{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,3,15]],"date-time":"2026-03-15T23:53:24Z","timestamp":1773618804120,"version":"3.50.1"},"reference-count":32,"publisher":"IEEE","license":[{"start":{"date-parts":[[2025,6,22]],"date-time":"2025-06-22T00:00:00Z","timestamp":1750550400000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-029"},{"start":{"date-parts":[[2025,6,22]],"date-time":"2025-06-22T00:00:00Z","timestamp":1750550400000},"content-version":"stm-asf","delay-in-days":0,"URL":"https:\/\/doi.org\/10.15223\/policy-037"}],"funder":[{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"DOI":"10.13039\/100015803","name":"Tencent","doi-asserted-by":"publisher","id":[{"id":"10.13039\/100015803","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2025,6,22]]},"DOI":"10.1109\/dac63849.2025.11132627","type":"proceedings-article","created":{"date-parts":[[2025,9,15]],"date-time":"2025-09-15T17:35:41Z","timestamp":1757957741000},"page":"1-7","source":"Crossref","is-referenced-by-count":2,"title":["GoPTX: Fine-grained GPU Kernel Fusion by PTX-level Instruction Flow Weaving"],"prefix":"10.1109","author":[{"given":"Kan","family":"Wu","sequence":"first","affiliation":[{"name":"Sun Yat-sen University,Guangzhou,China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Zejia","family":"Lin","sequence":"additional","affiliation":[{"name":"Sun Yat-sen University,Guangzhou,China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Mengyue","family":"Xi","sequence":"additional","affiliation":[{"name":"Sun Yat-sen University,Guangzhou,China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Zhongchun","family":"Zheng","sequence":"additional","affiliation":[{"name":"Sun Yat-sen University,Guangzhou,China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Wenxuan","family":"Pan","sequence":"additional","affiliation":[{"name":"Sun Yat-sen University,Guangzhou,China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Xianwei","family":"Zhang","sequence":"additional","affiliation":[{"name":"Sun Yat-sen University,Guangzhou,China"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Yutong","family":"Lu","sequence":"additional","affiliation":[{"name":"Sun Yat-sen University,Guangzhou,China"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"263","reference":[{"key":"ref1","article-title":"NVIDIA Multi-Process Service (MPS)"},{"key":"ref2","article-title":"NVIDIA Multi-Instance GPU (MIG)"},{"key":"ref3","doi-asserted-by":"publisher","DOI":"10.1109\/HPCA.2018.00027"},{"key":"ref4","doi-asserted-by":"publisher","DOI":"10.1145\/2451116.2451160"},{"key":"ref5","doi-asserted-by":"publisher","DOI":"10.1145\/3620665.3640366"},{"key":"ref6","first-page":"265","article-title":"Tensorflow: a system for large-scale machine learning","volume-title":"Proceedings of the 12th USENIX Conference on Operating Systems Design and Implementation, OSDI\u201916","author":"Abadi"},{"key":"ref7","doi-asserted-by":"publisher","DOI":"10.5555\/3291168.3291211"},{"key":"ref8","doi-asserted-by":"publisher","DOI":"10.1145\/3582016.3582047"},{"key":"ref9","doi-asserted-by":"publisher","DOI":"10.1109\/SC41404.2022.00077"},{"key":"ref10","doi-asserted-by":"publisher","DOI":"10.1109\/HPCA53966.2022.00064"},{"key":"ref11","doi-asserted-by":"publisher","DOI":"10.1145\/3368826.3377918"},{"key":"ref12","doi-asserted-by":"publisher","DOI":"10.1109\/CGO57630.2024.10444869"},{"key":"ref13","doi-asserted-by":"publisher","DOI":"10.1109\/GreenCom-CPSCom.2010.102"},{"key":"ref14","doi-asserted-by":"publisher","DOI":"10.1109\/CGO53902.2022.9741270"},{"key":"ref15","first-page":"105","article-title":"gpucc: an open-source gpgpu compiler","volume-title":"Proceedings of the 2016 International Symposium on Code Generation and Optimization, CGO \u201816","author":"Wu"},{"key":"ref16","doi-asserted-by":"publisher","DOI":"10.1109\/CGO.2004.1281665"},{"key":"ref17","volume-title":"grammars-v4"},{"key":"ref18","doi-asserted-by":"publisher","DOI":"10.1109\/HPEC55821.2022.9926299"},{"key":"ref19","doi-asserted-by":"publisher","DOI":"10.1109\/IISWC.2009.5306797"},{"key":"ref20","year":"2018","journal-title":"ONNX Runtime"},{"key":"ref21","article-title":"Cuda samples"},{"key":"ref22","doi-asserted-by":"publisher","DOI":"10.1109\/CGO.2019.8661176"},{"key":"ref23","doi-asserted-by":"publisher","DOI":"10.1109\/DAC18072.2020.9218711"},{"key":"ref24","first-page":"59","article-title":"StreamBox: A lightweight GPU SandBox for serverless inference workflow","volume-title":"2024 USENIX Annual Technical Conference (USENIX ATC","volume":"24","author":"Wu"},{"key":"ref25","article-title":"Cupti 12.5 documentation","year":"2024"},{"key":"ref26","doi-asserted-by":"publisher","DOI":"10.1109\/TPDS.2013.257"},{"key":"ref27","doi-asserted-by":"publisher","DOI":"10.1109\/ICCD58817.2023.00046"},{"key":"ref28","doi-asserted-by":"publisher","DOI":"10.1145\/3007787.3001161"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.1109\/HPCA.2016.7446078"},{"key":"ref30","article-title":"Enabling task parallelism in the cuda scheduler","author":"Guevara","year":"2009"},{"key":"ref31","first-page":"10","article-title":"Fine-grained resource sharing for concurrent gpgpu kernels","volume-title":"Proceedings of the 4th USENIX Conference on Hot Topics in Parallelism, HotPar\u201912, (USA)","author":"Gregg"},{"key":"ref32","doi-asserted-by":"publisher","DOI":"10.1109\/HPCA57654.2024.00086"}],"event":{"name":"2025 62nd ACM\/IEEE Design Automation Conference (DAC)","location":"San Francisco, CA, USA","start":{"date-parts":[[2025,6,22]]},"end":{"date-parts":[[2025,6,25]]}},"container-title":["2025 62nd ACM\/IEEE Design Automation Conference (DAC)"],"original-title":[],"link":[{"URL":"http:\/\/xplorestaging.ieee.org\/ielx8\/11132383\/11132091\/11132627.pdf?arnumber=11132627","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,9,16]],"date-time":"2025-09-16T05:58:48Z","timestamp":1758002328000},"score":1,"resource":{"primary":{"URL":"https:\/\/ieeexplore.ieee.org\/document\/11132627\/"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,6,22]]},"references-count":32,"URL":"https:\/\/doi.org\/10.1109\/dac63849.2025.11132627","relation":{},"subject":[],"published":{"date-parts":[[2025,6,22]]}}}