Control of data transfer between processing nodes
A data processing system 900 comprises nodes, each comprising a memory which stores an array 910 of data items, wherein each node belongs to at least two different sets 920, 930 of nodes. A processor of each node is configured to execute compute instructions during a compute phase and following a pr...
Gespeichert in:
| Hauptverfasser: | , , |
|---|---|
| Format: | Patent |
| Sprache: | eng |
| Schlagworte: | |
| Online-Zugang: | Volltext bestellen |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| container_end_page | |
|---|---|
| container_issue | |
| container_start_page | |
| container_title | |
| container_volume | |
| creator | Fabian Tschopp Lorenzo Cevolani Ola Tørudbakken |
| description | A data processing system 900 comprises nodes, each comprising a memory which stores an array 910 of data items, wherein each node belongs to at least two different sets 920, 930 of nodes. A processor of each node is configured to execute compute instructions during a compute phase and following a precompiled synchronisation barrier, enter an exchange phase, when it takes part in a reduce-scatter collective using the respective array of data items to obtain a reduced subset of an array of data items, wherein the reduce-scatter collective is performed between nodes of a different one of the respective different sets of processing nodes. Subsequently, the processor exchanges the respective reduce subset of the array of data items using an all-reduce collective with processing nodes of a further of the different sets of processing nodes to which the respective processing node belongs to obtain a further reduced subset of the array of data items. Subsequently, the processor takes part in an all-gather collective using the further reduced subset to obtain a reduced array of data items, wherein the all-gather collective is performed between nodes of one of the different ones of the respective sets of nodes. |
| format | Patent |
| fullrecord | <record><control><sourceid>epo_EVB</sourceid><recordid>TN_cdi_epo_espacenet_GB2593756A</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>GB2593756A</sourcerecordid><originalsourceid>FETCH-epo_espacenet_GB2593756A3</originalsourceid><addsrcrecordid>eNrjZDB0zs8rKcrPUchPU0hJLElUKClKzCtOSy1SSEotKU9NzVMoKMpPTi0uzsxLV8jLT0kt5mFgTUvMKU7lhdLcDPJuriHOHrqpBfnxqcUFicmpeakl8e5ORqaWxuamZo7GhFUAAPa-KwM</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>patent</recordtype></control><display><type>patent</type><title>Control of data transfer between processing nodes</title><source>esp@cenet</source><creator>Fabian Tschopp ; Lorenzo Cevolani ; Ola Tørudbakken</creator><creatorcontrib>Fabian Tschopp ; Lorenzo Cevolani ; Ola Tørudbakken</creatorcontrib><description>A data processing system 900 comprises nodes, each comprising a memory which stores an array 910 of data items, wherein each node belongs to at least two different sets 920, 930 of nodes. A processor of each node is configured to execute compute instructions during a compute phase and following a precompiled synchronisation barrier, enter an exchange phase, when it takes part in a reduce-scatter collective using the respective array of data items to obtain a reduced subset of an array of data items, wherein the reduce-scatter collective is performed between nodes of a different one of the respective different sets of processing nodes. Subsequently, the processor exchanges the respective reduce subset of the array of data items using an all-reduce collective with processing nodes of a further of the different sets of processing nodes to which the respective processing node belongs to obtain a further reduced subset of the array of data items. Subsequently, the processor takes part in an all-gather collective using the further reduced subset to obtain a reduced array of data items, wherein the all-gather collective is performed between nodes of one of the different ones of the respective sets of nodes.</description><language>eng</language><subject>CALCULATING ; COMPUTING ; COUNTING ; ELECTRIC DIGITAL DATA PROCESSING ; PHYSICS</subject><creationdate>2021</creationdate><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://worldwide.espacenet.com/publicationDetails/biblio?FT=D&date=20211006&DB=EPODOC&CC=GB&NR=2593756A$$EHTML$$P50$$Gepo$$Hfree_for_read</linktohtml><link.rule.ids>230,308,776,881,25542,76516</link.rule.ids><linktorsrc>$$Uhttps://worldwide.espacenet.com/publicationDetails/biblio?FT=D&date=20211006&DB=EPODOC&CC=GB&NR=2593756A$$EView_record_in_European_Patent_Office$$FView_record_in_$$GEuropean_Patent_Office$$Hfree_for_read</linktorsrc></links><search><creatorcontrib>Fabian Tschopp</creatorcontrib><creatorcontrib>Lorenzo Cevolani</creatorcontrib><creatorcontrib>Ola Tørudbakken</creatorcontrib><title>Control of data transfer between processing