{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,1,9]],"date-time":"2026-01-09T14:34:04Z","timestamp":1767969244902,"version":"3.49.0"},"publisher-location":"New York, NY, USA","reference-count":34,"publisher":"ACM","license":[{"start":{"date-parts":[[2015,11,15]],"date-time":"2015-11-15T00:00:00Z","timestamp":1447545600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/www.acm.org\/publications\/policies\/copyright_policy#Background"}],"content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":[],"published-print":{"date-parts":[[2015,11,15]]},"DOI":"10.1145\/2807591.2807602","type":"proceedings-article","created":{"date-parts":[[2015,10,27]],"date-time":"2015-10-27T13:07:31Z","timestamp":1445951251000},"page":"1-12","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":29,"title":["Improving concurrency and asynchrony in multithreaded MPI applications using software offloading"],"prefix":"10.1145","author":[{"given":"Karthikeyan","family":"Vaidyanathan","sequence":"first","affiliation":[{"name":"Parallel Computing Lab, Intel"}]},{"given":"Dhiraj D.","family":"Kalamkar","sequence":"additional","affiliation":[{"name":"Parallel Computing Lab, Intel"}]},{"given":"Kiran","family":"Pamnany","sequence":"additional","affiliation":[{"name":"Parallel Computing Lab, Intel"}]},{"given":"Jeff R.","family":"Hammond","sequence":"additional","affiliation":[{"name":"Parallel Computing Lab, Intel"}]},{"given":"Pavan","family":"Balaji","sequence":"additional","affiliation":[{"name":"Argonne National Laboratory"}]},{"given":"Dipankar","family":"Das","sequence":"additional","affiliation":[{"name":"Parallel Computing Lab, Intel"}]},{"given":"Jongsoo","family":"Park","sequence":"additional","affiliation":[{"name":"Parallel Computing Lab, Intel"}]},{"given":"B\u00e1lint","family":"Jo\u00f3","sequence":"additional","affiliation":[{"name":"Thomas Jefferson National Accelerator Facility"}]}],"member":"320","published-online":{"date-parts":[[2015,11,15]]},"reference":[{"key":"e_1_3_2_1_1_1","volume-title":"Cray","author":"Is Using","year":"2013","unstructured":"Using the GNI and DMAPP AP Is . Technical Report S-2446-5002 , Cray , Mar. 2013 . URL http:\/\/docs.cray.com\/books\/S-2446-5002\/S-2446-5002.pdf. Using the GNI and DMAPP APIs. Technical Report S-2446-5002, Cray, Mar. 2013. URL http:\/\/docs.cray.com\/books\/S-2446-5002\/S-2446-5002.pdf."},{"key":"e_1_3_2_1_2_1","unstructured":"Mellanox Technologies. http:\/\/www.mellanox.com.  Mellanox Technologies. http:\/\/www.mellanox.com."},{"key":"e_1_3_2_1_3_1","unstructured":"Trinity \/ NERSC-8 RFP. http:\/\/ofiwg.github.io\/libfabric\/. URL http:\/\/ofiwg.github.io\/libfabric\/.  Trinity \/ NERSC-8 RFP. http:\/\/ofiwg.github.io\/libfabric\/. URL http:\/\/ofiwg.github.io\/libfabric\/."},{"key":"e_1_3_2_1_4_1","unstructured":"Trinity \/ NERSC-8 RFP. http:\/\/www.nersc.gov\/users\/computational-systems\/cori\/nersc-8-procurement\/trinity-nersc-8-rfp\/. URL http:\/\/www.nersc.gov\/users\/computational-systems\/cori\/nersc-8-procurement\/trinity-nersc-8-rfp\/.  Trinity \/ NERSC-8 RFP. http:\/\/www.nersc.gov\/users\/computational-systems\/cori\/nersc-8-procurement\/trinity-nersc-8-rfp\/. URL http:\/\/www.nersc.gov\/users\/computational-systems\/cori\/nersc-8-procurement\/trinity-nersc-8-rfp\/."},{"key":"e_1_3_2_1_5_1","unstructured":"OSU Micro-benchmarks 4.4.1. http:\/\/mvapich.cse.ohio-state.edu\/benchmarks\/.  