Identifying Messaging Completion on a Parallel Computer

• US 20080195840A1
• Filed: 02/09/2007
• Published: 08/14/2008
• Est. Priority Date: 02/09/2007
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Wallenfelt, Brian P., Archer, Charles J., McCarthy, Patrick J., Hardwick, Camesha R.
• Granted Patent: US 7,552,312 B2
• Time in Patent Office: Days
• US Class Current: 712/1
• CPC Class Codes: G06F 15/17337 Direct connection machines, e.g. completely connected computers, point to point communication networks coupling between buses G06F13/4004

Abstract

Methods, parallel computers, and products are provided for identifying messaging completion on a parallel computer. The parallel computer includes a plurality of compute nodes, the compute nodes coupled for data communications by at least two independent data communications networks including a binary tree data communications network optimal for collective operations that organizes the nodes as a tree and a torus data communications network optimal for point to point operations that organizes the nodes as a torus. Embodiments include reading all counters at each node of the torus data communications network; calculating at each node a current node value in dependence upon the values read from the counters at each node; and determining for all nodes whether the current node value for each node is the same as a previously calculated node value for each node. If the current node is the same as the previously calculated node value for all nodes of the torus data communications network, embodiments include determining that messaging is complete and if the current node is not the same as the previously calculated node value for all nodes of the torus data communications network, embodiments include determining that messaging is currently incomplete.

Claims

1. A computer-implemented method for identifying messaging completion on a parallel computer, the parallel computer comprising a plurality of compute nodes, the compute nodes coupled for data communications by at least two independent data communications networks including a binary tree data communications network optimal for collective operations that organizes the nodes as a tree and a torus data communications network optimal for point to point operations that organizes the nodes as a torus, the method comprising:

reading all counters at each node of the torus data communications network;

calculating at each node a current node value in dependence upon the values read from the counters at each node; and

determining for all nodes whether the current node value for each node is the same as at least one previously calculated node value for each node; and

if the current node value is the same as the previously calculated node value for all nodes of the torus data communications network, determining that messaging is complete; and

if the current node value is not the same as the previously calculated node value for all nodes of the torus data communications network, determining that messaging is currently incomplete.

2. The method of claim 1 further comprising determining that all nodes have sent all queued messages prior to reading all counters at each node of the torus data communications network;

3. The method of claim 1 wherein calculating at each node a current node value in dependence upon the values read from the counters at each node further comprising summing at each node all values read from the counters for the node.

4. The method of claim 1 wherein determining for all nodes whether the current node value for each node is the same as at least one previously calculated node value for each node further comprises performing an allreduce operation on the binary tree data communications network of the difference between the current node value for each node and the previously calculated node value for each node.

5. The method of claim 1 wherein determining for all nodes whether the current node value for each node is the same as at least one previously calculated node value for each node further comprises determining for all nodes whether the current node value for each node is the same as a plurality of previously calculated node value for each node.

6. The method of claim 1 further comprising if the current node is not the same as the previously calculated node value for all nodes of the torus data communications network:

re-reading all counters at each node of the torus data communications network;

calculating at each node a new current node value in dependence upon the values read from the counters at each node; and

determining for all nodes whether the new current node value for each node is the same as at least one previously calculated node value for each node.

7. A parallel computer for identifying messaging completion on a parallel computer, the parallel computer comprising a plurality of compute nodes, the compute nodes coupled for data communications by at least two independent data communications networks including a binary tree data communications network optimal for collective operations that organizes the nodes as a tree and a torus data communications network optimal for point to point operations that organizes the nodes as a torus, the parallel computer further 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:

reading all counters at each node of the torus data communications network;

calculating at each node a current node value in dependence upon the values read from the counters at each node; and

determining for all nodes whether the current node value for each node is the same as at least one previously calculated node value for each node; and

if the current node value is the same as the previously calculated node value for all nodes of the torus data communications network, determining that messaging is complete; and

if the current node value is not the same as the previously calculated node value for all nodes of the torus data communications network, determining that messaging is currently incomplete.

8. The parallel computer of claim 7 wherein the computer memory also has disposed within it computer program instructions capable of determining that all nodes have sent all queued messages prior to reading all counters at each node of the torus data communications network;

9. The parallel computer of claim 7 wherein computer program instructions capable of calculating at each node a current node value in dependence upon the values read from the counters at each node further comprises computer program instructions capable of summing at each node all values read from the counters for the node.

10. The parallel computer of claim 7 wherein computer program instructions capable of determining for all nodes whether the current node value for each node is the same as at least one previously calculated node value for each node further comprise computer program instructions capable of performing an allreduce operation on the binary tree data communications network of the difference between the current node value for each node and the previously calculated node value for each node.

11. The parallel computer of claim 7 wherein computer program instructions capable of determining for all nodes whether the current node value for each node is the same as at least one previously calculated node value for each node further comprise computer program instructions capable of determining for all nodes whether the current node value for each node is the same as a plurality of previously calculated node value for each node.

12. The parallel computer of claim 7 wherein the computer memory also has disposed within it computer program instructions capable of:

re-reading all counters at each node of the torus data communications network if the current node is not the same as the previously calculated node value for all nodes of the torus data communications network;

calculating at each node a new current node value in dependence upon the values read from the counters at each node; and

determining for all nodes whether the new current node value for each node is the same as at least one previously calculated node value for each node.

13. A computer program product for identifying messaging completion on a parallel computer, the parallel computer comprising a plurality of compute nodes, the compute nodes coupled for data communications by at least two independent data communications networks including a binary tree data communications network optimal for collective operations that organizes the nodes as a tree and a torus data communications network optimal for point to point operations that organizes the nodes as a torus, the computer program product disposed upon a computer readable medium, the computer program product comprising:

computer program instructions for reading all counters at each node of the torus data communications network;

computer program instructions for calculating at each node a current node value in dependence upon the values read from the counters at each node; and

computer program instructions for determining for all nodes whether the current node value for each node is the same as at least one previously calculated node value for each node; and

computer program instructions for determining that messaging is complete if the current node value is the same as the previously calculated node value for all nodes of the torus data communications network; and

computer program instructions for determining that messaging is currently incomplete if the current node value is not the same as the previously calculated node value for all nodes of the torus data communications network.

14. The computer program product of claim 13 further comprising computer program instructions for determining that all nodes have sent all queued messages prior to reading all counters at each node of the torus data communications network;

15. The computer program product of claim 13 wherein computer program instructions for calculating at each node a current node value in dependence upon the values read from the counters at each node further comprises computer program instructions for summing at each node all values read from the counters for the node.

16. The computer program product of claim 13 wherein computer program instructions for determining for all nodes whether the current node value for each node is the same as at least one previously calculated node value for each node further comprise computer program instructions for performing an allreduce operation on the binary tree data communications network of the difference between the current node value for each node and the previously calculated node value for each node.

17. The computer program product of claim 13 wherein computer program instructions for determining for all nodes whether the current node value for each node is the same as at least one previously calculated node value for each node further comprise computer program instructions for determining for all nodes whether the current node value for each node is the same as a plurality of previously calculated node value for each node.

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

computer program instructions for re-reading all counters at each node of the torus data communications network if the current node is not the same as the previously calculated node value for all nodes of the torus data communications network;

computer program instructions for calculating at each node a new current node value in dependence upon the values read from the counters at each node; and

computer program instructions for determining for all nodes whether the new current node value for each node is the same as at least one previously calculated node value for each node.

19. The computer program product of claim 13 wherein the computer readable medium comprises a recordable medium.

20. The computer program product of claim 13 wherein the computer readable medium comprises a transmission medium.