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

• US 20090024830A1
• Filed: 07/19/2007
• Published: 01/22/2009
• Est. Priority Date: 07/19/2007
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Peters, Amanda, Budnik, Thomas A., McCarthy, Patrick J., Sexton, James C., Stewart, Gordon G., Mundy, Michael B., King, Alan J.
• Granted Patent: US 7,831,802 B2
• Time in Patent Office: Days
• US Class Current: 712/20
• CPC Class Codes: G06F 15/161 Computing infrastructure, e.g. computer clusters, blade chassis or hardware partitioning casings, cabinets, racks or drawers for data centers H05K5/00

Abstract

Executing Multiple Instructions Multiple Data (‘MIMD’) programs on a Single Instruction Multiple Data (‘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 a SIMD partition comprising a plurality of the compute nodes; booting the SIMD partition in MIMD mode; executing by launcher programs a plurality of MIMD programs on compute nodes in the SIMD partition; and re-executing a launcher program by an operating system on a compute node in the SIMD 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 a SIMD partition comprising a plurality of the compute nodes, the compute nodes in the 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 the SIMD partition;

booting the SIMD partition in MIMD mode, including;

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

loading onto the compute nodes of the SIMD partition a launcher program;

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

initializing, with a parallel processing barrier, parallel operations among the compute nodes of the SIMD partition, and executing a launcher program on each compute node in the SIMD partition;

executing by launcher programs a plurality of MIMD programs on two or more of the compute nodes in the SIMD partition, including replacing the launcher programs with the MIMD programs in process address space in computer memory of the 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 SIMD partition upon termination of the MIMD program executed by the launcher program.

2. The method of claim 1 further comprising communicating, by a dispatcher to a launcher program on a compute node in the SIMD partition booted in MIMD mode, the name of a MIMD program to execute by the launcher.

3. The method of claim 1 wherein loading a launcher program further comprises loading a launcher program by a control application on a service node through a service network through input/output nodes through a collective network into compute nodes of the SIMD partition booted in MIMD mode.

4. The method of claim 1 further comprising establishing on the SIMD machine a SIMD partition booted in SIMD mode, with the SIMD partition booted in SIMD mode running simultaneously with the SIMD partition booted in MIMD mode.

5. The method of claim 1 further comprising:

detecting by a compute node a processing error;

communicating by the compute node through a service network to a control application on a service node the existence of the processing error; and

booting, by the control application without affecting operations of other compute nodes, the compute node in MIMD mode, the booting including;

setting, in an operating system on the compute node, a flag indicating MIMD operation;

setting, in an operating system on the compute node, a flag indicating that the current boot of the compute node is a reboot;

loading and executing a launcher program on the compute node;

omitting the initializing of synchronous data communications on the network with link training and the initializing of parallel operations among the compute nodes with a parallel processing barrier.

17. The method of claim 1 further comprising communicating by a dispatcher to a launcher program on a compute node the name of a MIMD program to execute by the launcher.

18. The method of claim 1 wherein loading a launcher program further comprises loading a launcher program by a control application on a service node through a service network through i/o nodes through a collective network into compute nodes.

19. The method of claim 1 further comprising establishing on the SIMD machine a SIMD partition booted in SIMD mode and a SIMD partition booted in MIMD mode, with the SIMD partition booted in SIMD mode running simultaneously with the SIMD partition booted in MIMD mode.

20. The method of operating claim 1 further comprising:

detecting by a compute node a processing error;

communicating by the compute node through a service (JTAG) network to a service application on a service node the existence of the processing error; and

booting the compute node in MIMD mode, including setting in an operating system on the compute node a flag indicating MIMD operation, further including setting in an operating system on the compute node a flag indicating that the current boot of the compute node is a reboot, further including loading a launcher program on the compute node, omitting the initializing of synchronous data communications on the network with link training, and omitting the initializing of parallel operations among the compute nodes with a parallel processing barrier, the booting implemented without affecting operations of other compute nodes.

