MANAGING INTERNODE DATA COMMUNICATIONS FOR AN UNINITIALIZED PROCESS IN A PARALLEL COMPUTER

• US 20130124666A1
• Filed: 12/10/2012
• Published: 05/16/2013
• Est. Priority Date: 11/09/2011
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: ARCHER, CHARLES J., BLOCKSOME, MICHAEL A., MILLER, DOUGLAS R., PARKER, JEFFREY J., RATTERMAN, JOSEPH D., SMITH, BRIAN E.
• Granted Patent: US 8,732,726 B2
• Time in Patent Office: Days
• US Class Current: 709/212
• CPC Class Codes: G06F 15/167 using a common memory, e.g. mailbox
• CPC Class Codes: G06F 15/17318 Parallel communications techniques, e.g. gather, scatter, reduce, roadcast, multicast, all to all
• CPC Class Codes: G06F 9/544 Buffers; Shared memory; Pipes

Abstract

A parallel computer includes nodes, each having main memory and a messaging unit (MU). Each MU includes computer memory, which in turn includes, MU message buffers. Each MU message buffer is associated with an uninitialized process on the compute node. In the parallel computer, managing internode data communications for an uninitialized process includes: receiving, by an MU of a compute node, one or more data communications messages in an MU message buffer associated with an uninitialized process on the compute node; determining, by an application agent, that the MU message buffer associated with the uninitialized process is full prior to initialization of the uninitialized process; establishing, by the application agent, a temporary message buffer for the uninitialized process in main computer memory; and moving, by the application agent, data communications messages from the MU message buffer associated with the uninitialized process to the temporary message buffer in main computer memory.

Claims

1. A method of managing internode data communications for an uninitialized process in a parallel computer, the parallel computer comprising a plurality of compute nodes, each compute node comprising main computer memory and a messaging unit (MU), each MU comprising a module of automated computing machinery coupling compute nodes for data communications, each MU comprising computer memory, the computer memory of the MU comprising one or more MU message buffers, each MU message buffer associated with an uninitialized process on the compute node, the method comprising:

receiving, by an MU of a compute node, one or more data communications messages in an MU message buffer associated with an uninitialized process on the compute node;

determining, by an application agent, that the MU message buffer associated with the uninitialized process is full prior to initialization of the uninitialized process;

establishing, by the application agent, a temporary message buffer for the uninitialized process in main computer memory; and

moving, by the application agent, data communications messages from the MU message buffer associated with the uninitialized process to the temporary message buffer in main computer memory.

2. The method of claim 1 further comprising:

upon initialization of the uninitialized process, providing, by the application agent to the initialized process, a pointer to the temporary message buffer; and

consuming, by the initialized process in dependence upon the pointer, data communications messages from the temporary message buffer prior to consuming data communications messages stored in the MU message buffer.

3. The method of claim 1 wherein determining, by an application agent, that the MU message buffer associated with the uninitialized process is full further comprises:

periodically polling the messaging unit for information describing the size of all data communications messages presently stored in the MU message buffer; and

determining whether the size of all data communications messages presently stored in the MU message buffer exceeds a predetermined threshold size.

4. The method of claim 1 wherein determining, by an application agent, that the MU message buffer associated with the uninitialized process is full further comprises receiving a notification from the messaging unit that MU message buffer is full.

5. The method of claim 1 wherein moving, by the application agent, data communications messages from the MU message buffer associated with the uninitialized process to the temporary message buffer in main computer memory further comprises:

periodically moving, by the application agent, one or more messages from the MU message buffer to the temporary message buffer prior to initialization of the uninitialized process.

6. The method of claim 1 wherein each compute node comprises a multi-core computer processor and one core executes only the application agent.

7. The method of claim 1 wherein receiving, by an MU of a compute node, one or more data communications messages in an MU message buffer associated with an uninitialized process on the compute node, further comprises:

allocating, at boot time of the compute node by the MU, one or more MU message buffers in the MU'"'"'s computer memory;

receiving, by the MU from a process on another compute node, the one or more data communications message intended for the uninitialized process; and

storing the data communications message in the MU message buffer associated with the uninitialized process.

8. The method of claim 1 wherein:

the parallel computer comprises a parallel active messaging interface (‘

PAMI’

) and the plurality of compute nodes execute a parallel application, the PAMI comprises data communications endpoints, each endpoint comprising a specification of data communications parameters for a thread of execution on a compute node, including specifications of a client, a context, and a task, the endpoints coupled for data communications through the PAMI; and

each process comprises an endpoint.

9. The method of claim 8 wherein:

each client comprises a collection of data communications resources dedicated to the exclusive use of an application-level data processing entity;

each context comprises a subset of the collection of data processing resources of a client, context functions, and a work queue of data transfer instructions to be performed by use of the subset through the context functions operated by an assigned thread of execution; and

each task represents a process of execution of the parallel application.

10. The method of claim 8 wherein each context carries out, through post and advance functions, data communications for the parallel application on data communications resources in the exclusive possession of that context.

11. The method of claim 8 wherein each context carries out data communications operations independently and in parallel with other contexts.

12-20. -20. (canceled)