This paper highlights the most significant enhancements made to PaRSEC, a scalable task-based runtime system designed for hybrid machines, during the Exascale Computing Project (ECP). The enhancements focus on expanding the capabilities of PaRSEC to address the evolving landscape of parallel computing. Notable achievements include the integration of support for three major types of accelerators (NVIDIA, AMD, and Intel GPUs), the refinement and increased flexibility of the communication subsystem, and the introduction of new programming interfaces tailored for irregular applications. Additionally, the project resulted in the development of powerful debugging and performance analysis tools aimed at assisting users in understanding and optimizing their applications. We present a comprehensive demonstration of these advancements through a series of benchmarks and applications within ECP and beyond, thereby showcasing the enhanced capabilities of PaRSEC across the diverse architectures within the ECP, providing valuable insights into the runtime system’s adaptability and performance across varied computing environments.
AbdulahSCaoQPeiY, et al. (2021) Accelerating geostatistical modeling and prediction with mixed-precision computations: a high-productivity approach with PaRSEC. IEEE Transactions on Parallel and Distributed Systems33(4): 964–976.
2.
AkbudakKLtaiefHMikhalevA, et al. (2018) Exploiting data sparsity for large-scale matrix computations. In: European conference on parallel processing, Cham, 721–734. Springer.
3.
AugonnetCThibaultSNamystR, et al. (2011) StarPU: a unified platform for task scheduling on heterogeneous multicore architectures. Concurrency and Computation: Practice and Experience23: 187–198.
4.
BauerMTreichlerSSlaughterE, et al. (2012) Legion: expressing locality and independence with logical regions. In: International conference for high performance computing, networking, storage and analysis, Salt Lake City, UT, 10–16 November 2012, 1–11.
5.
BosilcaGBouteillerADanalisA, et al. (2011a) Flexible development of dense linear algebra algorithms on massively parallel architectures with DPLASMA. In: 25th IEEE international symposium on parallel and distributed processing - workshop proceedings, Anchorage, AK, 16–20 May 2011, 1432–1441. IEEE. DOI: 10.1109/IPDPS.2011.299.
6.
BosilcaGBouteillerAHeraultT, et al. (2011b) Performance portability of a GPU enabled factorization with the DAGuE framework. In: International conference on cluster computing (CLUSTER), Austin, TX, 26–30 September 2011, 395–402. IEEE. DOI: 10.1109/CLUSTER.2011.51.
7.
BosilcaGBouteillerADanalisA, et al. (2013) Parsec: exploiting heterogeneity to enhance scalability. Computer Science and Engineering15(6): 36–45. DOI: 10.1109/MCSE.2013.98.
8.
CaoCHeraultTBosilcaG, et al. (2015) Design for a soft error resilient dynamic task-based runtime. In: IPDPS, Hyderabad, 25–29 May 2015, 765–774. IEEE. DOI: 10.1109/IPDPS.2015.81.
9.
CaoQPeiYHeraultT, et al. (2019) Performance analysis of tile low-rank cholesky factorization using parsec instrumentation tools. In: 2019 IEEE/ACM international workshop on programming and performance visualization tools (ProTools), Denver, CO, 17 November 2019, 25–32. IEEE.
10.
CaoQBosilcaGWuW, et al. (2020a) Flexible data redistribution in a task-based runtime system. In: 2020 IEEE international conference on cluster computing (CLUSTER), Kobe, 14–17 September 2020, 221–225. IEEE.
11.
CaoQPeiYAkbudakK, et al. (2020b) Extreme-scale task-based cholesky factorization toward climate and weather prediction applications. In: Proceedings of the platform for advanced scientific computing conference, Geneva, 29 June 2020- 1 July 2020, 1–11.
12.
CaoQBosilcaGLosadaN, et al. (2021a) Evaluating data redistribution in PaRSEC. IEEE Transactions on Parallel and Distributed Systems33(8): 1856–1872.
13.
