{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,3,5]],"date-time":"2026-03-05T16:04:17Z","timestamp":1772726657901,"version":"3.50.1"},"publisher-location":"New York, NY, USA","reference-count":57,"publisher":"ACM","license":[{"start":{"date-parts":[[2024,5,30]],"date-time":"2024-05-30T00:00:00Z","timestamp":1717027200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":[],"published-print":{"date-parts":[[2024,5,30]]},"DOI":"10.1145\/3650200.3656616","type":"proceedings-article","created":{"date-parts":[[2024,6,3]],"date-time":"2024-06-03T14:11:54Z","timestamp":1717423914000},"page":"413-425","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":6,"title":["SmartFuse: Reconfigurable Smart Switches to Accelerate Fused Collectives in HPC Applications"],"prefix":"10.1145","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-2893-9194","authenticated-orcid":false,"given":"Pouya","family":"Haghi","sequence":"first","affiliation":[{"name":"University of Rochester, United States of America"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-3727-2889","authenticated-orcid":false,"given":"Cheng","family":"Tan","sequence":"additional","affiliation":[{"name":"Microsoft, USA"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-5872-4464","authenticated-orcid":false,"given":"Anqi","family":"Guo","sequence":"additional","affiliation":[{"name":"Boston University, USA"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0009-0006-2039-0853","authenticated-orcid":false,"given":"Chunshu","family":"Wu","sequence":"additional","affiliation":[{"name":"University of Rochester, USA"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-6995-4775","authenticated-orcid":false,"given":"Dongfang","family":"Liu","sequence":"additional","affiliation":[{"name":"Rochester Institute of Technology, USA"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-3734-9137","authenticated-orcid":false,"given":"Ang","family":"Li","sequence":"additional","affiliation":[{"name":"PNNL, United States of America"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-5252-6600","authenticated-orcid":false,"given":"Anthony","family":"Skjellum","sequence":"additional","affiliation":[{"name":"Tennessee Tech University, USA"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3644-2922","authenticated-orcid":false,"given":"Tong","family":"Geng","sequence":"additional","affiliation":[{"name":"University of Rochester, USA"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3443-9113","authenticated-orcid":false,"given":"Martin","family":"Herbordt","sequence":"additional","affiliation":[{"name":"Boston University, United States of America"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"320","published-online":{"date-parts":[[2024,6,3]]},"reference":[{"key":"e_1_3_2_1_1_1","doi-asserted-by":"publisher","DOI":"10.1147\/rd.395.0575"},{"key":"e_1_3_2_1_2_1","doi-asserted-by":"publisher","DOI":"10.1006\/jpdc.1997.1346"},{"key":"e_1_3_2_1_3_1","unstructured":"ECP\u00a0Proxy Applications. 2023. miniFE. https: \/\/proxyapps.exascaleproject.org\/app\/minife\/."},{"key":"e_1_3_2_1_4_1","doi-asserted-by":"publisher","DOI":"10.1109\/HOTI.2016.024"},{"key":"e_1_3_2_1_5_1","doi-asserted-by":"publisher","DOI":"10.1515\/jnma-2023-0089"},{"key":"e_1_3_2_1_6_1","volume-title":"High Performance Computing","author":"Bayatpour M.","unstructured":"M. Bayatpour, N. Sarkauskas, H. Subramoni, J.\u00a0Maqbool Hashmi, and D.\u00a0K. Panda. 2021. BluesMPI: Efficient MPI Non-blocking Alltoall Offloading Designs on Modern BlueField Smart NICs. In High Performance Computing. Springer International Publishing, 18\u201337."},{"key":"e_1_3_2_1_7_1","doi-asserted-by":"publisher","DOI":"10.1109\/SC.2012.50"},{"key":"e_1_3_2_1_8_1","doi-asserted-by":"publisher","DOI":"10.1109\/MCSE.2021.3083216"},{"key":"e_1_3_2_1_9_1","doi-asserted-by":"publisher","DOI":"10.1145\/3295500.3356201"},{"key":"e_1_3_2_1_10_1","doi-asserted-by":"publisher","DOI":"10.1145\/3458817.3476178"},{"key":"e_1_3_2_1_11_1","doi-asserted-by":"publisher","unstructured":"Susan Dickey and R. Kenner. 