6. 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 a SIMD partition comprising a plurality of the compute nodes, the compute nodes in the 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 the SIMD partition;

booting the SIMD partition in MIMD mode, including;

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

loading onto the compute nodes of the SIMD partition a launcher program;

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

initializing, with a parallel processing barrier, parallel operations among the compute nodes of the SIMD partition, and executing a launcher program on each compute node in the SIMD partition;

executing by launcher programs a plurality of MIMD programs on two or more of the compute nodes in the SIMD partition, including replacing the launcher programs with the MIMD programs in process address space in computer memory of the 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 SIMD partition upon termination of the MIMD program executed by the launcher program.

7. The apparatus of claim 6 further comprising computer program instructions capable of communicating, by a dispatcher to a launcher program on a compute node in the SIMD partition booted in MIMD mode, the name of a MIMD program to execute by the launcher.

8. The apparatus of claim 6 wherein loading a launcher program further comprises loading a launcher program by a control application on a service node through a service network through input/output nodes through a collective network into compute nodes of the SIMD partition booted in MIMD mode.

9. The apparatus of claim 6 further comprising computer program instructions capable of establishing on the SIMD machine a SIMD partition booted in SIMD mode, with the SIMD partition booted in SIMD mode running simultaneously with the SIMD partition booted in MIMD mode.

10. The apparatus of claim 6 further comprising computer program instructions capable of:

detecting by a compute node a processing error;

communicating by the compute node through a service network to a control application on a service node the existence of the processing error; and

booting, by the control application without affecting operations of other compute nodes, the compute node in MIMD mode, the booting including;

setting, in an operating system on the compute node, a flag indicating MIMD operation;

setting, in an operating system on the compute node, a flag indicating that the current boot of the compute node is a reboot;

loading and executing a launcher program on the compute node;

omitting the initializing of synchronous data communications on the network with link training and the initializing of parallel operations among the compute nodes with a parallel processing barrier.

11. 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 a SIMD partition comprising a plurality of the compute nodes, the compute nodes in the 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 the SIMD partition;

booting the SIMD partition in MIMD mode, including;

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

loading onto the compute nodes of the SIMD partition a launcher program;

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

initializing, with a parallel processing barrier, parallel operations among the compute nodes of the SIMD partition, and executing a launcher program on each compute node in the SIMD partition;

executing by launcher programs a plurality of MIMD programs on two or more of the compute nodes in the SIMD partition, including replacing the launcher programs with the MIMD programs in process address space in computer memory of the 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 SIMD partition upon termination of the MIMD program executed by the launcher program.

12. The computer program product of claim 11 further comprising computer program instructions capable of communicating, by a dispatcher to a launcher program on a compute node in the SIMD partition booted in MIMD mode, the name of a MIMD program to execute by the launcher.

13. The computer program product of claim 11 wherein loading a launcher program further comprises loading a launcher program by a control application on a service node through a service network through input/output nodes through a collective network into compute nodes of the SIMD partition booted in MIMD mode.

14. The computer program product of claim 11 further comprising computer program instructions capable of establishing on the SIMD machine a SIMD partition booted in SIMD mode, with the SIMD partition booted in SIMD mode running simultaneously with the SIMD partition booted in MIMD mode.

15. The computer program product of claim 11 further comprising computer program instructions capable of:

detecting by a compute node a processing error;

communicating by the compute node through a service network to a control application on a service node the existence of the processing error; and

booting, by the control application without affecting operations of other compute nodes, the compute node in MIMD mode, the booting including;

setting, in an operating system on the compute node, a flag indicating MIMD operation;

setting, in an operating system on the compute node, a flag indicating that the current boot of the compute node is a reboot;

loading and executing a launcher program on the compute node;

omitting the initializing of synchronous data communications on the network with link training and the initializing of parallel operations among the compute nodes with a parallel processing barrier.

