{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,5]],"date-time":"2026-08-05T18:25:42Z","timestamp":1785954342635,"version":"3.56.0"},"publisher-location":"New York, NY, USA","reference-count":72,"publisher":"ACM","license":[{"start":{"date-parts":[[2025,3,30]],"date-time":"2025-03-30T00:00:00Z","timestamp":1743292800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100006374","name":"National Science Foundation","doi-asserted-by":"publisher","award":["2235276, 2349144, 2349143, 2349582, 2349141"],"award-info":[{"award-number":["2235276, 2349144, 2349143, 2349582, 2349141"]}],"id":[{"id":"10.13039\/501100006374","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":["dl.acm.org"],"crossmark-restriction":true},"short-container-title":[],"published-print":{"date-parts":[[2025,3,30]]},"DOI":"10.1145\/3676641.3715984","type":"proceedings-article","created":{"date-parts":[[2025,3,27]],"date-time":"2025-03-27T16:47:32Z","timestamp":1743094052000},"page":"79-94","update-policy":"https:\/\/doi.org\/10.1145\/crossmark-policy","source":"Crossref","is-referenced-by-count":9,"title":["BQSim: GPU-accelerated Batch Quantum Circuit Simulation using Decision Diagram"],"prefix":"10.1145","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-0718-1003","authenticated-orcid":false,"given":"Shui","family":"Jiang","sequence":"first","affiliation":[{"name":"The Chinese University of Hong Kong, Shatin, Hong Kong and University of Wisconsin-Madison, Madison, WI, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0008-1064-2513","authenticated-orcid":false,"given":"Yi-Hua","family":"Chung","sequence":"additional","affiliation":[{"name":"University of Wisconsin-Madison, Madison, WI, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0006-0369-5495","authenticated-orcid":false,"given":"Chih-Chun","family":"Chang","sequence":"additional","affiliation":[{"name":"University of Wisconsin-Madison, Madison, WI, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-7348-5625","authenticated-orcid":false,"given":"Tsung-Yi","family":"Ho","sequence":"additional","affiliation":[{"name":"The Chinese University of Hong Kong, Shatin, Hong Kong"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-9768-3378","authenticated-orcid":false,"given":"Tsung-Wei","family":"Huang","sequence":"additional","affiliation":[{"name":"University of Wisconsin-Madison, Madison, WI, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"320","published-online":{"date-parts":[[2025,3,30]]},"reference":[{"key":"e_1_3_2_1_1_1","unstructured":"2024. cuQuantum. https:\/\/github.com\/NVIDIA\/cuQuantum"},{"key":"e_1_3_2_1_2_1","unstructured":"2024. FlatDD. https:\/\/github.com\/IDEA-CUHK\/FlatDD"},{"key":"e_1_3_2_1_3_1","unstructured":"2024. HIP Graph. https:\/\/rocm.docs.amd.com\/projects\/HIP\/en\/docsdevelop\/howto\/hipgraph.html"},{"key":"e_1_3_2_1_4_1","unstructured":"2024. Qiskit-Aer. https:\/\/github.com\/Qiskit\/qiskit-aer"},{"key":"e_1_3_2_1_5_1","volume-title":"Coefficient of variation. Encyclopedia of research design 1, 5","author":"Abdi Herv\u00e9","year":"2010","unstructured":"Herv\u00e9 Abdi. 2010. Coefficient of variation. Encyclopedia of research design 1, 5 (2010), 169--171."},{"key":"e_1_3_2_1_6_1","volume-title":"David A Buell, et al.","author":"Arute Frank","year":"2019","unstructured":"Frank Arute, Kunal Arya, Ryan Babbush, Dave Bacon, Joseph C Bardin, Rami Barends, Rupak Biswas, Sergio Boixo, Fernando GSL Brandao, David A Buell, et al. 2019. Quantum supremacy using a programmable superconducting processor. Nature 574, 7779 (2019), 505--510."