method of executing Multiple Instructions Multiple Data (‘

MIMD’

) programs on a Single Instruction Multiple Data (‘

SIMD’

) machine, the SIMD machine comprising a plurality of compute nodes, each compute node capable of executing only a single thread of execution, the compute nodes initially configured exclusively for SIMD operations, the SIMD machine further comprising a data communications network, the network comprising synchronous data communications links among the compute nodes, the method comprising;

establishing one or more SIMD partitions, each SIMD partition comprising a plurality of the compute nodes, the compute nodes in each SIMD partition electronically isolated from compute nodes in other partitions of the SIMD machine and coupled to one another through links of the network for synchronous data communications for parallel SIMD operations among the compute nodes in each SIMD partition;

booting one or more SIMD partitions in MIMD mode;

establishing a MIMD partition, the MIMD partition comprising a multiplicity of compute nodes from one or more SIMD partitions booted in MIMD mode, including assigning each compute node in the MIMD partition to receive MIMD programs from a same dispatcher;

executing by launcher programs a plurality of MIMD programs on two or more of the compute nodes of the MIMD partition, including replacing the launcher programs with the MIMD programs in process address space in computer memory of the two or more compute nodes, each MIMD program autonomously executing different instructions on different data; and

re-executing a launcher program by an operating system on a compute node in the MIMD partition upon termination of the MIMD program executed by the launcher program.

2. The method of claim 1 wherein booting one or more SIMD partitions in MIMD mode further comprises:

setting, in operating systems on the compute nodes of each SIMD partition booted in MIMD mode, flags indicating MIMD operation;

loading onto the compute nodes of each SIMD partition booted in MIMD mode a launcher program;

initializing with link training synchronous data communications among links of the network among compute nodes in each SIMD partition booted in MIMD mode;

initializing, with a parallel processing barrier, parallel operations among the compute nodes of each SIMD partition booted in MIMD mode, and

executing a launcher program on each compute node in each SIMD partition booted in MIMD mode;

3. The method of claim 1 wherein assigning each compute node in the MIMD partition to receive MIMD programs from a same dispatcher further comprises providing to each compute node in the MIMD partition the network address of the same dispatcher.

4. The method of claim 1 further comprising:

tracking performance parameters comprising a performance profile for the MIMD programs executing on the compute nodes of the MIMD partition; and

dynamically varying the number of compute nodes in the MIMD partition in dependence upon values of performance parameters in the performance profile.

5. The method of claim 4 wherein tracking performance parameters further comprises tracking total run time for each instance of the application, memory utilization, memory faults, cache misses, and total run time for all instances of the application.

6. The method of claim 4 wherein dynamically varying the number of compute nodes in the MIMD partition further comprises adding to or removing from the MIMD partition all the compute nodes of a SIMD partition booted in MIMD mode.

7. Apparatus for executing Multiple Instructions Multiple Data (‘

MIMD’

) programs on a Single Instruction Multiple Data (‘

SIMD’

) machine, the SIMD machine comprising a plurality of compute nodes, each compute node capable of executing only a single thread of execution, the compute nodes initially configured exclusively for SIMD operations, the SIMD machine further comprising a data communications network, the network comprising synchronous data communications links among the compute nodes, the apparatus comprising a computer processor, a computer memory operatively coupled to the computer processor, the computer memory having disposed within it computer program instructions capable of;

establishing one or more SIMD partitions, each SIMD partition comprising a plurality of the compute nodes, the compute nodes in each SIMD partition electronically isolated from compute nodes in other partitions of the SIMD machine and coupled to one another through links of the network for synchronous data communications for parallel SIMD operations among the compute nodes in each SIMD partition;

booting one or more SIMD partitions in MIMD mode;

establishing a MIMD partition, the MIMD partition comprising a multiplicity of compute nodes from one or more SIMD partitions booted in MIMD mode, including assigning each compute node in the MIMD partition to receive MIMD programs from a same dispatcher;

executing by launcher programs a plurality of MIMD programs on two or more of the compute nodes of the MIMD partition, including replacing the launcher programs with the MIMD programs in process address space in computer memory of the two or more compute nodes, each MIMD program autonomously executing different instructions on different data; and

re-executing a launcher program by an operating system on a compute node in the MIMD partition upon termination of the MIMD program executed by the launcher program.

