Paul Hargrove
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Paul Hargrove has a total of 121 co-authors. Here are the top co-authors by number of shared publications:
- Dan Bonachea
Lawrence Berkeley National Laboratory, United States
12 shared publications Costin Iancu
Lawrence Berkeley National, Laboratory
7 shared publicationsKoushik Sen
University of California, Berkeley
4 shared publicationsAndrew Lumsdaine
Indiana University
3 shared publicationsEric Roman
3 shared publicationsJason Duell
University of California at Berkeley and Lawrence Berkeley National Laboratory
3 shared publications
Latest Publications
- Open Access
- 23 May 2025
- Published by Office of Scientific and Technical Information (OSTI)
Investigation into the Performance Benefits of Exposing Network Backpressure in UPC++ and GASNet-EX
This document is a brief summary of the research, and supporting development efforts, conducted by the project "Investigation into Improving Dynamic Adaptivity to System-Level Asynchrony in UPC++". We tested the hypothesis "The UPC++ and GASNet-EX runtimes can expose information from the network stack that enables applications to dynamically adapt to congestion, improving total throughput". We present experimental results from both a microbenchmark and an application benchmark that support this hypothesis.
- 23 May 2025
- Published by Office of Scientific and Technical Information (OSTI)
Open AccessInvestigation into the Performance Benefits of Exposing Network Backpressure in UPC++ and GASNet-EX
This document is a brief summary of the research, and supporting development efforts, conducted by the project "Investigation into Improving Dynamic Adaptivity to System-Level Asynchrony in UPC++". We tested the hypothesis "The UPC++ and GASNet-EX runtimes can expose information from the network stack that enables applications to dynamically adapt to congestion, improving total throughput". We present experimental results from both a microbenchmark and an application benchmark that support this hypothesis.
- Open Access
- 12 November 2023
- Published by Association for Computing Machinery (ACM)
- p. 1171-1184
symPACK: A GPU-Capable Fan-Out Sparse Cholesky Solver
Sparse symmetric positive definite systems of equations are ubiquitous in scientific workloads and applications. Parallel sparse Cholesky factorization is the method of choice for solving such linear systems. Therefore, the development of parallel sparse Cholesky codes that can efficiently run on today’s large-scale heterogeneous distributed-memory platforms is of vital importance. Modern supercomputers offer nodes that contain a mix of CPUs and GPUs. To fully utilize the computing power of these nodes, scientific codes must be adapted to offload expensive computations to GPUs. We present symPACK, a GPU-capable parallel sparse Cholesky solver that uses one-sided communication primitives and remote procedure calls provided by the UPC++ library. We also utilize the UPC++ “memory kinds” feature to enable efficient communication of GPU-resident data. We show that on a number of large problems, symPACK outperforms comparable state-of-the-art GPU-capable Cholesky factorization codes by up to 14x on the NERSC Perlmutter supercomputer.
- 12 November 2023
- Published by Association for Computing Machinery (ACM)
- p. 1171-1184
Open AccesssymPACK: A GPU-Capable Fan-Out Sparse Cholesky Solver
Sparse symmetric positive definite systems of equations are ubiquitous in scientific workloads and applications. Parallel sparse Cholesky factorization is the method of choice for solving such linear systems. Therefore, the development of parallel sparse Cholesky codes that can efficiently run on today’s large-scale heterogeneous distributed-memory platforms is of vital importance. Modern supercomputers offer nodes that contain a mix of CPUs and GPUs. To fully utilize the computing power of these nodes, scientific codes must be adapted to offload expensive computations to GPUs. We present symPACK, a GPU-capable parallel sparse Cholesky solver that uses one-sided communication primitives and remote procedure calls provided by the UPC++ library. We also utilize the UPC++ “memory kinds” feature to enable efficient communication of GPU-resident data. We show that on a number of large problems, symPACK outperforms comparable state-of-the-art GPU-capable Cholesky factorization codes by up to 14x on the NERSC Perlmutter supercomputer.
- 30 October 2020
- Published by Office of Scientific and Technical Information (OSTI)
UPC++ v1.0 Programmer’s Guide, Revision 2020.10.0
UPC++ is a C++11 library that provides Partitioned Global Address Space (PGAS) programming. It is designed for writing parallel programs that run efficiently and scale well on distributed-memory parallel computers. The PGAS model is single program, multiple-data (SPMD), with each separate constituent process having access to local memory as it would in C++. However, PGAS also provides access to a global address space, which is allocated in shared segments that are distributed over the processes. UPC++ provides numerous methods for accessing and using global memory. In UPC++, all operations that access remote memory are explicit, which encourages programmers to be aware of the cost of communication and data movement. Moreover, all remote-memory access operations are by default asynchronous, to enable programmers to write code that scales well even on hundreds of thousands of cores.
- 30 October 2020
- Published by Office of Scientific and Technical Information (OSTI)
UPC++ v1.0 Programmer’s Guide, Revision 2020.10.0
UPC++ is a C++11 library that provides Partitioned Global Address Space (PGAS) programming. It is designed for writing parallel programs that run efficiently and scale well on distributed-memory parallel computers. The PGAS model is single program, multiple-data (SPMD), with each separate constituent process having access to local memory as it would in C++. However, PGAS also provides access to a global address space, which is allocated in shared segments that are distributed over the processes. UPC++ provides numerous methods for accessing and using global memory. In UPC++, all operations that access remote memory are explicit, which encourages programmers to be aware of the cost of communication and data movement. Moreover, all remote-memory access operations are by default asynchronous, to enable programmers to write code that scales well even on hundreds of thousands of cores.
