Notice

This record is currently in review state, the data hasn’t been validated yet.

Show simple item record

dc.contributor.author
Eller, Paul R.
dc.contributor.author
Hoefler, Torsten
dc.contributor.author
Gropp, William
dc.date.accessioned
2020-08-28T11:02:28Z
dc.date.available
2020-08-28T11:02:28Z
dc.date.issued
2019-01-01
dc.identifier.isbn
978-1-4503-6079-1
dc.identifier.other
10.1145/3330345.3330358
dc.identifier.uri
http://hdl.handle.net/20.500.11850/437299
dc.description.abstract
Krylov solvers are key kernels in many large-scale science and engineering applications for solving sparse linear systems. Applications running at scale can experience significant slowdown due to factors such as network congestion, off-node congestion, network distance, and performance variation across processes. Performance models can help us better understand factors limiting performance, however simple models fail to capture slowdowns often occurring at scale and performance variation across multiple runs of the same code. This work develops performance models that capture behavior found at scale and uses these models to guide optimizations for Krylov solvers and related kernels using both blocking and non-blocking communication for structured grid problems at scale. We use detailed performance analysis with network performance counters to show how network behavior relates to observed performance and guide the development of performance models that capture the runtime impact of network congestion, network distance, communication and computation overlap, and process mappings. These models guide us to optimize kernels using MPI protocol changes, node-aware communication, and topology-aware communication. The resulting tools and analysis provide us with a better understanding of how to improve performance at scale that can benefit a wider range of applications.
dc.publisher
ASSOC COMPUTING MACHINERY
dc.subject
Krylov Solvers
dc.subject
Performance Modeling
dc.subject
Performance Analysis
dc.title
Using Performance Models to Understand Scalable Krylov Solver Performance at Scale for Structured Grid Problems
dc.type
Conference Paper
ethz.journal.title
INTERNATIONAL CONFERENCE ON SUPERCOMPUTING (ICS 2019)
ethz.pages.start
138
ethz.pages.end
149
ethz.event
33rd ACM International Conference on Supercomputing (ICS)
ethz.event.location
Phoenix, AZ
ethz.event.date
JUN 26-28, 2019
ethz.identifier.wos
ethz.publication.place
NEW YORK
ethz.source
WOS
ethz.COinS
ctx_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.atitle=Using%20Performance%20Models%20to%20Understand%20Scalable%20Krylov%20Solver%20Performance%20at%20Scale%20for%20Structured%20Grid%20Problems&rft.jtitle=INTERNATIONAL%20CONFERENCE%20ON%20SUPERCOMPUTING%20(ICS%202019)&rft.date=2019-01-01&rft.spage=138&rft.epage=149&rft.au=Eller,%20Paul%20R.&Hoefler,%20Torsten&Gropp,%20William&rft.isbn=978-1-4503-6079-1&rft.genre=proceeding&rft_id=info:doi/10.1145/3330345.3330358&
 Search print copy at ETH Library

Files in this item

FilesSizeFormatOpen in viewer

There are no files associated with this item.

Publication type

Show simple item record