},{"key":"e_1_3_2_1_7_1","doi-asserted-by":"publisher","DOI":"10.1109\/QCE57702.2023.00119"},{"key":"e_1_3_2_1_8_1","doi-asserted-by":"publisher","DOI":"10.1145\/1654059.1654078"},{"key":"e_1_3_2_1_9_1","doi-asserted-by":"publisher","DOI":"10.1109\/DAC18072.2020.9218563"},{"key":"e_1_3_2_1_10_1","doi-asserted-by":"publisher","DOI":"10.1109\/TC.2018.2885044"},{"key":"e_1_3_2_1_11_1","volume-title":"An Experimental Study of SYCL Task Graph Parallelism for Large-Scale Machine Learning Workloads. In European Conference on Parallel Processing. Springer, 468--479","author":"Chiu Cheng-Hsiang","year":"2021","unstructured":"Cheng-Hsiang Chiu, Dian-Lun Lin, and Tsung-Wei Huang. 2021. An Experimental Study of SYCL Task Graph Parallelism for Large-Scale Machine Learning Workloads. In European Conference on Parallel Processing. Springer, 468--479."},{"key":"e_1_3_2_1_12_1","unstructured":"cuStateVec Functions NVIDIA cuQuantum 24.08.0 documentation. https:\/\/docs.nvidia.com\/cuda\/cuquantum\/latest\/custatevec\/api\/functions.html"},{"key":"e_1_3_2_1_13_1","doi-asserted-by":"publisher","DOI":"10.1109\/HPCA51647.2021.00026"},{"key":"e_1_3_2_1_14_1","doi-asserted-by":"publisher","DOI":"10.5555\/2011422.2011423"},{"key":"e_1_3_2_1_15_1","volume-title":"Quantum cryptography. Reviews of modern physics 74, 1","author":"Gisin Nicolas","year":"2002","unstructured":"Nicolas Gisin, Gr\u00e9goire Ribordy, Wolfgang Tittel, and Hugo Zbinden. 2002. Quantum cryptography. Reviews of modern physics 74, 1 (2002), 145."},{"key":"e_1_3_2_1_16_1","doi-asserted-by":"publisher","DOI":"10.22331\/q-2022-05-03-706"},{"key":"e_1_3_2_1_17_1","doi-asserted-by":"publisher","DOI":"10.5555\/3014904.3015003"},{"key":"e_1_3_2_1_18_1","unstructured":"Adam Hayes. 2024. Coefficient of Variation: Meaning and How to Use It. https:\/\/www.investopedia.com\/terms\/c\/coefficientofvariation.asp"},{"key":"e_1_3_2_1_19_1","doi-asserted-by":"publisher","DOI":"10.1038\/s42254-023-00603-1"},{"key":"e_1_3_2_1_20_1","doi-asserted-by":"publisher","DOI":"10.1109\/ASP-DAC47756.2020.9045711"},{"key":"e_1_3_2_1_21_1","doi-asserted-by":"publisher","DOI":"10.1145\/3514355"},{"key":"e_1_3_2_1_22_1","unstructured":"Hiroshi Horii and Jun Doi. 2021. Optimization of Quantum Computing Simulation with Gate Fusion. IPSJ SIG Technical Report (2021)."},{"key":"e_1_3_2_1_23_1","unstructured":"Hiroshi Horii Christopher Wood et al. 2023. Efficient techniques to gpu accelerations of multi-shot quantum computing simulations. arXiv preprint arXiv:2308.03399 (2023)."},{"key":"e_1_3_2_1_24_1","volume-title":"Quantum entanglement. Reviews of modern physics 81, 2","author":"Horodecki Ryszard","year":"2009","unstructured":"Ryszard Horodecki, Pawe\u0140 Horodecki, Micha\u0140 Horodecki, and Karol Horodecki. 2009. Quantum entanglement. Reviews of modern physics 81, 2 (2009), 865--942."},{"key":"e_1_3_2_1_25_1","doi-asserted-by":"publisher","DOI":"10.1145\/3508352.3549382"},{"key":"e_1_3_2_1_26_1","doi-asserted-by":"publisher","DOI":"10.1109\/IPDPS54959.2023.00080"},{"key":"e_1_3_2_1_27_1","volume-title":"Wong","author":"Huang T.-W.","year":"2019","unstructured":"T.-W. Huang, C.-X. Lin, Guannan Guo, and Martin D. F. Wong. 2019. Cpp-Taskflow: Fast Task-based Parallel Programming using Modern C. IEEE IPDPS, 974--983."},{"key":"e_1_3_2_1_28_1","doi-asserted-by":"publisher","DOI":"10.1109\/TPDS.2021.3104255"},{"key":"e_1_3_2_1_29_1","doi-asserted-by":"publisher","DOI":"10.1145\/3445814.3446750"},{"key":"e_1_3_2_1_30_1","doi-asserted-by":"publisher","DOI":"10.1145\/3498325"},{"key":"e_1_3_2_1_31_1","volume-title":"Gambetta","author":"Javadi-Abhari Ali","year":"2024","unstructured":"Ali Javadi-Abhari, Matthew Treinish, Kevin Krsulich, Christopher J. Wood, Jake Lishman, Julien Gacon, Simon Martiel, Paul D. Nation, Lev S. Bishop, Andrew W. Cross, Blake R. Johnson, and Jay M. Gambetta. 