Performing an allreduce operation on a plurality of compute nodes of a parallel computer
원문보기
IPC분류정보
국가/구분
United States(US) Patent
등록
국제특허분류(IPC7판)
G06F-015/80
G06F-009/30
G06F-009/302
출원번호
US-0124763
(2008-05-21)
등록번호
US-8375197
(2013-02-12)
발명자
/ 주소
Faraj, Ahmad
출원인 / 주소
International Business Machines Corporation
인용정보
피인용 횟수 :
0인용 특허 :
42
초록▼
Methods, apparatus, and products are disclosed for performing an allreduce operation on a plurality of compute nodes of a parallel computer, each node including at least two processing cores, that include: performing, for each node, a local reduction operation using allreduce contribution data for t
Methods, apparatus, and products are disclosed for performing an allreduce operation on a plurality of compute nodes of a parallel computer, each node including at least two processing cores, that include: performing, for each node, a local reduction operation using allreduce contribution data for the cores of that node, yielding, for each node, a local reduction result for one or more representative cores for that node; establishing one or more logical rings among the nodes, each logical ring including only one of the representative cores from each node; performing, for each logical ring, a global allreduce operation using the local reduction result for the representative cores included in that logical ring, yielding a global allreduce result for each representative core included in that logical ring; and performing, for each node, a local broadcast operation using the global allreduce results for each representative core on that node.
대표청구항▼
1. A method of performing an allreduce operation on a plurality of compute nodes of a parallel computer, each compute node comprising at least two processing cores, each processing core having contribution data for the allreduce operation, the method comprising: performing, for each compute node, a
1. A method of performing an allreduce operation on a plurality of compute nodes of a parallel computer, each compute node comprising at least two processing cores, each processing core having contribution data for the allreduce operation, the method comprising: performing, for each compute node, a local reduction operation using the contribution data for the processing cores of that compute node, yielding, for each compute node, a local reduction result for one or more representative cores for that compute node;establishing one or more logical rings among the compute nodes, each logical ring including only one of the representative cores from each compute node, wherein in a logical ring, all data that is transferred between processing cores travels from one core of one node to a next core in a next node in a circular manner;performing, for each logical ring, a global allreduce operation using the local reduction result for the representative cores included in that logical ring, yielding a global allreduce result for each representative core included in that logical ring; andperforming, for each compute node, a local broadcast operation using the global allreduce results for each representative core on that compute node. 2. The method of claim 1 wherein establishing one or more logical rings among the compute nodes, each logical ring including only one of the representative cores from each compute node further comprises establishing a single logical ring using a single representative core from each compute node. 3. The method of claim 1 wherein performing, for each logical ring, a global allreduce operation using the local reduction result for the representative cores included in that logical ring, yielding a global allreduce result for each representative core further comprising: performing, for each logical ring, a reduce scatter operation using the local reduction result for each representative core included in that logical ring, yielding reduce scatter results for each representative core; andperforming, for each logical ring, an allgather operation using the reduce scatter results for each representative core included in that logical ring, yielding the global allreduce result for each representative core. 4. The method of claim 1 further comprising: assigning, on each compute node, the processing cores on that compute node to one of a plurality of logical ring domains, each logical ring domain uniquely corresponding to one of the logical rings; andselecting, for each logical ring domain, one of the processing cores assigned to that logical ring domain as the representative core for that logical ring domain. 5. The method of claim 4 wherein: performing, for each compute node, a local reduction operation using the contribution data for the processing cores of that compute node, yielding, for each compute node, a local reduction result for one or more representative cores for that compute node further comprises performing, for each logical ring domain, the local reduction operation using the contribution data for the processing cores assigned to that logical ring domain, yielding, for each logical ring domain, the local reduction result for the representative core of that logical ring domain;establishing one or more logical rings among the compute nodes, each logical ring including only one of the representative cores from each compute node further comprises establishing, for each logical ring domain, a logical ring including the representative cores assigned to that logical ring domain; andperforming, for each compute node, a local broadcast operation using the global allreduce results for each representative core on that compute node further comprises: performing, for each compute node, a local allreduce operation using the global allreduce results for each representative core on that compute node, yielding local allreduce results for each representative core on that compute node, and performing, for each logical ring domain for each compute node, the local broadcast operation using the local allreduce results for each representative core included in that logical ring domain. 6. The method of claim 1 wherein the plurality of compute nodes are connected using a plurality of data communications networks, at least one of the data communications networks configured for point to point operations, and at least one data communications networks optimized for collective operations. 