- 13 November 2019
- Published by Springer Nature
- p. 138-158
GASNet-EX: A High-Performance, Portable Communication Library for Exascale
- 13 November 2019
- Published by Springer Nature
- p. 138-158
GASNet-EX: A High-Performance, Portable Communication Library for Exascale
- 31 March 2018
- Published by Office of Scientific and Technical Information (OSTI)
UPC++ Programmer’s Guide, v1.0-2018.3.0
UPC++ is a C++11 library that provides Partitioned Global Address Space (PGAS) programming. It is designed for writing parallel programs that run efficiently and scale well on distributed-memory parallel computers. The PGAS model is single program, multiple-data (SPMD), with each separate thread of execution (referred to as a rank, a term borrowed from MPI) having access to local memory as it would in C++. However, PGAS also provides access to a global address space, which is allocated in shared segments that are distributed over the ranks. UPC++ provides numerous methods for accessing and using global memory. In UPC++, all operations that access remote memory are explicit, which encourages programmers to be aware of the cost of communication and data movement. Moreover, all remote-memory access operations are by default asynchronous, to enable programmers to write code that scales well even on hundreds of thousands of cores.
- 31 March 2018
- Published by Office of Scientific and Technical Information (OSTI)
UPC++ Programmer’s Guide, v1.0-2018.3.0
UPC++ is a C++11 library that provides Partitioned Global Address Space (PGAS) programming. It is designed for writing parallel programs that run efficiently and scale well on distributed-memory parallel computers. The PGAS model is single program, multiple-data (SPMD), with each separate thread of execution (referred to as a rank, a term borrowed from MPI) having access to local memory as it would in C++. However, PGAS also provides access to a global address space, which is allocated in shared segments that are distributed over the ranks. UPC++ provides numerous methods for accessing and using global memory. In UPC++, all operations that access remote memory are explicit, which encourages programmers to be aware of the cost of communication and data movement. Moreover, all remote-memory access operations are by default asynchronous, to enable programmers to write code that scales well even on hundreds of thousands of cores.
- Open Access
- 12 November 2017
- Published by Association for Computing Machinery (ACM)
- p. 1-4
The UPC++ PGAS library for Exascale Computing
- 12 November 2017
- Published by Association for Computing Machinery (ACM)
- p. 1-4
Open AccessThe UPC++ PGAS library for Exascale Computing
- 1 November 2017
- Published by Elsevier in Journal of Parallel and Distributed Computing
- Vol. 109, 350-362
DINO: Divergent node cloning for sustained redundancy in HPC
- 1 November 2017
- Published by Elsevier in Journal of Parallel and Distributed Computing
- Vol. 109, 350-362
DINO: Divergent node cloning for sustained redundancy in HPC
- 31 August 2017
- Published by Office of Scientific and Technical Information (OSTI)
GASNet Specification, v1.8.1
The U.S. Department of Energy's Office of Scientific and Technical Information
- 31 August 2017
- Published by Office of Scientific and Technical Information (OSTI)
GASNet Specification, v1.8.1
The U.S. Department of Energy's Office of Scientific and Technical Information
- Open Access
- 1 June 2016
- Published by Association for Computing Machinery (ACM)
- p. 1-13
SReplay
Replay of parallel execution is required by HPC debuggers and resilience mechanisms. Up-to-date, there is no existing deterministic replay solution for one-sided communication. The essential problem is that the readers of updated data do not have any information on which remote threads produced the updates, the conventional happens-before based ordering tracking techniques are challenging to work at scale. This paper presents SReplay, the first software tool for sub-group deterministic record and replay for one-sided communication. SReplay allows the user to specify and record the execution of a set of threads of interest (sub-group), and then deterministically replays the execution of the sub-group on a local machine without starting the remaining threads. SReplay ensures sub-group determinism using a hybrid data- and order-replay technique. SReplay maintains scalability by a combination of local logging and approximative event order tracking within sub-group. Our evaluation on deterministic and nondeterministic UPC programs shows that SReplay introduces an overhead ranging from 1.3x to 29x, when running on 1,024 cores and tracking up to 16 threads.
- 1 June 2016
- Published by Association for Computing Machinery (ACM)
- p. 1-13
Open AccessSReplay
Replay of parallel execution is required by HPC debuggers and resilience mechanisms. Up-to-date, there is no existing deterministic replay solution for one-sided communication. The essential problem is that the readers of updated data do not have any information on which remote threads produced the updates, the conventional happens-before based ordering tracking techniques are challenging to work at scale. This paper presents SReplay, the first software tool for sub-group deterministic record and replay for one-sided communication. SReplay allows the user to specify and record the execution of a set of threads of interest (sub-group), and then deterministically replays the execution of the sub-group on a local machine without starting the remaining threads. SReplay ensures sub-group determinism using a hybrid data- and order-replay technique. SReplay maintains scalability by a combination of local logging and approximative event order tracking within sub-group. Our evaluation on deterministic and nondeterministic UPC programs shows that SReplay introduces an overhead ranging from 1.3x to 29x, when running on 1,024 cores and tracking up to 16 threads.