Location: Santa Fe, NM
Date: May 28–30, 2019
Venue: Inn and Spa at Loretto, Santa Fe, NM
Logistics information: This Workshop will be hosted by the Texas Advanced Computing Center with local coordination by the Data Science at Scale team at Los Alamos National Laboratory.
Simulation processing capability continues to outpace disk I/O capacity, making it increasingly intractable and impractical to save checkpoints or large results files for later analysis. This issue is particularly acute for Extreme Performance users, as modern Intel processors offer a high degree of parallelism and a large memory space per node, which increases both the time and opportunity cost of saving to disk. As a result, scientists must reduce data per timestep or save fewer time steps to utilize post-run analysis. Using less data than available or desired can affect a researcher’s ability to draw insights that drive discovery. To maximize opportunity and minimize ‘time to’ discovery, researchers can incorporate data and visual analysis as part of simulation runs, while the results data is still resident in memory, rather than saving time steps for later use. This event will provide HPC simulation teams the opportunity to collaborate with in situ analysis experts to insert in situ capabilities directly into their simulation codes to provide online "in situ" analysis with minimal impact to simulation execution.
This three-day event builds on successful events in 2017 and 2018. It is aimed at HPC simulation developers and domain experts with simulation implementation expertise. After an initial orientation to in situ capabilities, the workshop will combine domain expert attendees with in situ analysis experts from TACC, LANL, Kitware, Intelligent Light, and Intel, either to implement new in situ analysis capabilities or enhance and expand existing in situ analysis in target simulation codes. The workshop will provide functional analysis code on which to continue in situ development, as well as identifying barriers to in situ analysis for particular scientific domains.
Presentations will be linked in the agenda below.
Hotel: Inn and Spa at Loretto. We have secured a special rate of $179/night.
To reserve by phone, contact the Reservations Department at 1-866-582-1646. Mention either the group code "190524IXPUG" or mention the "IXPUG Group" to gain access to the blocked rooms and special group rate.
To book online, use the IXPUG Group Link
Reservation deadline is May 1st, 2019.
Transportation: We recommend that you use the Sandia Shuttle service or the Regional trains. Santa Fe is an hour drive from Albuquerque airport.
Agenda:
Tuesday, May 28
Start |
End |
Title |
8:00 a.m. |
9:00 a.m. |
Continental Breakfast |
9:00 a.m. |
9:30 a.m. |
Welcome and Opening Remarks |
9:30 a.m. |
10:00 a.m. |
Stampede2 Access Orientation |
10:00 a.m. |
11:00 a.m. |
Keynote: In Situ Analysis State of the Art Report |
11:00 a.m. |
12:00 p.m. |
Tiger Team Organization: • Sync with established TT to confirm goals for hackathon • Create impromptu TT as needed / desired |
12:00 p.m. |
2:00 p.m. |
Lunch and TT Planning Sessions |
2:00 p.m. |
5:30 p.m. |
Hackathon part 1
|
6:00 p.m. |
Networking Event – TBD |
Wednesday, May 29
Start |
End |
Title |
8:00 a.m. |
9:00 a.m. |
Continental Breakfast |
9:00 a.m. |
9:30 a.m. |
Tiger Team progress update: • Short (< 2 min) report on day 1 work from each TT • Identify and triage any issues • Adjust TT as needed for next push |
9:30 a.m. |
12:00 p.m. |
Hackathon Part 2 |
12:00 p.m. |
1:30 p.m. |
Lunch and hacking |
1:30 p.m. |
6:00 p.m. |
Hackathon part 3 |
6:00 p.m. |
Pizza Delivery and Continued Hacking |
Thursday, May 30
Questions? Email Dr. This email address is being protected from spambots. You need JavaScript enabled to view it. at TACC or Drs. This email address is being protected from spambots. You need JavaScript enabled to view it. and This email address is being protected from spambots. You need JavaScript enabled to view it. at LANL.