7. A parallel computer for performing an allreduce operation on a plurality of compute nodes, each compute node comprising at least two processing cores, each processing core having contribution data for the allreduce operation, the parallel computer comprising computer memory operatively coupled to the processing cores of the parallel computer, the computer memory having disposed within it computer program instructions capable of: performing, for each compute node, a local reduction operation using the contribution data for the processing cores of that compute node, yielding, for each compute node, a local reduction result for one or more representative cores for that compute node;establishing one or more logical rings among the compute nodes, each logical ring including only one of the representative cores from each compute node, wherein in a logical ring, all data that is transferred between processing cores travels from one core of one node to a next core in a next node in a circular manner;performing, for each logical ring, a global allreduce operation using the local reduction result for the representative cores included in that logical ring, yielding a global allreduce result for each representative core included in that logical ring; andperforming, for each compute node, a local broadcast operation using the global allreduce results for each representative core on that compute node. 8. The parallel computer of claim 7 wherein establishing one or more logical rings among the compute nodes, each logical ring including only one of the representative cores from each compute node further comprises establishing a single logical ring using a single representative core from each compute node. 9. The parallel computer of claim 7 wherein performing, for each logical ring, a global allreduce operation using the local reduction result for the representative cores included in that logical ring, yielding a global allreduce result for each representative core further comprising: performing, for each logical ring, a reduce scatter operation using the local reduction result for each representative core included in that logical ring, yielding reduce scatter results for each representative core; andperforming, for each logical ring, an allgather operation using the reduce scatter results for each representative core included in that logical ring, yielding the global allreduce result for each representative core. 10. The parallel computer of claim 7 wherein the computer memory has disposed within it computer program instructions capable of: assigning, on each compute node, the processing cores on that compute node to one of a plurality of logical ring domains, each logical ring domain uniquely corresponding to one of the logical rings; andselecting, for each logical ring domain, one of the processing cores assigned to that logical ring domain as the representative core for that logical ring domain. 11. The parallel computer of claim 10 wherein: performing, for each compute node, a local reduction operation using the contribution data for the processing cores of that compute node, yielding, for each compute node, a local reduction result for one or more representative cores for that compute node further comprises performing, for each logical ring domain, the local reduction operation using the contribution data for the processing cores assigned to that logical ring domain, yielding, for each logical ring domain, the local reduction result for the representative core of that logical ring domain; establishing one or more logical rings among the compute nodes, each logical ring including only one of the representative cores from each compute node further comprises establishing, for each logical ring domain, a logical ring including the representative cores assigned to that logical ring domain; andperforming, for each compute node, a local broadcast operation using the global allreduce results for each representative core on that compute node further comprises: performing, for each compute node, a local allreduce operation using the global allreduce results for each representative core on that compute node, yielding local allreduce results for each representative core on that compute node, and performing, for each logical ring domain for each compute node, the local broadcast operation using the local allreduce results for each representative core included in that logical ring domain. 12. The parallel computer of claim 7 wherein the plurality of compute nodes are connected using a plurality of data communications networks, at least one of the data communications networks optimized for point to point operations, and at least one data communications networks configured for collective operations. 13. A computer program product for performing an allreduce operation on a plurality of compute nodes of a parallel computer, each compute node comprising at least two processing cores, each processing core having contribution data for the allreduce operation, the computer program product disposed upon a computer readable recordable medium, the computer program product comprising computer program instructions capable of: performing, for each compute node, a local reduction operation using the contribution data for the processing cores of that compute node, yielding, for each compute node, a local reduction result for one or more representative cores for that compute node;establishing one or more logical rings among the compute nodes, each logical ring including only one of the representative cores from each compute node, wherein in a logical ring, all data that is transferred between processing cores travels from one core of one node to a next core in a next node in a circular manner;performing, for each logical ring, a global allreduce operation using the local reduction result for the representative cores included in that logical ring, yielding a global allreduce result for each representative core included in that logical ring; andperforming, for each compute node, a local broadcast operation using the global allreduce results for each representative core on that compute node. 14. The computer program product of claim 13 wherein establishing one or more logical rings among the compute nodes, each logical ring including only one of the representative cores from each compute node further