{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,20]],"date-time":"2026-07-20T11:07:25Z","timestamp":1784545645426,"version":"3.55.0"},"publisher-location":"New York, NY, USA","reference-count":15,"publisher":"ACM","license":[{"start":{"date-parts":[[2011,11,12]],"date-time":"2011-11-12T00:00:00Z","timestamp":1321056000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/www.acm.org\/publications\/policies\/copyright_policy#Background"}],"funder":[{"DOI":"10.13039\/501100002855","name":"Ministry of Science and Technology of the People's Republic of China","doi-asserted-by":"publisher","award":["2011CB302500"],"award-info":[{"award-number":["2011CB302500"]}],"id":[{"id":"10.13039\/501100002855","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100002855","name":"Ministry of Science and Technology of the People's Republic of China","doi-asserted-by":"publisher","award":["2009AA01A129"],"award-info":[{"award-number":["2009AA01A129"]}],"id":[{"id":"10.13039\/501100002855","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["6.08E+31"],"award-info":[{"award-number":["6.08E+31"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":[],"published-print":{"date-parts":[[2011,11,12]]},"DOI":"10.1145\/2063384.2063431","type":"proceedings-article","created":{"date-parts":[[2011,11,8]],"date-time":"2011-11-08T13:32:09Z","timestamp":1320759129000},"page":"1-11","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":75,"title":["Fast implementation of DGEMM on Fermi GPU"],"prefix":"10.1145","author":[{"given":"Guangming","family":"Tan","sequence":"first","affiliation":[{"name":"Key Laboratory of Computer Architecture, Institute of Computing Technology"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Linchuan","family":"Li","sequence":"additional","affiliation":[{"name":"Key Laboratory of Computer Architecture, Institute of Computing Technology"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Sean","family":"Triechle","sequence":"additional","affiliation":[{"name":"Nvidia Corporation"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Everett","family":"Phillips","sequence":"additional","affiliation":[{"name":"Nvidia Corporation"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yungang","family":"Bao","sequence":"additional","affiliation":[{"name":"Key Laboratory of Computer Architecture, Institute of Computing Technology"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ninghui","family":"Sun","sequence":"additional","affiliation":[{"name":"Key Laboratory of Computer Architecture, Institute of Computing Technology"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"320","published-online":{"date-parts":[[2011,11,12]]},"reference":[{"key":"e_1_3_2_1_1_1","volume-title":"August","author":"AMD.","year":"2010","unstructured":"AMD. ATI Stream Computing OpenCL Programming Guide, rev1.05 , August 2010 . AMD. ATI Stream Computing OpenCL Programming Guide, rev1.05, August 2010."},{"key":"e_1_3_2_1_3_1","doi-asserted-by":"publisher","DOI":"10.1109\/JPROC.2004.840848"},{"key":"e_1_3_2_1_4_1","doi-asserted-by":"publisher","DOI":"10.1145\/77626.79170"},{"key":"e_1_3_2_1_5_1","doi-asserted-by":"publisher","DOI":"10.1145\/1356052.1356053"},{"key":"e_1_3_2_1_6_1","unstructured":"C. Jang. Gatlas gpu automatically tuned linear algebra software. http:\/\/golem5.org\/gatlas\/.  C. Jang. Gatlas gpu automatically tuned linear algebra software. http:\/\/golem5.org\/gatlas\/."},{"key":"e_1_3_2_1_7_1","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-642-01970-8_89"},{"key":"e_1_3_2_1_8_1","doi-asserted-by":"publisher","DOI":"10.1145\/1964218.1964227"},{"key":"e_1_3_2_1_10_1","unstructured":"NVIDIA. Cuda Community Showcase. http:\/\/www.nvidia.com\/object\/cuda_apps_flash_new.html.  NVIDIA. Cuda Community Showcase. http:\/\/www.nvidia.com\/object\/cuda_apps_flash_new.html."},{"key":"e_1_3_2_1_11_1","unstructured":"NVIDIA. Nvidia's next generation cuda compute architecture: Fermi. http:\/\/www.nvidia.com\/object\/fermi_architecture.html 2009.  NVIDIA. Nvidia's next generation cuda compute architecture: Fermi. http:\/\/www.nvidia.com\/object\/fermi_architecture.html 2009."},{"key":"e_1_3_2_1_12_1","volume-title":"Version 3.2","author":"NVIDIA.","year":"2010","unstructured":"NVIDIA. CUDA C Programming Guide , Version 3.2 , 2010 . NVIDIA. CUDA C Programming Guide, Version 3.2, 2010."},{"key":"e_1_3_2_1_13_1","doi-asserted-by":"publisher","DOI":"10.1145\/1345206.1345220"},{"key":"e_1_3_2_1_14_1","doi-asserted-by":"publisher","DOI":"10.1145\/1356058.1356084"},{"key":"e_1_3_2_1_15_1","doi-asserted-by":"publisher","DOI":"10.5555\/1413370.1413402"},{"key":"e_1_3_2_1_16_1","doi-asserted-by":"publisher","DOI":"10.1109\/ISPASS.2010.5452013"},{"key":"e_1_3_2_1_17_1","doi-asserted-by":"publisher","DOI":"10.1109\/JPROC.2004.840444"}],"event":{"name":"SC '11: International Conference for High Performance Computing, Networking, Storage and Analysis","location":"Seattle Washington","acronym":"SC '11","sponsor":["SIGARCH ACM Special Interest Group on Computer Architecture","IEEE-CS Computer Society"]},"container-title":["Proceedings of 2011 International Conference for High Performance Computing, Networking, Storage and Analysis"],"original-title":[],"link":[{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/2063384.2063431","content-type":"unspecified","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/2063384.2063431","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,6,18]],"date-time":"2025-06-18T10:06:07Z","timestamp":1750241167000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/2063384.2063431"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2011,11,12]]},"references-count":15,"alternative-id":["10.1145\/2063384.2063431","10.1145\/2063384"],"URL":"https:\/\/doi.org\/10.1145\/2063384.2063431","relation":{},"subject":[],"published":{"date-parts":[[2011,11,12]]},"assertion":[{"value":"2011-11-12","order":2,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}