2024. Quantum computing with Qiskit. https:\/\/doi.org\/10. 48550\/arXiv.2405.08810 arXiv:2405.08810 [quant-ph]"},{"key":"e_1_3_2_1_32_1","doi-asserted-by":"publisher","DOI":"10.1145\/3673038.3673073"},{"key":"e_1_3_2_1_33_1","volume-title":"A co-design framework of neural networks and quantum circuits towards quantum advantage. Nature communications 12, 1","author":"Jiang Weiwen","year":"2021","unstructured":"Weiwen Jiang, Jinjun Xiong, and Yiyu Shi. 2021. A co-design framework of neural networks and quantum circuits towards quantum advantage. Nature communications 12, 1 (2021), 579."},{"key":"e_1_3_2_1_34_1","doi-asserted-by":"publisher","DOI":"10.1145\/3605573.3605631"},{"key":"e_1_3_2_1_35_1","unstructured":"Colin King. 2015. powerstat - a tool to measure power consumption. https:\/\/manpages.ubuntu.com\/manpages\/xenial\/man8\/ powerstat.8.html"},{"key":"e_1_3_2_1_36_1","unstructured":"Los Alamos National Lab. 2014. ELL format. https:\/\/www.lanl.gov\/Caesar\/node223.html"},{"key":"e_1_3_2_1_37_1","doi-asserted-by":"publisher","DOI":"10.1145\/3458817.3476169"},{"key":"e_1_3_2_1_38_1","volume-title":"Sc20: international conference for high performance computing, networking, storage and analysis","author":"Li Ang","unstructured":"Ang Li, Omer Subasi, Xiu Yang, and Sriram Krishnamoorthy. 2020. Density matrix quantum circuit simulation via the BSP machine on modern GPU clusters. In Sc20: international conference for high performance computing, networking, storage and analysis. IEEE, 1--15."},{"key":"e_1_3_2_1_39_1","doi-asserted-by":"publisher","DOI":"10.1145\/3297858.3304023"},{"key":"e_1_3_2_1_40_1","doi-asserted-by":"publisher","DOI":"10.1109\/DAC18072.2020.9218666"},{"key":"e_1_3_2_1_41_1","doi-asserted-by":"publisher","DOI":"10.1109\/ICCAD51958.2021.9643470"},{"key":"e_1_3_2_1_42_1","doi-asserted-by":"publisher","DOI":"10.1109\/HPEC43674.2020.9286218"},{"key":"e_1_3_2_1_43_1","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-030-85665-6_27"},{"key":"e_1_3_2_1_44_1","doi-asserted-by":"publisher","DOI":"10.1109\/TPDS.2021.3138856"},{"key":"e_1_3_2_1_45_1","doi-asserted-by":"publisher","DOI":"10.1109\/ICPP.2013.10"},{"key":"e_1_3_2_1_46_1","doi-asserted-by":"publisher","DOI":"10.1109\/IPDPSW.2014.106"},{"key":"e_1_3_2_1_47_1","doi-asserted-by":"publisher","DOI":"10.1145\/3624062.3624223"},{"key":"e_1_3_2_1_48_1","doi-asserted-by":"publisher","DOI":"10.1109\/TCAD.2015.2459034"},{"key":"e_1_3_2_1_49_1","unstructured":"NVIDIA. [n. d.]. nvidia-smi. https:\/\/docs.nvidia.com\/deploy\/nvidiasmi\/index.html"},{"key":"e_1_3_2_1_50_1","doi-asserted-by":"publisher","DOI":"10.1109\/JETCAS.2022.3202204"},{"key":"e_1_3_2_1_51_1","volume-title":"MQT Bench: Benchmarking Software and Design Automation Tools for Quantum Computing. Quantum","author":"Quetschlich Nils","year":"2023","unstructured":"Nils Quetschlich, Lukas Burgholzer, and Robert Wille. 2023. MQT Bench: Benchmarking Software and Design Automation Tools for Quantum Computing. Quantum (2023). MQT Bench is available at https:\/\/www.cda.cit.tum.de\/mqtbench\/."},{"key":"e_1_3_2_1_52_1","doi-asserted-by":"publisher","DOI":"10.1145\/3567955.3567958"},{"key":"e_1_3_2_1_53_1","volume-title":"Solving elliptic problems using ELLPACK","author":"Rice John R","unstructured":"John R Rice and Ronald F Boisvert. 2012. Solving elliptic problems using ELLPACK. Vol. 2. Springer Science & Business Media."},{"key":"e_1_3_2_1_54_1","doi-asserted-by":"publisher","DOI":"10.1145\/1345206.1345220"},{"key":"e_1_3_2_1_55_1","unstructured":"SixSigma. 