16. 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;

booting the SIMD machine in MIMD mode, including;

setting, in operating systems on the compute nodes, flags indicating MIMD operation;

loading onto the compute nodes a launcher program;

initializing, with link training, synchronous data communications among links of the network among compute nodes in the SIMD partition;

initializing, with a parallel processing barrier, parallel operations among the compute nodes; and

executing a launcher program on each compute node;

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

re-executing a launcher program by an operating system upon termination of the MIMD program executed by the launcher program.

21. 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;

booting the SIMD machine in MIMD mode, including;

setting, in operating systems on the compute nodes, flags indicating MIMD operation;

loading onto the compute nodes a launcher program;

initializing, with link training, synchronous data communications among links of the network among compute nodes in the SIMD partition;

initializing, with a parallel processing barrier, parallel operations among the compute nodes; and

executing a launcher program on each compute node;

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

re-executing a launcher program by an operating system upon termination of the MIMD program executed by the launcher program.

22. The apparatus of claim 21 further comprising computer program instructions capable of communicating by a dispatcher to a launcher program on a compute node the name of a MIMD program to execute by the launcher.

23. The apparatus of claim 21 wherein loading a launcher program further comprises loading a launcher program by a control application on a service node through a service network through i/o nodes through a collective network into compute nodes.

24. The apparatus of claim 21 further comprising computer program instructions capable of establishing on the SIMD machine a SIMD partition booted in SIMD mode and a SIMD partition booted in MIMD mode, with the SIMD partition booted in SIMD mode running simultaneously with the SIMD partition booted in MIMD mode.

25. The apparatus of claim 21 further comprising computer program instructions capable of:

detecting by a compute node a processing error;

communicating by the compute node through a service (JTAG) network to a service application on a service node the existence of the processing error; and

booting the compute node in MIMD mode, including setting in an operating system on the compute node a flag indicating MIMD operation, further including setting in an operating system on the compute node a flag indicating that the current boot of the compute node is a reboot, further including loading a launcher program on the compute node, omitting the initializing of synchronous data communications on the network with link training, and omitting the initializing of parallel operations among the compute nodes with a parallel processing barrier, the booting implemented without affecting operations of other compute nodes.

26. 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;

booting the SIMD machine in MIMD mode, including;

setting, in operating systems on the compute nodes, flags indicating MIMD operation;

loading onto the compute nodes a launcher program;

initializing, with link training, synchronous data communications among links of the network among compute nodes in the SIMD partition;

initializing, with a parallel processing barrier, parallel operations among the compute nodes; and

executing a launcher program on each compute node;

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

re-executing a launcher program by an operating system upon termination of the MIMD program executed by the launcher program.

27. The computer program product of claim 26 further comprising computer program instructions capable of communicating by a dispatcher to a launcher program on a compute node the name of a MIMD program to execute by the launcher.

28. The computer program product of claim 26 wherein loading a launcher program further comprises loading a launcher program by a control application on a service node through a service network through i/o nodes through a collective network into compute nodes.

29. The computer program product of claim 26 further comprising computer program instructions capable of establishing on the SIMD machine a SIMD partition booted in SIMD mode and a SIMD partition booted in MIMD mode, with the SIMD partition booted in SIMD mode running simultaneously with the SIMD partition booted in MIMD mode.

30. The computer program product of claim 26 further comprising computer program instructions capable of:

detecting by a compute node a processing error;

communicating by the compute node through a service (JTAG) network to a service application on a service node the existence of the processing error; and

booting the compute node in MIMD mode, including setting in an operating system on the compute node a flag indicating MIMD operation, further including setting in an operating system on the compute node a flag indicating that the current boot of the compute node is a reboot, further including loading a launcher program on the compute node, omitting the initializing of synchronous data communications on the network with link training, and omitting the initializing of parallel operations among the compute nodes with a parallel processing barrier, the booting implemented without affecting operations of other compute nodes.