1992. Combining switches for the NYU Ultracomputer. 521 \u2013 523. https:\/\/doi.org\/10.1109\/FMPC.1992.234864","DOI":"10.1109\/FMPC.1992.234864"},{"key":"e_1_3_2_1_12_1","unstructured":"New\u00a0Wave DV. 2024. 32-Port Programmable Switch. https:\/\/newwavedv.com\/ products\/appliances\/ 32-port-programmable-switch\/."},{"key":"e_1_3_2_1_13_1","doi-asserted-by":"publisher","DOI":"10.1109\/MCSE.2006.105"},{"key":"e_1_3_2_1_14_1","doi-asserted-by":"publisher","DOI":"10.1109\/HOTI.2009.12"},{"key":"e_1_3_2_1_15_1","doi-asserted-by":"publisher","DOI":"10.1109\/COMHPC.2016.006"},{"key":"e_1_3_2_1_16_1","doi-asserted-by":"publisher","DOI":"10.1016\/0167-8191(96)00024-5"},{"key":"e_1_3_2_1_17_1","doi-asserted-by":"publisher","DOI":"10.1109\/FPL57034.2022.00071"},{"key":"e_1_3_2_1_18_1","doi-asserted-by":"publisher","DOI":"10.1145\/3577193.3593724"},{"key":"e_1_3_2_1_19_1","doi-asserted-by":"publisher","DOI":"10.1109\/FCCM48280.2020.00028"},{"key":"e_1_3_2_1_20_1","doi-asserted-by":"publisher","DOI":"10.1109\/ICFPT51103.2020.00030"},{"key":"e_1_3_2_1_21_1","doi-asserted-by":"publisher","DOI":"10.1109\/HPEC49654.2021.9622847"},{"key":"e_1_3_2_1_22_1","doi-asserted-by":"publisher","DOI":"10.1109\/HPEC43674.2020.9286200"},{"key":"e_1_3_2_1_23_1","doi-asserted-by":"publisher","DOI":"10.1002\/cpe.6769"},{"key":"e_1_3_2_1_24_1","doi-asserted-by":"publisher","DOI":"10.1145\/3577193.3593739"},{"key":"e_1_3_2_1_25_1","volume-title":"2014 51st ACM\/EDAC\/IEEE Design Automation Conference (DAC). 1\u20136.","author":"Hamzeh Mahdi","year":"2014","unstructured":"Mahdi Hamzeh, Aviral Shrivastava, and Sarma Vrudhula. 2014. Branch-aware loop mapping on CGRAs. In 2014 51st ACM\/EDAC\/IEEE Design Automation Conference (DAC). 1\u20136."},{"key":"e_1_3_2_1_26_1","volume-title":"OCT: The Open Cloud FPGA Testbed. In 31st International Conference on Field Programmable Logic and Applications (FPL).","author":"Handagala S.","unstructured":"S. Handagala, M.C. Herbordt, and M. Leeser. 2021. OCT: The Open Cloud FPGA Testbed. In 31st International Conference on Field Programmable Logic and Applications (FPL)."},{"key":"e_1_3_2_1_27_1","volume-title":"IEEE INFOCOM 2022 - IEEE Conference on Computer Communications Workshops (INFOCOM WKSHPS). 1\u20136.","author":"Handagala S.","unstructured":"S. Handagala, M. Leeser, K. Patle, and M. Zink. 2022. Network Attached FPGAs in the Open Cloud Testbed (OCT). In IEEE INFOCOM 2022 - IEEE Conference on Computer Communications Workshops (INFOCOM WKSHPS). 1\u20136."},{"key":"e_1_3_2_1_28_1","doi-asserted-by":"publisher","DOI":"10.1109\/FPL53798.2021.00040"},{"key":"e_1_3_2_1_29_1","unstructured":"Michael\u00a0A Heroux Douglas\u00a0W Doerfler Paul\u00a0S Crozier James\u00a0M Willenbring H\u00a0Carter Edwards Alan Williams Mahesh Rajan Eric\u00a0R Keiter Heidi\u00a0K Thornquist and Robert\u00a0W Numrich. 2009. Improving Performance via Mini-applications. Technical Report SAND2009-5574. Sandia National Laboratories."},{"key":"e_1_3_2_1_30_1","doi-asserted-by":"publisher","DOI":"10.1145\/1089014.1089021"},{"key":"e_1_3_2_1_31_1","unstructured":"HYPRE. 2024. Scalable Linear Solvers and Multigrid Methods. https:\/\/computing.llnl.gov\/projects\/hypre-scalable-linear-solvers-multigrid-methods."},{"key":"e_1_3_2_1_32_1","doi-asserted-by":"publisher","DOI":"10.1109\/ISCA.2005.35"},{"key":"e_1_3_2_1_33_1","doi-asserted-by":"publisher","DOI":"10.1109\/HOTI51249.2020.00018"},{"key":"e_1_3_2_1_34_1","doi-asserted-by":"publisher","DOI":"10.1145\/3295500.3356181"},{"key":"e_1_3_2_1_35_1","doi-asserted-by":"publisher","DOI":"10.5555\/977395.977673"},{"key":"e_1_3_2_1_36_1","doi-asserted-by":"publisher","DOI":"10.1145\/3307650.3322259"},{"key":"e_1_3_2_1_37_1","series-title":"SIAM Journal on Scientific Computing 35 (01","volume-title":"Efficient implementation of finite element methods on non-matching and overlapping meshes in 3D","author":"Massing Andre","year":"2013","unstructured":"Andre Massing, Mats Larson, and Anders Logg. 2013. Efficient implementation of finite element methods on non-matching and overlapping meshes in 3D. SIAM Journal on Scientific Computing 35 (01 2013), C23\u2013C47."