2024. Coefficient of Variation: Mastering Relative Variability in Statistics. https:\/\/www.6sigma.us\/six-sigma-in-focus\/coefficient-of-variation\/"},{"key":"e_1_3_2_1_56_1","volume-title":"Nicolas PD Sawaya, and Al\u00e1n Aspuru-Guzik","author":"Smelyanskiy Mikhail","year":"2016","unstructured":"Mikhail Smelyanskiy, Nicolas PD Sawaya, and Al\u00e1n Aspuru-Guzik. 2016. qHiPSTER: The quantum high performance software testing environment. arXiv preprint arXiv:1601.07195 (2016)."},{"key":"e_1_3_2_1_57_1","doi-asserted-by":"publisher","DOI":"10.22331\/q-2021-10-06-559"},{"key":"e_1_3_2_1_58_1","doi-asserted-by":"publisher","DOI":"10.1145\/3624062.3624226"},{"key":"e_1_3_2_1_59_1","doi-asserted-by":"publisher","DOI":"10.1145\/3620666.3651360"},{"key":"e_1_3_2_1_60_1","doi-asserted-by":"publisher","unstructured":"Quantum AI team and collaborators. 2020. qsim. https:\/\/doi.org\/10.5281\/zenodo.4023103","DOI":"10.5281\/zenodo.4023103"},{"key":"e_1_3_2_1_61_1","unstructured":"CUDA Toolkit v12.6.2. [n. d.]. https:\/\/docs.nvidia.com\/cuda\/cudaruntime-api\/group__CUDART__GRAPH.html"},{"key":"e_1_3_2_1_62_1","doi-asserted-by":"publisher","DOI":"10.1109\/ICST49551.2021.00061"},{"key":"e_1_3_2_1_63_1","doi-asserted-by":"publisher","DOI":"10.1109\/ASE51524.2021.9678792"},{"key":"e_1_3_2_1_64_1","doi-asserted-by":"publisher","DOI":"10.1109\/ASE51524.2021.9678798"},{"key":"e_1_3_2_1_65_1","unstructured":"Wikipedia. [n. d.]. Coefficient of Variation. https:\/\/en.wikipedia.org\/wiki\/Coefficient_of_variation"},{"key":"e_1_3_2_1_66_1","doi-asserted-by":"publisher","DOI":"10.1145\/3489517.3530627"},{"key":"e_1_3_2_1_67_1","doi-asserted-by":"publisher","DOI":"10.1145\/3295500.3356155"},{"key":"e_1_3_2_1_68_1","doi-asserted-by":"publisher","DOI":"10.1145\/3447818.3460357"},{"key":"e_1_3_2_1_69_1","volume-title":"SC22: International Conference for High Performance Computing, Networking, Storage and Analysis. IEEE, 1--16","author":"Zhang Chen","year":"2022","unstructured":"Chen Zhang, Haojie Wang, Zixuan Ma, Lei Xie, Zeyu Song, and Jidong Zhai. 2022. UniQ: a unified programming model for efficient quantum circuit simulation. In SC22: International Conference for High Performance Computing, Networking, Storage and Analysis. IEEE, 1--16."},{"key":"e_1_3_2_1_70_1","doi-asserted-by":"publisher","DOI":"10.1109\/HPCA53966.2022.00059"},{"key":"e_1_3_2_1_71_1","doi-asserted-by":"publisher","DOI":"10.1109\/ICCAD45719.2019.8942057"},{"key":"e_1_3_2_1_72_1","doi-asserted-by":"publisher","DOI":"10.1109\/TCAD.2018.2834427"}],"event":{"name":"ASPLOS '25: 30th ACM International Conference on Architectural Support for Programming Languages and Operating Systems","location":"Rotterdam Netherlands","acronym":"ASPLOS '25","sponsor":["SIGPLAN ACM Special Interest Group on Programming Languages","SIGOPS ACM Special Interest Group on Operating Systems","SIGARCH ACM Special Interest Group on Computer Architecture"]},"container-title":["Proceedings of the 30th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 2"],"original-title":[],"link":[{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3676641.3715984","content-type":"unspecified","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/dl.acm.org\/doi\/pdf\/10.1145\/3676641.3715984","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,8,21]],"date-time":"2025-08-21T11:13:16Z","timestamp":1755774796000},"score":1,"resource":{"primary":{"URL":"https:\/\/dl.acm.org\/doi\/10.1145\/3676641.3715984"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,3,30]]},"references-count":72,"alternative-id":["10.1145\/3676641.3715984","10.1145\/3676641"],"URL":"https:\/\/doi.org\/10.1145\/3676641.3715984","relation":{},"subject":[],"published":{"date-parts":[[2025,3,30]]},"assertion":[{"value":"2025-03-30","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}]}}