Processors interconnect fabric with relay broadcasting and accumulation of partial responses

• US 7,254,694 B2
• Filed: 02/10/2005
• Issued: 08/07/2007
• Est. Priority Date: 02/10/2005
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Clark, Leo J., Starke, William J., Fields, James S. Jr., Stuecheli, Jeffrey A., Guthrie, Guy L.
• Primary Examiner(s): Kim; Kenneth S.
• Application Number:
  US11/055,297
• Publication Number:
  US 20060179272A1
• Time in Patent Office: 908 Days
• Field of Search: None
• US Class Current: 712/28
• CPC Class Codes: G06F 13/385 for adaptation of a particular data processing system to different peripheral devices
• CPC Class Codes: G06F 9/546 Message passing systems or structures, e.g. queues

Abstract

A data processing system includes a plurality of processing units each having a respective point-to-point communication link with each of multiple others of the plurality of processing units but fewer than all of the plurality of processing units. Each of the plurality of processing units includes interconnect logic, coupled to each point-to-point communication link of that processing unit, that broadcasts requests received from one of the multiple others of the plurality of processing units to one or more of the plurality of processing units. The interconnect logic includes a partial response data structure including a plurality of entries each associating a partial response field with a plurality of flags respectively associated with each processing unit containing a snooper from which that processing unit will receive a partial response. The interconnect logic accumulates partial responses of processing units by reference to the partial response field to obtain an accumulated partial response, and when the plurality of flags indicate that all processing units from which partial responses are expected have returned a partial response, outputs the accumulated partial response.

Claims

1. A data processing system, comprising:

a plurality of processing units each having a respective point-to-point communication link with each of multiple others of said plurality of processing units but fewer than all of said plurality of processing units;

each of said plurality of processing units including interconnect logic, coupled to each point-to-point communication link of that processing unit, that broadcasts requests received from one of said multiple others of said plurality of processing units to one or more of said plurality of processing units;

said interconnect logic including a partial response data structure including a plurality of entries each associating a partial response field with a plurality of flags respectively associated with each processing unit containing a snooper from which that processing unit will receive a partial response, wherein said interconnect logic accumulates partial responses of processing units by reference to said partial response field to obtain an accumulated partial response, and when said plurality of flags indicate that all processing units from which partial responses are expected have returned a partial response, outputs said accumulated partial response.

2. The data processing system of claim 1, wherein:

said plurality of entries in said partial response data structure comprises a plurality of time-ordered entries allocated to requests in order of receipt; and

said interconnect logic deallocates the entry in response to determining all processing units from which partial responses are expected have returned a partial response.

3. The data processing system of claim 1, wherein:

said partial response data structure comprises a first partial response data structure utilized to accumulate partial responses for requests received directly from an originating processing unit via a point-to-point communication link; and

said interconnect logic comprises a second partial response data structure utilized to accumulate partial responses for requests received from a processing unit other than the originating processing unit.

4. The data processing system of claim 1, wherein said plurality of flags includes a flag for the processing unit containing the interconnect logic.

5. The data processing system of claim 1, wherein said interconnect logic outputs said accumulated partial response to another processing unit among said plurality of processing units for further partial response accumulation.

6. The data processing system of claim 1, wherein:

said data processing system further comprises combined response logic; and

said interconnect logic outputs said accumulated partial response to combined response logic to generate a combined response representing a system-wide response to said request.

19. The data processing system of claim 1, wherein:

said requests comprise data access requests; and

said interconnect logic outputs said accumulated partial response to another processing unit to which a processing unit containing the interconnect logic is connected by a point-to-point communication link.

7. A processing unit for a data processing system including a plurality of processing units, said processing unit comprising:

one or more processor cores; and

interconnect logic, coupled to each communication link of that processing unit, that broadcasts requests received from one of multiple others of said plurality of processing units to one or more of said plurality of processing units;

said interconnect logic including a partial response data structure including a plurality of entries each associating a partial response field with a plurality of flags respectively associated with each processing unit containing a snooper from which that processing unit will receive a partial response, wherein said interconnect logic accumulates partial responses of processing units by reference to said partial response field to obtain an accumulated partial response, and when said plurality of flags indicate that all processing units from which partial responses are expected have returned a partial response, outputs said accumulated partial response.

8. The processing unit of claim 7, wherein:

said plurality of entries in said partial response data structure comprises a plurality of time-ordered entries allocated to requests in order of receipt; and

said interconnect logic deallocates the entry in response to determining all processing units from which partial responses are expected have returned a partial response.

9. The processing unit of claim 7, wherein:

said partial response data structure comprises a first partial response data structure utilized to accumulate partial responses for requests received directly from an originating processing unit via a point-to-point communication link; and

said interconnect logic comprises a second partial response data structure utilized to accumulate partial responses for requests received from a processing unit other than the originating processing unit.

10. The processing unit of claim 7, wherein said plurality of flags includes a flag for the processing unit.

11. The processing unit of claim 7, wherein said interconnect logic outputs said accumulated partial response to another processing unit among said plurality of processing units for further partial response accumulation.

12. The processing unit of claim 7, wherein said data processing system includes combined response logic, and wherein said interconnect logic outputs said accumulated partial response to combined response logic to generate a combined response representing a system-wide response to said request.

20. The processing unit of claim 7, wherein:

said requests comprise data access requests; and

said interconnect logic outputs said accumulated partial response to another processing unit via a point-to-point communication link.

13. A method of data processing in a data processing system including a plurality of processing units, said method comprising:

broadcasting requests received from one of multiple others of said plurality of processing units to one or more of said plurality of processing units; and

in a partial response data structure including a plurality of entries each associating a partial response field with a plurality of flags respectively associated with each processing unit containing a snooper from which that processing unit will receive a partial response, accumulating partial responses of processing units by reference to said partial response field to obtain an accumulated partial response;

setting one or more flags among said plurality of flags to indicate processing units from which partial responses have been received; and

when said plurality of flags indicate that all processing units from which partial responses are expected have returned a partial response, outputting said accumulated partial response.

14. The method of claim 13, wherein:

said plurality of entries in said partial response data structure comprises a plurality of time-ordered entries;

said method further comprises;

allocating entries to requests in order of receipt; and

deallocating an entry in response to determining all processing units from which partial responses are expected have returned a partial response.

15. The method of claim 13, wherein:

said partial response data structure comprises a first partial response data structure;

said accumulating comprises;

accumulating, in said first partial response data structure, partial responses for requests received directly from an originating processing unit via a point-to-point communication link; and

accumulating, in a second partial response data structure, partial responses for requests received from a processing unit other than the originating processing unit.

16. The method of claim 13, wherein said plurality of flags includes a flag for the processing unit.

17. The method of claim 13, wherein said outputting comprises outputting said accumulated partial response to another processing unit among said plurality of processing units for further partial response accumulation.

18. The method of claim 13, wherein said outputting comprises outputting said accumulated partial response to combined response logic to generate a combined response representing a system-wide response to said request.

21. The method of claim 13, wherein:

said requests comprise data access requests;

the broadcasting, accumulating, setting and outputting steps are all performed by a particular one of said plurality of processing units; and

the step of outputting the accumulated partial response comprises the particular processing unit outputting the accumulated partial response to another processing unit via a point-to-point communication link.