},{"key":"e_1_3_2_1_38_1","doi-asserted-by":"publisher","DOI":"10.1109\/DATE.2003.1253623"},{"key":"e_1_3_2_1_39_1","unstructured":"Arista Networks. 2024. 7130 FPGA-enabled Network Switches - Quick Look. www.arista.com\/ en\/products\/ 7130-fpga-enabled-network-switches-quick-look."},{"key":"e_1_3_2_1_40_1","unstructured":"NPB. 2023. NAS Parallel Benchmarks. https:\/\/www.nas.nasa.gov\/software\/npb.html."},{"key":"e_1_3_2_1_41_1","unstructured":"L. Page S. Brin R. Motwani and T. Winograd. 1999. The PageRank Citation Ranking: Bringing Order to the Web.TR 1999-66. Stanford InfoLab. http:\/\/ilpubs.stanford.edu:8090\/422\/"},{"key":"e_1_3_2_1_42_1","doi-asserted-by":"publisher","DOI":"10.1145\/2807591.2807603"},{"key":"e_1_3_2_1_43_1","doi-asserted-by":"publisher","DOI":"10.1145\/3079856.3080256"},{"key":"e_1_3_2_1_44_1","doi-asserted-by":"publisher","DOI":"10.1137\/0907058"},{"key":"e_1_3_2_1_45_1","doi-asserted-by":"publisher","DOI":"10.1145\/1862648.1862652"},{"key":"e_1_3_2_1_46_1","doi-asserted-by":"publisher","DOI":"10.1145\/3295500.3356153"},{"key":"e_1_3_2_1_47_1","doi-asserted-by":"publisher","unstructured":"A. Skjellum and et al. 2020. ExaMPI: A Modern Design and Implementation to Accelerate Message Passing Interface Innovation. Communications in Computer and Information Science 1087 CCIS (2020) 153\u2013169. https:\/\/doi.org\/10.1007\/978-3-030-41005-6_11","DOI":"10.1007\/978-3-030-41005-6_11"},{"key":"e_1_3_2_1_48_1","doi-asserted-by":"publisher","DOI":"10.1145\/3093338.3093385"},{"key":"e_1_3_2_1_49_1","doi-asserted-by":"publisher","DOI":"10.1145\/3503222.3507726"},{"key":"e_1_3_2_1_50_1","doi-asserted-by":"publisher","DOI":"10.1177\/1094342005051521"},{"key":"e_1_3_2_1_51_1","unstructured":"Texas\u00a0A&M University. 2024. SuiteSparse Matrix Collection. https:\/\/sparse.tamu.edu\/."},{"key":"e_1_3_2_1_52_1","doi-asserted-by":"publisher","DOI":"10.1002\/(SICI)1096-9128(199704)9:4<255::AID-CPE250>3.0.CO;2-2"},{"key":"e_1_3_2_1_53_1","doi-asserted-by":"publisher","DOI":"10.1109\/HiPC.2015.50"},{"key":"e_1_3_2_1_54_1","doi-asserted-by":"publisher","DOI":"10.1109\/HPCA47549.2020.00063"},{"key":"e_1_3_2_1_55_1","unstructured":"Xilinx. 2023. XUP Vitis Network Example (VNx). https:\/\/github.com\/Xilinx\/xup_vitis_network_example."},{"key":"e_1_3_2_1_56_1","doi-asserted-by":"publisher","DOI":"10.1109\/FCCM48280.2020.00046"},{"key":"e_1_3_2_1_57_1","doi-asserted-by":"publisher","DOI":"10.1109\/IPDPSW.2017.65"}],"event":{"name":"ICS '24: 2024 International Conference on Supercomputing","location":"Kyoto Japan","acronym":"ICS '24","sponsor":["SIGARCH ACM Special Interest Group on Computer Architecture"]},"container-title":["Proceedings of the 38th ACM International Conference on Supercomputing"],"original-title":[],"link":[{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3650200.3656616","content-type":"unspecified","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/3650200.3656616","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,8,22]],"date-time":"2025-08-22T15:24:51Z","timestamp":1755876291000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3650200.3656616"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,5,30]]},"references-count":57,"alternative-id":["10.1145\/3650200.3656616","10.1145\/3650200"],"URL":"https:\/\/doi.org\/10.1145\/3650200.3656616","relation":{},"subject":[],"published":{"date-parts":[[2024,5,30]]},"assertion":[{"value":"2024-06-03","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}