8. The apparatus of claim 7 wherein booting one or more SIMD partitions in MIMD mode further comprises:

setting, in operating systems on the compute nodes of each SIMD partition booted in MIMD mode, flags indicating MIMD operation;

loading onto the compute nodes of each SIMD partition booted in MIMD mode a launcher program;

initializing with link training synchronous data communications among links of the network among compute nodes in each SIMD partition booted in MIMD mode;

initializing, with a parallel processing barrier, parallel operations among the compute nodes of each SIMD partition booted in MIMD mode, and

executing a launcher program on each compute node in each SIMD partition booted in MIMD mode;

9. The apparatus of claim 7 wherein assigning each compute node in the MIMD partition to receive MIMD programs from a same dispatcher further comprises providing to each compute node in the MIMD partition the network address of the same dispatcher.

10. The apparatus of claim 7 further comprising computer program instructions capable of:

tracking performance parameters comprising a performance profile for the MIMD programs executing on the compute nodes of the MIMD partition; and

dynamically varying the number of compute nodes in the MIMD partition in dependence upon values of performance parameters in the performance profile.

11. The apparatus of claim 10 wherein tracking performance parameters further comprises tracking total run time for each instance of the application, memory utilization, memory faults, cache misses, and total run time for all instances of the application.

12. The apparatus of claim 10 wherein dynamically varying the number of compute nodes in the MIMD partition further comprises adding to or removing from the MIMD partition all the compute nodes of a SIMD partition booted in MIND mode

13. A computer program product for executing Multiple Instructions Multiple Data (‘

MIMD’

) programs on a Single Instruction Multiple Data (‘

SIMD’

) machine, the SIMD machine comprising a plurality of compute nodes, each compute node capable of executing only a single thread of execution, the compute nodes initially configured exclusively for SIMD operations, the SIMD machine further comprising a data communications network, the network comprising synchronous data communications links among the compute nodes, the computer program product disposed in a computer readable, signal bearing medium, the computer program product comprising computer program instructions capable of;

establishing one or more SIMD partitions, each SIMD partition comprising a plurality of the compute nodes, the compute nodes in each SIMD partition electronically isolated from compute nodes in other partitions of the SIMD machine and coupled to one another through links of the network for synchronous data communications for parallel SIMD operations among the compute nodes in each SIMD partition;

booting one or more SIMD partitions in MIMD mode;

establishing a MIMD partition, the MIMD partition comprising a multiplicity of compute nodes from one or more SIMD partitions booted in MIMD mode, including assigning each compute node in the MIMD partition to receive MIMD programs from a same dispatcher;

executing by launcher programs a plurality of MIMD programs on two or more of the compute nodes of the MIMD partition, including replacing the launcher programs with the MIMD programs in process address space in computer memory of the two or more compute nodes, each MIMD program autonomously executing different instructions on different data; and

re-executing a launcher program by an operating system on a compute node in the MIMD partition upon termination of the MIMD program executed by the launcher program.

14. The computer program product of claim 13 wherein the signal bearing medium comprises a recordable medium.

15. The computer program product of claim 13 wherein the signal bearing medium comprises a transmission medium.

16. The computer program product of claim 13 wherein booting one or more SIMD partitions in MIMD mode further comprises:

setting, in operating systems on the compute nodes of each SIMD partition booted in MIMD mode, flags indicating MIMD operation;

loading onto the compute nodes of each SIMD partition booted in MIMD mode a launcher program;

initializing with link training synchronous data communications among links of the network among compute nodes in each SIMD partition booted in MIMD mode;

initializing, with a parallel processing barrier, parallel operations among the compute nodes of each SIMD partition booted in MIMD mode, and

executing a launcher program on each compute node in each SIMD partition booted in MIMD mode;

17. The computer program product of claim 13 wherein assigning each compute node in the MIMD partition to receive MIMD programs from a same dispatcher further comprises providing to each compute node in the MIMD partition the network address of the same dispatcher.

18. The computer program product of claim 13 further comprising computer program instructions capable of:

tracking performance parameters comprising a performance profile for the MIMD programs executing on the compute nodes of the MIMD partition; and

dynamically varying the number of compute nodes in the MIMD partition in dependence upon values of performance parameters in the performance profile.

19. The computer program product of claim 18 wherein tracking performance parameters further comprises tracking total run time for each instance of the application, memory utilization, memory faults, cache misses, and total run time for all instances of the application.

20. The computer program product of claim 18 wherein dynamically varying the number of compute nodes in the MIMD partition further comprises adding to or removing from the MIMD partition all the compute nodes of a SIMD partition booted in MIMD mode.