OSU Micro-benchmarks 4.4.1. http:\/\/mvapich.cse.ohio-state.edu\/benchmarks\/."},{"key":"e_1_3_2_1_6_1","doi-asserted-by":"publisher","DOI":"10.1145\/2688500.2688522"},{"key":"e_1_3_2_1_7_1","doi-asserted-by":"publisher","DOI":"10.1109\/SC.2010.40"},{"key":"e_1_3_2_1_8_1","doi-asserted-by":"publisher","DOI":"10.1109\/CLUSTR.2005.347068"},{"key":"e_1_3_2_1_9_1","volume-title":"Proceedings of Science, Lattice Field Theory","author":"Boyle P.","year":"2012","unstructured":"P. Boyle . The BlueGene\/Q supercomputer. Proceedings of Science, Lattice Field Theory , 2012 . P. Boyle. The BlueGene\/Q supercomputer. Proceedings of Science, Lattice Field Theory, 2012."},{"key":"e_1_3_2_1_10_1","doi-asserted-by":"publisher","DOI":"10.1109\/CONECT.2005.24"},{"key":"e_1_3_2_1_11_1","doi-asserted-by":"publisher","DOI":"10.5555\/1413370.1413396"},{"issue":"2","key":"e_1_3_2_1_12_1","first-page":"297","volume":"19","author":"Cooley J. W.","year":"1965","unstructured":"J. W. Cooley and J. W. Tukey . An Algorithm for the Machine Computation of Complex Fourier Series. MATHCOMP , 19 ( 2 ): 297 -- 301 , 1965 . J. W. Cooley and J. W. Tukey. An Algorithm for the Machine Computation of Complex Fourier Series. MATHCOMP, 19(2):297--301, 1965.","journal-title":"MATHCOMP"},{"key":"e_1_3_2_1_14_1","doi-asserted-by":"publisher","DOI":"10.5555\/1894122.1894125"},{"key":"e_1_3_2_1_15_1","doi-asserted-by":"publisher","DOI":"10.1109\/CONECT.2005.30"},{"key":"e_1_3_2_1_16_1","doi-asserted-by":"publisher","DOI":"10.1145\/2442516.2442534"},{"key":"e_1_3_2_1_17_1","first-page":"216","volume":"93","author":"Frigo M.","year":"2005","unstructured":"M. Frigo and S. G. Johnson . The Design and Implementation of FFTW. IEEEP , 93 : 216 -- 231 , 2005 . M. Frigo and S. G. Johnson. The Design and Implementation of FFTW. IEEEP, 93:216--231, 2005.","journal-title":"The Design and Implementation of FFTW. IEEEP"},{"key":"e_1_3_2_1_18_1","doi-asserted-by":"publisher","DOI":"10.1002\/cpe.1881"},{"key":"e_1_3_2_1_19_1","doi-asserted-by":"publisher","DOI":"10.6028\/jres.049.044"},{"key":"e_1_3_2_1_20_1","volume-title":"Proceedings of ISC13: International Conference for Super Computing","author":"Joo B.","year":"2013","unstructured":"B. Joo , D. D. Kalamkar , K. Vaidyanathan , M. Smelyanskiy , K. Pamnany , V. W. Lee , P. Dubey , and W. W. III1. Lattice QCD on Intel Xeon Phi . In Proceedings of ISC13: International Conference for Super Computing , 2013 . B. Joo, D. D. Kalamkar, K. Vaidyanathan, M. Smelyanskiy, K. Pamnany, V. W. Lee, P. Dubey, and W. W. III1. Lattice QCD on Intel Xeon Phi. In Proceedings of ISC13: International Conference for Super Computing, 2013."},{"key":"e_1_3_2_1_21_1","doi-asserted-by":"publisher","DOI":"10.1109\/ICPP-W.2008.34"},{"key":"e_1_3_2_1_22_1","volume-title":"Computing Research Repository (CoRR)","author":"Krizhevsky A.","year":"2014","unstructured":"A. Krizhevsky . One weird trick for parallelizing convolutional neural networks . In Computing Research Repository (CoRR) , 2014 . A. Krizhevsky. One weird trick for parallelizing convolutional neural networks. In Computing Research Repository (CoRR), 2014."