{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2024,10,23]],"date-time":"2024-10-23T05:40:48Z","timestamp":1729662048157,"version":"3.28.0"},"reference-count":39,"publisher":"IEEE","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2015,5]]},"DOI":"10.1109\/ipdpsw.2015.55","type":"proceedings-article","created":{"date-parts":[[2015,10,1]],"date-time":"2015-10-01T18:05:10Z","timestamp":1443722710000},"page":"998-1007","source":"Crossref","is-referenced-by-count":0,"title":["Modeling Cooperative Threads to Project GPU Performance for Adaptive Parallelism"],"prefix":"10.1109","author":[{"given":"Jiayuan","family":"Meng","sequence":"first","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Thomas D.","family":"Uram","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Vitali","family":"Morozov","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Venkatram","family":"Vishwanath","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Kalyan","family":"Kumaran","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"263","reference":[{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.1109\/HPCA.2011.5749745"},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.1109\/SC.2005.20"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.1109\/SC.2002.10004"},{"key":"ref32","doi-asserted-by":"crossref","DOI":"10.1145\/2145816.2145819","article-title":"GPUPerf: A performance analysis framework for identifying potential benefits in GPGPU applications","author":"sim","year":"2012","journal-title":"PPoPP"},{"key":"ref31","first-page":"6","article-title":"Conceptual design of a proton computed tomography system for applications in proton radiation therapy","volume":"51","author":"reinhard","year":"2004","journal-title":"IEEE Trans Nuclear Science"},{"key":"ref30","doi-asserted-by":"publisher","DOI":"10.1146\/annurev.nucl.51.101701.132506"},{"key":"ref37","doi-asserted-by":"publisher","DOI":"10.1145\/1735688.1735697"},{"key":"ref36","article-title":"CUDA-Lite: Reducing GPU programming complexity","author":"ueng","year":"2008","journal-title":"LCPC"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.1109\/PACT.2009.18"},{"key":"ref34","doi-asserted-by":"publisher","DOI":"10.1145\/773056.773060"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1109\/SC.Companion.2012.32"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.1145\/1995896.1995932"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1145\/1995896.1995924"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.1137\/040609458"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1109\/71.86110"},{"key":"ref15","doi-asserted-by":"publisher","DOI":"10.1145\/2063348.2063356"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.1145\/1555754.1555775"},{"key":"ref17","article-title":"Modern potentials and the properties of condensed 4He","author":"kalos","year":"1981","journal-title":"Phys Rev C 66 044310–1 14"},{"key":"ref18","first-page":"37","article-title":"Predictive performance and scalability modeling of a large-scale application","author":"kerbyson","year":"2001","journal-title":"SC"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1109\/JVA.2006.39"},{"year":"0","key":"ref28","article-title":"NVIDIA Corporation. NVIDIA Corporation. NVIDIA Visual Profiler"},{"key":"ref4","article-title":"Automatic C-to-CUDA code generation for affine programs","author":"baskaran","year":"2010","journal-title":"CC"},{"key":"ref27","doi-asserted-by":"publisher","DOI":"10.1145\/1542275.1542313"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.1109\/MC.2009.372"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1109\/IPDPSW.2013.236"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.1103\/PhysRevC.66.044310"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1145\/1375527.1375562"},{"key":"ref8","article-title":"Model-driven auto tuning of sparse matrix-vector multiply on GPUs","author":"choi","year":"2010","journal-title":"PPoPP"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.1145\/1995896.1995928"},{"key":"ref2","doi-asserted-by":"publisher","DOI":"10.1145\/1693453.1693470"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1145\/1854273.1854318"},{"year":"2012","key":"ref1","article-title":"NVIDIA's next generation CUDA compute architecture: Kepler GK110"},{"key":"ref20","article-title":"PyCUDA: GPU run-time code generation for high-performance computing","author":"kl\u00f6ckner","year":"2009","journal-title":"CoRR"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1145\/1168857.1168881"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.1109\/HIPC.2009.5433179"},{"key":"ref24","doi-asserted-by":"publisher","DOI":"10.1109\/SC.2010.36"},{"key":"ref23","doi-asserted-by":"crossref","DOI":"10.1145\/1229428.1229479","article-title":"Methods of inference and learning for performance modeling of parallel applications","author":"lee","year":"2007","journal-title":"PPoPP"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1109\/SC.2012.42"},{"key":"ref25","article-title":"GROPHECY: GPU performance projection from CPU code skeletons","author":"meng","year":"2011","journal-title":"SC"}],"event":{"name":"2015 IEEE International Parallel and Distributed Processing Symposium Workshop (IPDPSW)","start":{"date-parts":[[2015,5,25]]},"location":"Hyderabad, India","end":{"date-parts":[[2015,5,29]]}},"container-title":["2015 IEEE International Parallel and Distributed Processing Symposium Workshop"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx7\/7275194\/7284273\/07284419.pdf?arnumber=7284419","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2019,8,31]],"date-time":"2019-08-31T01:13:44Z","timestamp":1567214024000},"score":1,"resource":{"primary":{"URL":"http:\/\/ieeexplore.ieee.org\/document\/7284419\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2015,5]]},"references-count":39,"URL":"https:\/\/doi.org\/10.1109\/ipdpsw.2015.55","relation":{},"subject":[],"published":{"date-parts":[[2015,5]]}}}