Abstract
Profiling tools are indispensable for performance optimization of parallel codes. They allow users to understand where a code spends its time, and to focus optimization efforts on the most time consuming regions. However, two sources of complexity make them difficult to use on large-scale parallel applications. First, the code complexity of parallel applications is increasingly, and identifying the problematic regions in the code is difficult. Traditional profilers show either a flat view or a calling context tree view. Second, most tools average performance data across processes, losing per-process behavior. Diagnosing problems like load imbalance requires profiles from many processes, and manually analyzing profiles from hundreds or thousands of processes is infeasible. We introduce VIPACT, a visualization tool to identify different behaviors in multiple processes. VIPACT introduces "halo nodes" that concisely encode the distributions of runtimes from all processes, and a hybrid tree view that combines the advantages of calling context trees with those of flat profiles. We combine these with approaches such as ring charts, as well as the filtering and subselection of nodes, and we apply these techniques to a production multi-physics code.
| Original language | English (US) |
|---|---|
| Title of host publication | Proceedings of VPA 2016 |
| Subtitle of host publication | 3rd Workshop on Visual Performance Analysis - Held in conjunction with SC 2016: The International Conference for High Performance Computing, Networking, Storage and Analysis |
| Publisher | Institute of Electrical and Electronics Engineers Inc. |
| Pages | 25-28 |
| Number of pages | 4 |
| ISBN (Electronic) | 9781509052264 |
| DOIs | |
| State | Published - Jan 30 2017 |
| Event | 3rd Workshop on Visual Performance Analysis, VPA 2016 - Salt Lake City, United States Duration: Nov 18 2016 → … |
Other
| Other | 3rd Workshop on Visual Performance Analysis, VPA 2016 |
|---|---|
| Country | United States |
| City | Salt Lake City |
| Period | 11/18/16 → … |
Fingerprint
ASJC Scopus subject areas
- Modeling and Simulation
- Software
Cite this
VIPACT : A visualization interface for analyzing calling context trees. / Nguyen, Huu Tan; Wei, Lai; Bhatele, Abhinav; Gamblin, Todd; Boehme, David; Schulz, Martin; Ma, Kwan-Liu; Bremer, Peer Timo.
Proceedings of VPA 2016: 3rd Workshop on Visual Performance Analysis - Held in conjunction with SC 2016: The International Conference for High Performance Computing, Networking, Storage and Analysis. Institute of Electrical and Electronics Engineers Inc., 2017. p. 25-28 7836553.Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
}
TY - GEN
T1 - VIPACT
T2 - A visualization interface for analyzing calling context trees
AU - Nguyen, Huu Tan
AU - Wei, Lai
AU - Bhatele, Abhinav
AU - Gamblin, Todd
AU - Boehme, David
AU - Schulz, Martin
AU - Ma, Kwan-Liu
AU - Bremer, Peer Timo
PY - 2017/1/30
Y1 - 2017/1/30
N2 - Profiling tools are indispensable for performance optimization of parallel codes. They allow users to understand where a code spends its time, and to focus optimization efforts on the most time consuming regions. However, two sources of complexity make them difficult to use on large-scale parallel applications. First, the code complexity of parallel applications is increasingly, and identifying the problematic regions in the code is difficult. Traditional profilers show either a flat view or a calling context tree view. Second, most tools average performance data across processes, losing per-process behavior. Diagnosing problems like load imbalance requires profiles from many processes, and manually analyzing profiles from hundreds or thousands of processes is infeasible. We introduce VIPACT, a visualization tool to identify different behaviors in multiple processes. VIPACT introduces "halo nodes" that concisely encode the distributions of runtimes from all processes, and a hybrid tree view that combines the advantages of calling context trees with those of flat profiles. We combine these with approaches such as ring charts, as well as the filtering and subselection of nodes, and we apply these techniques to a production multi-physics code.
AB - Profiling tools are indispensable for performance optimization of parallel codes. They allow users to understand where a code spends its time, and to focus optimization efforts on the most time consuming regions. However, two sources of complexity make them difficult to use on large-scale parallel applications. First, the code complexity of parallel applications is increasingly, and identifying the problematic regions in the code is difficult. Traditional profilers show either a flat view or a calling context tree view. Second, most tools average performance data across processes, losing per-process behavior. Diagnosing problems like load imbalance requires profiles from many processes, and manually analyzing profiles from hundreds or thousands of processes is infeasible. We introduce VIPACT, a visualization tool to identify different behaviors in multiple processes. VIPACT introduces "halo nodes" that concisely encode the distributions of runtimes from all processes, and a hybrid tree view that combines the advantages of calling context trees with those of flat profiles. We combine these with approaches such as ring charts, as well as the filtering and subselection of nodes, and we apply these techniques to a production multi-physics code.
UR - http://www.scopus.com/inward/record.url?scp=85015217935&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85015217935&partnerID=8YFLogxK
U2 - 10.1109/VPA.2016.9
DO - 10.1109/VPA.2016.9
M3 - Conference contribution
SP - 25
EP - 28
BT - Proceedings of VPA 2016
PB - Institute of Electrical and Electronics Engineers Inc.
ER -