Actions

Applications Workshop

From Modelado Foundation

Revision as of 15:50, September 25, 2014 by imported>WilliamFeiereisen
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

Last edit - (9/25/14)

Date: September 30, 2014 - October 2, 2014

Location: Hillsboro, Oregon, Ronler Acres Bldg. #2, 2501 NW 229th Ave, Hillsboro, OR, click here for Google Map --> [1] We will be in the MPR room. Please come to the visitor desk and we will have badges and wifi credentials ready for you.


Purpose

Bring application scientists and computer architects together to implement physics codes in the exascale-capable Traleika Glacier programming environment. Assess progress, review plans for the remaining year of the project and Do some hands-on coding together.

The 3rd Applications Workshop comes at the 2/3 mark of the project. The TG architecture was frozen as of July 1. and is captured by version 4.10 of FSIM. OCR is at version 0.9 and runs on x86 as well as on the TG architecture as represented by FSIM. We have several full proxy applications and numerous kernels implemented on OCR, either directly to the API or through one of the higher level notations, CnC, HTA, R Stream. We now have the ability to do performance and data movement studies comparing applications in their original form with refactored versions on OCR and OCR-on-TG. This workshop will allow us to work on these assessments and begin to make statements about performance and energy projections to exascale resulting from the revolutionary software and hardware architectures of this project.

Goals

  • Show TG's part in the entire strategy to exa-scale
  • "Sync-up" the entire team on status of the hardware/software architectures and the applications
  • Define the metrics by which we will measure progress
  • Let's code ...

Agenda

As of: 9/15/14

Tuesday, September 30

  • Morning: All
    • 8:30a - 9:15a: Shekhar - Exascale Vision ... how Xstack fits with Fast Forward and Design Forward in its implementation
    • 9:15a - 12:00p: Status talks: Each group gives a update on their progress (consolidated from earlier flash talks)
      • Intel TG: (Bill Feiereisen, Dave Dunning, Romain Cledat, Ivan Ganev, Bala Seshasayee, Sanjay Chatterjee, Josh Fryman),
        • TG goals, structure, status, and Intel internal work
      • Intel Labs: (Rob Knauerhase, Tim Mattson, Rob Van Der Wijngaart)
      • UCSD: (Laura Carrington/Pietro Cicotti/Manu), PNNL: (Ellen Porter/John Feo), Reservoir Labs: (Rich Lethin/Benoit Meister/Muthu Baskaran/Preston Briggs), Rice: (Vivek Sarkar/Zoran Budimlic/Vincent Cavé/Sagnak Tasirlar/Kath Knobe), UIUC Evaluation: (Josep Torrellas/Adam Smith/Chih-Chieh Yang), UIUC HTA: (David Padua/Adam Smith/Roger Golliver, LLNL: (Aaron Landswehr/Dave Richards)
    • 12:00p - 1:00p Lunch - "How to get started with OCR" Romain
  • Afternoon: All
    • 1:00p - 5:00p: Detailed applications discussion (how-to, lessons-learned)
      • CoMD, Laura, Pietro
      • LULESH, Ellen, Kath, Roger
      • HPGMG 1, Tom H
      • HPGMG 2, Ellen, Kath, John,
      • AMR proxy development, Roger
      • Kernels, Brian Womack
  • Evening: All (optional)
    • 6:00p-8:00p Dinner off site.

Wednesday, October 1

  • Morning: All
  • Afternoon: Small groups
    • 1:00p - 4:00p Coding
      • Proxies/kernels and BYO-apps
      • Health of OCR - status & next steps
        • Performance comparison and improvements
        • New features & their priorities
        • Freezing OCR features and timeline
    • 4:00p - 5:00p Coding: Wrap-up and plans (Definition of metrics, tools, workshop #4 ... ), Shekhar, Wilf, Bill

Thursday, October 2

  • All day: All participants
    • 8:30a - 4:00p: Morning and Afternoon: Coding and Small Groups

all three days: coffee breaks mid morning and mid afternoon, lunch in the facility


Publication Acknowledgement and Disclaimer

Material presented here (including uploaded documents) that is created under the scope of work of the Traleika Glacier X-Stack project needs to include the following attribution on the first page:

Acknowledgment: This material is based upon work supported by the Department of Energy [Office of Science] under Award Number DE-SC0008717.

Disclaimer: This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

Confidential Material

No confidential material may be posted unless all parties who have access are under NDA. For safety, don't post confidential material. See general notice on home page.

Privacy Notice

As a convenience to meeting invitees, we seek to record and post the presentation and audio of these meetings. Your attendance indicates your awareness of recording and consent to that. Please refer to Intel's Privacy Notice for further information. Presenters outside Intel Corporation will be requested to sign a Personal Release form prior to the meeting and return it to Jim Stone, if the meeting is to be recorded.