CaoQPeiYAkbudakK, et al. (2021b) Leveraging PaRSEC runtime support to tackle challenging 3D data-sparse matrix problems. In: IPDPS, Portland, OR, 17–21 May 2021, 79–89. IEEE.
14.
CaoQAbdulahSAlomairyR, et al. (2022a) Reshaping geostatistical modeling and prediction for extreme-scale environmental applications. In: SC22: International conference for high performance computing, networking, storage and analysis, Dallas, TX, 13–18 November 2022, 1–12. IEEE.
15.
CaoQAlomairyRPeiY, et al. (2022b) A framework to exploit data sparsity in tile low-rank cholesky factorization. In: IPDPS, Lyon, 30 May 2022 - 03 June 2022, 414–424. IEEE.
16.
CaoQAbdulahSLtaiefH, et al. (2023) Reducing data motion and energy consumption of geospatial modeling applications using automated precision conversion. In: 2023 IEEE international conference on cluster computing (CLUSTER), Santa Fe, NM, 31 October 2023 - 03 November 2023, 330–342. IEEE.
17.
CaoQHeraultTBouteillerASJ, et al. (2024) Evaluating PaRSEC through matrix computations in scientific applications. In: Workshop on asynchronous many-task systems and applications (WAMTA), Cham.
18.
DanalisABosilcaGBouteillerA, et al. (2014) PTG: an abstraction for unhindered parallelism. In: Proceedings of the fourth Intl. WS on domain-specific languages and high-level frameworks for HPC, WOLFHPC ’14, New Orleans, LA, 17 November 2014, 21–30. IEEE. DOI: 10.1109/WOLFHPC.2014.8.
19.
DanalisAJagodeHHeraultT, et al. (2019) Software-defined events through PAPI. In: IPDPS workshops, Rio de Janeiro, 20–24 May 2019, 363–372. IEEE.
20.
De BoerAVan der SchootMSBijlH (2007) Mesh deformation based on radial basis function interpolation. Computers & Structures85: 784–795. DOI: 10.1016/j.compstruc.2007.01.013.
21.
DuranAFerrerRAyguadéE, et al. (2009) A proposal to extend the OpenMP tasking model with dependent tasks. International Journal of Parallel Programming37(3): 292–305.
22.
EschweilerDWagnerMGeimerM, et al. (2012) Open trace format 2: the next generation of scalable trace formats and support libraries. In: Applications, tools and techniques on the road to exascale computing, Amsterdam, 481–490. IOS Press.
23.
FertigA (2021) Programming With C++ 20: Concepts, Coroutines, Ranges, and More. Topeka, KS: Fertig Publications.
24.
GamblinTLeGendreMColletteMR, et al. (2015) The spack package manager: bringing order to HPC software chaos. In: Proceedings of the international conference for high performance computing, networking, storage and analysis, Austin, TX, 15–20 November 2015, 1–12.
25.
GatesMKurzakJChararaA, et al. (2019) SLATE: design of a modern distributed and accelerated linear algebra library. In: Proceedings of the international conference for high performance computing, networking, storage and analysis, Denver, CO, November 17 - 19, 2019, 1–18.
26.
HellerTKaiserHIglbergerK (2013) Application of the ParalleX execution model to stencil-based problems. Computer Science - Research and Development28(2-3): 253–261. DOI: 10.1007/s00450-012-0217-1.
27.
HeraultTRobertYBosilcaG, et al. (2019) Generic matrix multiplication for multi-GPU accelerated distributed-memory platforms over PaRSEC. In: ScalA@SC, Denver, CO, 18 November 2019, 33–41. IEEE. DOI: 10.1109/SCALA49573.2019.00010.
28.
HeraultTRobertYBosilcaG, et al. (2021) Distributed-memory multi-GPU block-sparse tensor contraction for electronic structure. In: International parallel and distributed processing symposium, Portland, OR, 17–21 May 2021, 537–546. IEEE. DOI: 10.1109/IPDPS49936.2021.00062.
29.