nodes</title><description>A data processing system 900 comprises nodes, each comprising a memory which stores an array 910 of data items, wherein each node belongs to at least two different sets 920, 930 of nodes. A processor of each node is configured to execute compute instructions during a compute phase and following a precompiled synchronisation barrier, enter an exchange phase, when it takes part in a reduce-scatter collective using the respective array of data items to obtain a reduced subset of an array of data items, wherein the reduce-scatter collective is performed between nodes of a different one of the respective different sets of processing nodes. Subsequently, the processor exchanges the respective reduce subset of the array of data items using an all-reduce collective with processing nodes of a further of the different sets of processing nodes to which the respective processing node belongs to obtain a further reduced subset of the array of data items. Subsequently, the processor takes part in an all-gather collective using the further reduced subset to obtain a reduced array of data items, wherein the all-gather collective is performed between nodes of one of the different ones of the respective sets of nodes.</description><subject>CALCULATING</subject><subject>COMPUTING</subject><subject>COUNTING</subject><subject>ELECTRIC DIGITAL DATA PROCESSING</subject><subject>PHYSICS</subject><fulltext>true</fulltext><rsrctype>patent</rsrctype><creationdate>2021</creationdate><recordtype>patent</recordtype><sourceid>EVB</sourceid><recordid>eNrjZDB0zs8rKcrPUchPU0hJLElUKClKzCtOSy1SSEotKU9NzVMoKMpPTi0uzsxLV8jLT0kt5mFgTUvMKU7lhdLcDPJuriHOHrqpBfnxqcUFicmpeakl8e5ORqaWxuamZo7GhFUAAPa-KwM</recordid><startdate>20211006</startdate><enddate>20211006</enddate><creator>Fabian Tschopp</creator><creator>Lorenzo Cevolani</creator><creator>Ola Tørudbakken</creator><scope>EVB</scope></search><sort><creationdate>20211006</creationdate><title>Control of data transfer between processing nodes</title><author>Fabian Tschopp ; Lorenzo Cevolani ; Ola Tørudbakken</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-epo_espacenet_GB2593756A3</frbrgroupid><rsrctype>patents</rsrctype><prefilter>patents</prefilter><language>eng</language><creationdate>2021</creationdate><topic>CALCULATING</topic><topic>COMPUTING</topic><topic>COUNTING</topic><topic>ELECTRIC DIGITAL DATA PROCESSING</topic><topic>PHYSICS</topic><toplevel>online_resources</toplevel><creatorcontrib>Fabian Tschopp</creatorcontrib><creatorcontrib>Lorenzo Cevolani</creatorcontrib><creatorcontrib>Ola Tørudbakken</creatorcontrib><collection>esp@cenet</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Fabian Tschopp</au><au>Lorenzo Cevolani</au><au>Ola Tørudbakken</au><format>patent</format><genre>patent</genre><ristype>GEN</ristype><title>Control of data transfer between processing nodes</title><date>2021-10-06</date><risdate>2021</risdate><abstract>A data processing system 900 comprises nodes, each comprising a memory which stores an array 910 of data items, wherein each node belongs to at least two different sets 920, 930 of nodes. A processor of each node is configured to execute compute instructions during a compute phase and following a precompiled synchronisation barrier, enter an exchange phase, when it takes part in a reduce-scatter collective using the respective array of data items to obtain a reduced subset of an array of data items, wherein the reduce-scatter collective is performed between nodes of a different one of the respective different sets of processing nodes. Subsequently, the processor exchanges the respective reduce subset of the array of data items using an all-reduce collective with processing nodes of a further of the different sets of processing nodes to which the respective processing node belongs to obtain a further reduced subset of the array of data items. Subsequently, the processor takes part in an all-gather collective using the further reduced subset to obtain a reduced array of data items, wherein the all-gather collective is performed between nodes of one of the different ones of the respective sets of nodes.</abstract><oa>free_for_read</oa></addata></record> |
| fulltext | fulltext_linktorsrc |
| identifier | |
| ispartof | |
| issn | |
| language | eng |
| recordid | cdi_epo_espacenet_GB2593756A |
| source | esp@cenet |
| subjects | CALCULATING COMPUTING COUNTING ELECTRIC DIGITAL DATA PROCESSING PHYSICS |
| title | Control of data transfer between processing nodes |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-14T03%3A24%3A43IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-epo_EVB&rft_val_fmt=info:ofi/fmt:kev:mtx:patent&rft.genre=patent&rft.au=Fabian%20Tschopp&rft.date=2021-10-06&rft_id=info:doi/&rft_dat=%3Cepo_EVB%3EGB2593756A%3C/epo_EVB%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true |