comprises establishing a single logical ring using a single representative core from each compute node. 15. The computer program product of claim 13 wherein performing, for each logical ring, a global allreduce operation using the local reduction result for the representative cores included in that logical ring, yielding a global allreduce result for each representative core further comprising: performing, for each logical ring, a reduce scatter operation using the local reduction result for each representative core included in that logical ring, yielding reduce scatter results for each representative core; andperforming, for each logical ring, an allgather operation using the reduce scatter results for each representative core included in that logical ring, yielding the global allreduce result for each representative core. 16. The computer program product of claim 13 further comprising computer program instructions capable of: assigning, on each compute node, the processing cores on that compute node to one of a plurality of logical ring domains, each logical ring domain uniquely corresponding to one of the logical rings; andselecting, for each logical ring domain, one of the processing cores assigned to that logical ring domain as the representative core for that logical ring domain. 17. The computer program product of claim 16 wherein: performing, for each compute node, a local reduction operation using the contribution data for the processing cores of that compute node, yielding, for each compute node, a local reduction result for one or more representative cores for that compute node further comprises performing, for each logical ring domain, the local reduction operation using the contribution data for the processing cores assigned to that logical ring domain, yielding, for each logical ring domain, the local reduction result for the representative core of that logical ring domain; establishing one or more logical rings among the compute nodes, each logical ring including only one of the representative cores from each compute node further comprises establishing, for each logical ring domain, a logical ring including the representative cores assigned to that logical ring domain; andperforming, for each compute node, a local broadcast operation using the global allreduce results for each representative core on that compute node further comprises: performing, for each compute node, a local allreduce operation using the global allreduce results for each representative core on that compute node, yielding local allreduce results for each representative core on that compute node, and performing, for each logical ring domain for each compute node, the local broadcast operation using the local allreduce results for each representative core included in that logical ring domain. 18. The computer program product of claim 13 wherein the plurality of compute nodes are connected using a plurality of data communications networks, at least one of the data communications networks optimized for point to point operations, and at least one data communications networks configured for collective operations. 19. A computer program product for performing an allreduce operation on a plurality of compute nodes of a parallel computer, each compute node comprising at least two processing cores, each processing core having contribution data for the allreduce operation, the computer program product disposed upon a computer readable transmission medium, the computer program product comprising computer program instructions capable of: performing, for each compute node, a local reduction operation using the contribution data for the processing cores of that compute node, yielding, for each compute node, a local reduction result for one or more representative cores for that compute node;establishing one or more logical rings among the compute nodes, each logical ring including only one of the representative cores from each compute node, wherein in a logical ring, all data that is transferred between processing cores travels from one core of one node to a next core in a next node in a circular manner;performing, for each logical ring, a global allreduce operation using the local reduction result for the representative cores included in that logical ring, yielding a global allreduce result for each representative core included in that logical ring; andperforming, for each compute node, a local broadcast operation using the global allreduce results for each representative core on that compute node.
연구과제 타임라인
LOADING...
LOADING...
LOADING...
LOADING...
LOADING...
이 특허에 인용된 특허 (42)
Scott Steven L. ; Pribnow Richard D. ; Logghe Peter G. ; Kunkel Daniel L. ; Schwoerer Gerald A., Adaptive congestion control mechanism for modular computer networks.
Kato Sadayuki,JPX ; Ishihata Hiroaki,JPX ; Horie Takeshi,JPX ; Inano Satoshi,JPX ; Shimizu Toshiyuki,JPX, Data gathering/scattering system for a plurality of processors in a parallel computer.
Michael Olivier, Dynamically matching users for group communications based on a threshold degree of matching of sender and recipient predetermined acceptance criteria.
Cypher Robert E. (Los Gatos CA) Sanz Jorge L. C. (Los Gatos CA), Hierarchical interconnection network architecture for parallel processing, having interconnections between bit-addressib.
Flaig Charles M. (Pasadena CA) Seitz Charles L. (San Luis Rey CA), Inter-computer message routing system with each computer having separate routinng automata for each dimension of the net.
Carmichael Richard D. ; Ward Joel M. ; Winchell Michael A., Method and apparatus for controlling (N+I) I/O channels with (N) data managers in a homogenous software programmable en.
Krishnamoorthy Ashok V. (11188 Caminito Rodar San Diego CA 92126) Kiamilev Fouad (c/o UNC Charlotte ; Dept. of EE ; Smith Hall Room 332 Charlotte NC 28223), Packet-switched self-routing multistage interconnection network having contention-free fanout, low-loss routing, and fan.
Yasuda Yoshiko,JPX ; Tanaka Teruo,JPX, Parallel computer system using properties of messages to route them through an interconnect network and to select virtua.
Wilkinson Paul Amba ; Dieffenderfer James Warren ; Kogge Peter Michael ; Schoonover Nicholas Jerome, Partitioning of processing elements in a SIMD/MIMD array processor.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.