Efficient Execution of OpenMP on GPUs

Basit Öğe Kaydı
dc.authoridGeorgakoudis, Giorgis/0000-0001-6542-3555
dc.authorwosidGeorgakoudis, Giorgis/AAF-6033-2020
dc.contributor.authorHuber, Joseph
dc.contributor.authorCornelius, Melanie
dc.contributor.authorGeorgakoudis, Giorgis
dc.contributor.authorTian, Shilei
dc.contributor.authorDiaz, Jose M. Monsalve
dc.contributor.authorDinel, Kuter
dc.contributor.authorChapman, Barbara
dc.date.accessioned2023-07-26T11:50:35Z
dc.date.available2023-07-26T11:50:35Z
dc.date.issued2022
dc.departmentRektörlük, Rektörlüğe Bağlı Birimleren_US
dc.description20th IEEE/ACM International Symposium on Code Generation and Optimization (CGO) -- APR 02-06, 2022 -- Seoul, SOUTH KOREAen_US
dc.description.abstractOpenMP is the preferred choice for CPU parallelism in High-Performance-Computing (HPC) applications written in C, C++, or Fortran. As HPC systems became heterogeneous, OpenMP introduced support for accelerator offloading via the target directive. This allowed porting existing (CPU) code onto GPUs, including well established CPU parallelism paradigms. However, there are architectural differences between CPU and GPU execution which make common patterns, like forking and joining threads, single threaded execution, or sharing of local (stack) variables, in general costly on the latter. So far it was left to the user to identify and avoid non-efficient code patterns, most commonly by writing their OpenMP offloading codes in a kernel-language style which resembles CUDA more than it does traditional OpenMP. In this work we present OpenMP-aware program analyses and optimizations that allow efficient execution of the generic, CPU-centric parallelism model provided by OpenMP on GPUs. Our implementation in LLVM/Clang maps various common OpenMP patterns found in real world applications efficiently to the GPU. As static analysis is inherently limited we provide actionable and informative feedback to the user about the performed and missed optimizations, together with ways for the user to annotate the program for better results. Our extensive evaluation using several HPC proxy applications shows significantly improved GPU kernel times and reduction in resources requirements, such as GPU registers.en_US
dc.description.sponsorshipIEEE,Assoc Comp Machinery,ACM SIGPLAN,ACM SIGMICRO,IEEE Comp Soc,Arm,Meta,Huawei,Microsoft,Google,Samsung,Seoul Natl Univen_US
dc.description.sponsorshipExascale Computing Project, U.S. Department of Energy organization (Office of Science) [17-SC-20-SC]; Lawrence Livermore National Security, LLC (LLNS) via MPO [B642066]; DOE Office of Science User Facility [DE-AC05-00OR22725]; LLNL under LLNL-LDRD Program [DE-AC52-07NA27344 (LLNL-CONF-826728), 21-ERD-018]; Exascale Computing Project, U.S. Department of Energy organization (National Nuclear Security Administration) [17-SC-20-SC]en_US
dc.description.sponsorshipPart of this research was supported by the Exascale Computing Project (17-SC-20-SC), a collaborative effort of two U.S. Department of Energy organizations (Office of Science and the National Nuclear Security Administration) responsible for the planning and preparation of a capable exascale ecosystem, including software, applications, hardware, advanced system engineering, and early testbed platforms, in support of the nation's exascale computing imperative. Part of this research was supported by the Lawrence Livermore National Security, LLC (LLNS) via MPO No. B642066. We gratefully acknowledge the computing resources provided and operated by the Joint Laboratory for System Evaluation (JLSE) at Argonne National Laboratory. This research used resources of the Oak Ridge Leadership Computing Facility, which is a DOE Office of Science User Facility supported under Contract DE-AC05-00OR22725. This work was partially supported by LLNL under Contract DE-AC52-07NA27344 (LLNL-CONF-826728) through the LLNL-LDRD Program Project No. 21-ERD-018.en_US
dc.identifier.doi10.1109/CGO53902.2022.9741290
dc.identifier.endpage52en_US
dc.identifier.isbn978-1-6654-0584-3
dc.identifier.issn2164-2397
dc.identifier.scopus2-s2.0-85128418491en_US
dc.identifier.scopusqualityN/Aen_US
dc.identifier.startpage41en_US
dc.identifier.urihttps://doi.org/10.1109/CGO53902.2022.9741290
dc.identifier.urihttps://hdl.handle.net/20.500.12684/12381
dc.identifier.wosWOS:000827636600004en_US
dc.identifier.wosqualityN/Aen_US
dc.indekslendigikaynakWeb of Scienceen_US
dc.indekslendigikaynakScopusen_US
dc.institutionauthorDinel, Kuter
dc.language.isoenen_US
dc.publisherIeee Computer Socen_US
dc.relation.ispartofCgo '22: Proceedings of The 2022 Ieee/Acm International Symposium on Code Generation and Optimization (Cgo)en_US
dc.relation.publicationcategoryKonferans Öğesi - Uluslararası - Kurum Öğretim Elemanıen_US
dc.rightsinfo:eu-repo/semantics/openAccessen_US
dc.snmz$2023V1Guncelleme$en_US
dc.subjectOpenmp; Offloading; Optimization; Llvm; Gpuen_US
dc.titleEfficient Execution of OpenMP on GPUsen_US
dc.typeConference Objecten_US

Dosyalar

Orijinal paket
Listeleniyor 1 - 1 / 1
Küçük Resim
İsim:
12381.pdf
Boyut:
241.23 KB
Biçim:
Adobe Portable Document Format
Açıklama:
Tam Metin / Full Text
İndir

Koleksiyon

WoS İndeksli Yayınlar Koleksiyonu
Scopus İndeksli Yayınlar Koleksiyonu