HerouxMA (2019) The Extreme-Scale Scientific Software Stack (E4S). Technical report. Albuquerque, NM: Sandia National Laboratory (SNL-NM).
30.
HoqueRHeraultTBosilcaG, et al. (2017) Dynamic task discovery in PaRSEC: a data-flow task-based runtime. In: ScalA@SC, Denver, CO, November 12 - 17, 2017, 6:1–6:8. ACM. DOI: 10.1145/3148226.3148233.
31.
JagodeHDanalisAAnztH, et al. (2019) PAPI software-defined events for in-depth performance analysis. The International Journal of High Performance Computing Applications33(6): 1113–1127.
32.
KowalskiKBairRBaumanNP, et al. (2021) From NWChem to NWChemEx: evolving with the computational chemistry landscape. Chemical Reviews121(8): 4962–4998.
33.
LordanFTejedorEEjarqueJ, et al. (2014) Servicess: an interoperable programming framework for the cloud. Journal of Grid Computing12(1): 67–91.
34.
MiniskarNRMonilMAHValero-LaraP, et al. (2023) Iris-dmem: efficient memory management for heterogeneous computing. In: 2023 IEEE high performance extreme computing conference (HPEC), Boston, MA, 25–29 September 2023, 1–7. IEEE.
35.
MonilMAHMiniskarNRTeranishiK, et al. (2023) Matris: multi-level math library abstraction for heterogeneity and performance portability using iris runtime. In: Proceedings of the SC’23 workshops of the international conference on high performance computing, network, storage, and analysis, 1081–1092.
36.
MorOBosilcaGSnirM (2023) Improving the scaling of an asynchronous many-task runtime with a lightweight communication engine. In: Proceedings of the 52nd international conference on parallel processing, Salt Lake City, UT, August 7 - 10, 2023, 153–162.
PeiYCaoQBosilcaG, et al. (2020) Communication avoiding 2d stencil implementations over parsec task-based runtime. In: IPDPS workshops, New Orleans, LA, 18-22 May 2020, 721–729. IEEE Computer Society.
39.
PengCCalvinJAPavosevicF, et al. (2016) Massively parallel implementation of explicitly correlated coupled-cluster singles and doubles using TiledArray framework. The Journal of Physical Chemistry A120(51): 10231–10244.
40.
PengCLewisCAWangX, et al. (2020) Massively parallel quantum chemistry: a high-performance research platform for electronic structure. The Journal of Chemical Physics153(4): 044120.
41.
SchöneRTschüterRIlscheT, et al. (2011) The VampirTrace plugin counter interface: introduction and examples. In: Euro-Par parallel processing workshops, Cham, 501–511. Springer.
42.
SchuchartJNookalaPHeraultT, et al. (2022a) Pushing the boundaries of small tasks: scalable low-overhead data-flow programming in TTG. In: International conference on cluster computing, Heidelberg, 05-08 September 2022, 117–128. IEEE. DOI: 10.1109/CLUSTER51413.2022.00026.
43.
SchuchartJNookalaPJavanmardMM, et al. (2022b) Generalized flow-graph programming using template task-graphs: initial implementation and assessment. In: IPDPS, Lyon, 30 May 2022 - 03 June 2022, 839–849. IEEE. DOI: 10.1109/IPDPS53621.2022.00086.
44.
SlaughterEWuWFuY, et al. (2020) Task bench: a parameterized benchmark for evaluating parallel runtime performance. In: SC’20, Atlanta, GA, 09-19 November 2020, 1–15. IEEE.
45.
Valero-LaraPKimJHernandezO, et al. (2021) Openmp target task: Tasking and target offloading on heterogeneous systems. In: European conference on parallel processing, Cham, 445–455. Springer.
46.
WuWBouteillerABosilcaG, et al. (2015) Hierarchical DAG scheduling for hybrid distributed systems. In: IPDPS, Hyderabad, 25–29 May 2015, 156–165. IEEE. DOI: 10.1109/IPDPS.2015.56.
