{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,12,19]],"date-time":"2025-12-19T07:08:23Z","timestamp":1766128103183,"version":"3.48.0"},"reference-count":43,"publisher":"IEEE","license":[{"start":{"date-parts":[[2025,11,3]],"date-time":"2025-11-03T00:00:00Z","timestamp":1762128000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2025,11,3]],"date-time":"2025-11-03T00:00:00Z","timestamp":1762128000000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2025,11,3]]},"DOI":"10.1109\/pact65351.2025.00043","type":"proceedings-article","created":{"date-parts":[[2025,12,16]],"date-time":"2025-12-16T18:30:30Z","timestamp":1765909830000},"page":"401-415","source":"Crossref","is-referenced-by-count":0,"title":["Generating Two-Level, GPU-Aware Mappings for Distributed Tensor Computations"],"prefix":"10.1109","author":[{"given":"Botao","family":"Wu","sequence":"first","affiliation":[{"name":"The Ohio State University,Dept. of Computer Science and Engineering,Columbus,OH,USA"}]},{"given":"Martin","family":"Kong","sequence":"additional","affiliation":[{"name":"The Ohio State University,Dept. of Computer Science and Engineering,Columbus,OH,USA"}]}],"member":"263","reference":[{"key":"ref1","doi-asserted-by":"publisher","DOI":"10.1103\/revmodphys.79.291"},{"key":"ref2","doi-asserted-by":"crossref","DOI":"10.1145\/3295500.3356181","article-title":"Red-blue pebbling revisited: Near optimal parallel matrix-matrix multiplication","volume-title":"Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis, ser. SC \u201819.","author":"Kwasniewski"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.1109\/IPDPS.2013.112"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1109\/ScalA49573.2019.00010"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1145\/3581784.3607096"},{"key":"ref6","doi-asserted-by":"publisher","DOI":"10.1109\/MM.2017.37"},{"article-title":"Z3 prover","year":"2022","author":"Research","key":"ref7"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.1145\/3519939.3523437"},{"journal-title":"A cellular computer to implement the Kalman filter algorithm","author":"Cannon","key":"ref9"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1002\/(SICI)1096-9128(199704)9:4<255::AID-CPE250>3.0.CO;2-2"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.1016\/0003-2697(86)90241-1"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-642-23397-5_10"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.1145\/800076.802486"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-34356-9_28"},{"key":"ref15","article-title":"Nccl 2.0","volume-title":"GPU Technology Conference (GTC)","volume":"2","author":"Jeaugey"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.1109\/TPDS.2019.2928289"},{"key":"ref17","doi-asserted-by":"publisher","DOI":"10.1109\/SC.2018.00055"},{"article-title":"Twiddling hacks","year":"2025","author":"Eron Anderson","key":"ref18"},{"journal-title":"Microsoft","article-title":"Z3py guide","year":"2025","key":"ref19"},{"article-title":"Performance reported by nccl-tests","volume-title":"NVIDIA","year":"2018","key":"ref20"},{"journal-title":"Microsoft","article-title":"Z3 arrays","year":"2025","key":"ref21"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-34356-9_28"},{"article-title":"Ascend Supercomputer","year":"2022","author":"Center","key":"ref23"},{"article-title":"Pitzer cluster","year":"2018","author":"Center","key":"ref24"},{"key":"ref25","doi-asserted-by":"publisher","DOI":"10.1109\/IPDPSW.2018.00091"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-540-78800-3_24"},{"key":"ref27","doi-asserted-by":"publisher","DOI":"10.1109\/IPDPS49936.2021.00053"},{"article-title":"Generating two-level, gpu-aware mappings for distributed tensor computations - pact\u201925 artifact","year":"2025","author":"Wu","key":"ref28"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.1109\/CGO57630.2024.10444795"},{"article-title":"Osu micro-benchmarks","year":"2024","author":"Laboratory","key":"ref30"},{"article-title":"Nvidia nsight systems","year":"2025","author":"Corporation","key":"ref31"},{"journal-title":"E. Z. S. N. S. C. (CSCS)","article-title":"Tiled-mm: Matrix multiplication on gpus for matrices stored on a cpu","year":"2025","key":"ref32"},{"key":"ref33","doi-asserted-by":"publisher","DOI":"10.5555\/3291168.3291211"},{"first-page":"863","article-title":"Ansor: Generating high-performance tensor programs for deep learning","author":"Zheng","key":"ref34"},{"key":"ref35","doi-asserted-by":"publisher","DOI":"10.1145\/3575693.3576933"},{"key":"ref36","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-78713-4_12"},{"article-title":"Communication lower bounds for tensor contraction algorithms","year":"2015","author":"Solomonik","key":"ref37"},{"key":"ref38","doi-asserted-by":"publisher","DOI":"10.1109\/IPDPS.2018.00065"},{"key":"ref39","doi-asserted-by":"publisher","DOI":"10.1145\/155090.155102"},{"key":"ref40","doi-asserted-by":"publisher","DOI":"10.1109\/ipdps.2006.1639500"},{"key":"ref41","doi-asserted-by":"publisher","DOI":"10.1145\/2503210.2503289"},{"key":"ref42","doi-asserted-by":"publisher","DOI":"10.1145\/3712285.3759869"},{"author":"Center","key":"ref43","article-title":"Ohio supercomputer center"}],"event":{"name":"2025 34th International Conference on Parallel Architectures and Compilation Techniques (PACT)","start":{"date-parts":[[2025,11,3]]},"location":"Irvine, CA, USA","end":{"date-parts":[[2025,11,6]]}},"container-title":["2025 34th International Conference on Parallel Architectures and Compilation Techniques (PACT)"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/11282903\/11282500\/11282935.pdf?arnumber=11282935","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,12,19]],"date-time":"2025-12-19T07:03:23Z","timestamp":1766127803000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11282935\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,11,3]]},"references-count":43,"URL":"https:\/\/doi.org\/10.1109\/pact65351.2025.00043","relation":{},"subject":[],"published":{"date-parts":[[2025,11,3]]}}}