Broadcasting Data In A Hybrid Computing Environment

• US 20100191822A1
• Filed: 01/29/2009
• Published: 07/29/2010
• Est. Priority Date: 01/29/2009
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Sanders, Philip J, Carey, James E., Archer, Charles J., Markland, Matthew W., Schimke, Timothy J.
• Granted Patent: US 8,001,206 B2
• Time in Patent Office: Days
• US Class Current: 709/213
• CPC Class Codes: H04L 12/4641 Virtual LANs, VLANs, e.g. virtual private networks [VPN] LAN interconnection over a bridge based backbone H04L12/462; encapsulation techniques H04L12/4633; routing of packets H04L45/00; packet switches H04L49/00; virtual private networks for security H04L63/0272

Abstract

Methods, apparatus, and products for broadcasting data in a hybrid computing environment that includes a host computer, a number of accelerators, the host computer and the accelerators adapted to one another for data communications by a system level message passing module, the host computer having local memory shared remotely with the accelerators, the accelerators having local memory for the accelerators shared remotely with the host computer, where broadcasting data according to embodiments of the present invention includes: writing, by the host computer remotely to the shared local memory for the accelerators, the data to be broadcast; reading, by each of the accelerators from the shared local memory for the accelerators, the data; and notifying the host computer, by the accelerators, that the accelerators have read the data.

Claims

1. A method of broadcasting data in a hybrid computing environment, the hybrid computing environment comprising a host computer having a host computer architecture, a plurality of accelerators having an accelerator architecture, the accelerator architecture optimized, with respect to the host computer architecture, for speed of execution of a particular class of computing functions, the host computer and the accelerators adapted to one another for data communications by a system level message passing module, the host computer having local memory shared remotely with the accelerators, the accelerators having local memory for the plurality of accelerators shared remotely with the host computer, the method comprising:

writing, by the host computer remotely to the shared local memory for the plurality of accelerators, the data to be broadcast;

reading, by each of the accelerators from the shared local memory for the plurality of accelerators, the data; and

notifying the host computer, by the plurality of accelerators, that the plurality of accelerators have read the data.

2. The method of claim 1 wherein:

notifying the host computer that the plurality of accelerators have read the data further comprises incrementing remotely, by each accelerator upon reading the data, a counter residing in the shared local memory of the host computer; and

the method further comprises determining, by the host computer, that the value of the counter exceeds a predetermined threshold.

3. The method of claim 2 wherein incrementing remotely a counter residing in the shared local memory of the host computer further comprises:

locking by each accelerator the counter;

reading the value of the counter;

incrementing the value of the counter; and

writing the incremented value to the shared local memory of the host computer.

4. The method of claim 1 wherein:

each accelerator has a dedicated portion of memory in the shared local memory of the host computer;

notifying the host computer that the plurality of accelerators have read the data further comprises writing, by each accelerator to the accelerator'"'"'s dedicated portion, upon reading the data, a value different from a previous value in the dedicated portion; and

the method further comprises determining, by the host computer, that the current value residing in each dedicated portion of memory in the shared local memory of the host computer has changed.

5. The method of claim 1 wherein:

writing, by the host computer, remotely to the shared local memory for the plurality of accelerators the data to be broadcast further comprises initializing a value of a counter in the shared local memory for the plurality of accelerators;

notifying the host computer that the plurality of accelerators have read the data further comprises decrementing locally, by each accelerator upon reading the data, the counter residing in the shared local memory for the accelerators; and

the method further comprises determining, by the host computer, that the value of the counter is less than a predetermined threshold.

6. The method of claim 5 wherein decrementing locally, by each accelerator upon reading the data, the counter residing in the shared local memory for the accelerators further comprises:

locking by each accelerator the counter;

reading the value of the counter;

decrementing the value of the counter; and

writing the decremented value to the local memory of the accelerator.

7. A hybrid computing environment for broadcasting data, the hybrid computing environment comprising a host computer having a host computer architecture, a plurality of accelerators having an accelerator architecture, the accelerator architecture optimized, with respect to the host computer architecture, for speed of execution of a particular class of computing functions, the host computer and the accelerators adapted to one another for data communications by a system level message passing module, the host computer having local memory shared remotely with the accelerators, the accelerators having local memory for the plurality of accelerators shared remotely with the host computer,

the host computer comprising computer program instructions capable of writing, by the host computer remotely to the shared local memory for the plurality of accelerators, the data to be broadcast;

and

the accelerators comprising computer program instructions capable of reading, by each of the accelerators from the shared local memory for the plurality of accelerators, the data; and

notifying the host computer, by the plurality of accelerators, that the plurality of accelerators have read the data.