},{"key":"e_1_3_2_1_23_1","doi-asserted-by":"publisher","DOI":"10.1145\/1375527.1375544"},{"key":"e_1_3_2_1_24_1","doi-asserted-by":"publisher","DOI":"10.1109\/IPDPS.2012.73"},{"key":"e_1_3_2_1_25_1","doi-asserted-by":"publisher","DOI":"10.1109\/IPDPS.2013.83"},{"key":"e_1_3_2_1_26_1","doi-asserted-by":"publisher","DOI":"10.5555\/1896300.1896315"},{"key":"e_1_3_2_1_27_1","doi-asserted-by":"publisher","DOI":"10.5555\/572719.876332"},{"key":"e_1_3_2_1_28_1","volume-title":"Proceedings of the Cray User Group Conference","author":"Pritchard H.","year":"2012","unstructured":"H. Pritchard , D. Roweth , D. Henseler , and P. Cassella . Leveraging the cray linux environment core specialization feature to realize mpi asynchronous progress on cray xe systems . In Proceedings of the Cray User Group Conference , 2012 . H. Pritchard, D. Roweth, D. Henseler, and P. Cassella. Leveraging the cray linux environment core specialization feature to realize mpi asynchronous progress on cray xe systems. In Proceedings of the Cray User Group Conference, 2012."},{"key":"e_1_3_2_1_29_1","doi-asserted-by":"publisher","DOI":"10.1007\/s11263-015-0816-y"},{"key":"e_1_3_2_1_30_1","doi-asserted-by":"publisher","DOI":"10.1109\/IPDPS.2015.35"},{"key":"e_1_3_2_1_31_1","volume-title":"Blue Gene\/P application development redbook","author":"Solution I. S. B. G.","year":"2008","unstructured":"I. S. B. G. Solution . Blue Gene\/P application development redbook , 2008 . http:\/\/www.redbooks.ibm.com\/abstracts\/sg247287.html. I. S. B. G. Solution. Blue Gene\/P application development redbook, 2008. http:\/\/www.redbooks.ibm.com\/abstracts\/sg247287.html."},{"key":"e_1_3_2_1_32_1","doi-asserted-by":"publisher","DOI":"10.5555\/2388996.2389054"},{"key":"e_1_3_2_1_33_1","doi-asserted-by":"publisher","DOI":"10.1109\/IPDPS.2014.113"},{"key":"e_1_3_2_1_34_1","doi-asserted-by":"publisher","DOI":"10.1137\/0913035"},{"key":"e_1_3_2_1_35_1","volume-title":"Computing Research Repository","author":"Wu R.","year":"2015","unstructured":"R. Wu , S. Yan , Y. Shan , Q. Dang , and G. Sun . Deep Image: Scaling up Image Recognition . In Computing Research Repository , 2015 . R. Wu, S. Yan, Y. Shan, Q. Dang, and G. Sun. Deep Image: Scaling up Image Recognition. In Computing Research Repository, 2015."}],"event":{"name":"SC15: The International Conference for High Performance Computing, Networking, Storage and Analysis","location":"Austin Texas","acronym":"SC15","sponsor":["SIGHPC ACM Special Interest Group on High Performance Computing, Special Interest Group on High Performance Computing","SIGARCH ACM Special Interest Group on Computer Architecture","IEEE-CS Computer Society"]},"container-title":["Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis"],"original-title":[],"link":[{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/2807591.2807602","content-type":"unspecified","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/2807591.2807602","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,6,18]],"date-time":"2025-06-18T05:43:42Z","timestamp":1750225422000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/2807591.2807602"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2015,11,15]]},"references-count":34,"alternative-id":["10.1145\/2807591.2807602","10.1145\/2807591"],"URL":"https:\/\/doi.org\/10.1145\/2807591.2807602","relation":{},"subject":[],"published":{"date-parts":[[2015,11,15]]},"assertion":[{"value":"2015-11-15","order":2,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}