8. The hybrid computing environment of claim 7 wherein:

notifying the host computer that the plurality of accelerators have read the data further comprises incrementing remotely, by each accelerator upon reading the data, a counter residing in the shared local memory of the host computer; and

the host computer further comprises computer program instructions capable of determining, by the host computer, that the value of the counter exceeds a predetermined threshold.

9. The hybrid computing environment of claim 8 wherein incrementing remotely a counter residing in the shared local memory of the host computer further comprises:

locking by each accelerator the counter;

reading the value of the counter;

incrementing the value of the counter; and

writing the incremented value to the shared local memory of the host computer.

10. The hybrid computing environment of claim 7 wherein:

each accelerator has a dedicated portion of memory in the shared local memory of the host computer;

notifying the host computer that the plurality of accelerators have read the data further comprises writing, by each accelerator to the accelerator'"'"'s dedicated portion, upon reading the data, a value different from a previous value in the dedicated portion; and

the host computer further comprises computer program instructions capable of determining, by the host computer, that the current value residing in each dedicated portion of memory in the shared local memory of the host computer has changed.

11. The hybrid computing environment of claim 7 wherein:

writing, by the host computer, remotely to the shared local memory for the plurality of accelerators the data to be broadcast further comprises initializing a value of a counter in the shared local memory for the plurality of accelerators;

notifying the host computer that the plurality of accelerators have read the data further comprises decrementing locally, by each accelerator upon reading the data, the counter residing in the shared local memory for the accelerators; and

the host computer further comprises computer program instructions capable of determining, by the host computer, that the value of the counter is less than a predetermined threshold.

12. The hybrid computing environment of claim 11 wherein decrementing locally, by each accelerator upon reading the data, the counter residing in the shared local memory for the accelerators further comprises:

locking by each accelerator the counter;

reading the value of the counter;

decrementing the value of the counter; and

writing the decremented value to the local memory of the accelerator.

13. A computer program product for broadcasting data in a hybrid computing environment, the hybrid computing environment comprising a host computer having a host computer architecture, a plurality of accelerators having an accelerator architecture, the accelerator architecture optimized, with respect to the host computer architecture, for speed of execution of a particular class of computing functions, the host computer and the accelerators adapted to one another for data communications by a system level message passing module, the host computer having local memory shared remotely with the accelerators, the accelerators having local memory for the plurality of accelerators shared remotely with the host computer, the computer program product disposed in a computer readable, signal bearing medium, the computer program product comprising computer program instructions capable of:

writing, by the host computer remotely to the shared local memory for the plurality of accelerators, the data to be broadcast;

reading, by each of the accelerators from the shared local memory for the plurality of accelerators, the data; and

notifying the host computer, by the plurality of accelerators, that the plurality of accelerators have read the data.

14. The computer program product of claim 13 wherein:

notifying the host computer that the plurality of accelerators have read the data further comprises incrementing remotely, by each accelerator upon reading the data, a counter residing in the shared local memory of the host computer; and

the computer program product further comprises computer program instructions capable of determining, by the host computer, that the value of the counter exceeds a predetermined threshold.

15. The computer program product of claim 14 wherein incrementing remotely a counter residing in the shared local memory of the host computer further comprises:

locking by each accelerator the counter;

reading the value of the counter;

incrementing the value of the counter; and

writing the incremented value to the shared local memory of the host computer.

16. The computer program product of claim 13 wherein:

each accelerator has a dedicated portion of memory in the shared local memory of the host computer;

notifying the host computer that the plurality of accelerators have read the data further comprises writing, by each accelerator to the accelerator'"'"'s dedicated portion, upon reading the data, a value different from a previous value in the dedicated portion; and

the computer program product further comprises computer program instructions capable of determining, by the host computer, that the current value residing in each dedicated portion of memory in the shared local memory of the host computer has changed.

17. The computer program product of claim 13 wherein:

writing, by the host computer, remotely to the shared local memory for the plurality of accelerators the data to be broadcast further comprises initializing a value of a counter in the shared local memory for the plurality of accelerators;

notifying the host computer that the plurality of accelerators have read the data further comprises decrementing locally, by each accelerator upon reading the data, the counter residing in the shared local memory for the accelerators; and

the computer program product further comprises computer program instructions capable of determining, by the host computer, that the value of the counter is less than a predetermined threshold.

18. The computer program product of claim 17 wherein decrementing locally, by each accelerator upon reading the data, the counter residing in the shared local memory for the accelerators further comprises:

locking by each accelerator the counter;

reading the value of the counter;

decrementing the value of the counter; and

writing the decremented value to the local memory of the accelerator.

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

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