IXPUG Annual Meeting 2015


IXPUG 2015

IXPUG 2015 is full Monday through Thursday. If you have questions or want to be on a waiting list, please email This email address is being protected from spambots. You need JavaScript enabled to view it..

September 28 - October 2, 2015

Berkeley, CA



Monday, September 28: Hands-on Tutorials

Link to Tau material

Time (PDT)TopicPresenterLocation
9:00-12:00 Real-World Example of Optimizing a Code Kernel
Remote broadcast: https://zoom.us/j/506858130
Jack Deslippe, Rebecca Hartman-Baker, Scott French, NERSC  CRT 4th Floor South

OpenMP Basics and MPI/OpenMP Scaling
Remote broadcast: https://zoom.us/j/370300202

Zhengji Zhao, NERSC; David Martin, Argonne National Lab CRT Conference Room 3101
12:00-1:30 Lunch    
1:30-5:00 Real-World Example of Optimizing a Code Kernel (Continued)
Remote broadcast: https://zoom.us/j/506858130
Jack Deslippe, Rebecca Hartman-Baker, Scott French, NERSC CRT 4th Floor South
1:30-5:00 Advanced MPI
Remote broadcast: https://zoom.us/j/370300202
Pavan Balaji, Argonne National Lab  CRT Conference Room 3101

Tuesday, September 29: Plenary day - Keynotes, General presentations

Remote broadcast: https://zoom.us/j/506858130

Time (PDT)TopicPresenterLocation
8:45 Welcome from IXPUG, NERSC and Berkeley Lab   CRT Top Floor South 
9:00 Keynote: From Program Genesis to Future Revelations: Many Core at Intel Joe Curley, Intel   CRT Top Floor South 
9:45 Keynote: Cori and the NERSC Exascale Science Application Program Sudip Dosanjh, NERSC Director  CRT Top Floor South
10:15 Break    CRT Top Floor South
10:45 Aurora and the ALCF Early Science Program Susan Coghlan, Argonne Leadership Computing Center Division Deputy Director CRT Top Floor South
11:15 Stampede and Follow-Ons at TACC Dan Stanzione, Executive Director Texas Advanced Computing Center  CRT Top Floor South
11:45 Coding for the future: Knights Landing and beyond C.J. Newburn, Intel  CRT Top Floor South
12:30 Lunch    
 2:00 IXPUG: A community collaborating on creating high-performance applications Richard Gerber,  IXPUG President, NERSC Senior Science Advisor  CRT Top Floor South
 2:15 High Performance Parallelism Pearls: An example of collaborative community engagements for application performance Joe Curley, Intel CRT Top Floor South
 2:45 A collaborative optimization case study: data pre-conditioning C.J. Newburn, Intel  CRT Top Floor South
 3:15 Break    CRT Top Floor South
 3:45 Compiler considerations for programming the Xeon Phi; LLVM and QDP-JIT Rakesh Krishnaiyer, Intel; Balint Joo, Jefferson National Accelerator Laboratory CRT Top Floor South
 4:30 Software-defined visualization using the Phi Nathan Schultz, Intel CRT Top Floor South
 5:00 The new CRT Building at Berkeley Lab Jeff Broughton, NERSC CRT Top Floor South
 5:15 IXPUG Community All-Hands Meeting  All  CRT Top Floor South
 6:00 Reception     LBNL Cafeteria
 7:00 IXPUG 2015 Dinner    LBNL Cafeteria

Wednesday, September 30: Phi Optimization Workshop

Remote broadcast: https://zoom.us/j/506858130

 9:00 Intel(R) Xeon Phi(TM) Processor codenamed Knights Landing: Architectural Overview Avinash Sodani, Lead KNL Architect, Intel

 CRT Top Floor South

 9:45 Metrics on effective vectorization and latency sensitivity Hideki Saito & Karthik Raman, Intel  CRT Top Floor South
 10:15 Break    
 10:45 Intel Performance Tools Mark Lubin, Intel  CRT Top Floor South
 11:15 Understanding Application Data Movement Characteristics using Intel’s VTune Amplifier and Software Development Emulator tools  Doug Doerfler, NERSC  CRT Top Floor South
 11:30 Lunch    
Session 1: Vectorization and data locality (subject to change!)
 1:00 Tales from the trenches Session chair: CJ Newburn, Intel CRT Top Floor South
  GROMACS vector packing Roland Schulz, UTK CRT Top Floor South
  Grid: data parallel library for QCD Peter Boyle, U Edinburgh CRT Top Floor South
  Lessons learned from the Dungeon: optimizing the XGC1 code for Intel hardware Stephane Ethier, PPPL CRT Top Floor South
  OpenCL: there and back again Matthias Noack, ZIB CRT Top Floor South
  Explicit SIMD Vectorization in VASP Florian Wende, ZIB CRT Top Floor South
  High-Performance X-Ray Scattering Data Analysis Abhinav Sarje, Berkeley CRT Top Floor South
  Barriers and Memory Bandwidth Balint Joo, Jefferson Lab CRT Top Floor South
  Numerical Weather Prediction Optimization Tom Henderson, NOAA CRT Top Floor South
  Runtime Kernel Compilation for efficient vectorisation.  Matthias Noack, ZIB  CRT Top Floor South
 3:30 Break    
Session 2: Thread and cluster scaling, heterogeneous support (subject to change!)
 4:00 Tales from the trenches  Session chair: CJ Newburn, Intel CRT Top Floor South
  The Quantum Espresso NESAP Project Taylor Barnes, LBNL CRT Top Floor South
  Improving Thread Parallelism and Asynchronous Communication in VASP Florian Wende, ZIB CRT Top Floor South
  On-Node Parallelism of a Climate Mini-Application: Comparison of threading strategies on Intel Hans Johannsen, LBNL CRT Top Floor South
  Tiling with OpenMP in block-structured, finite-volume PDE codes Brian Friesen, LBNL CRT Top Floor South
 5:00 Adjourn    

Thursday, October 1: Workshops on Libraries and Programming Models 


Programming Models for Exascale: Challenges and Opportunities

Remote broadcast: https://zoom.us/j/506858130 

Kathy Yelick, LBNL Associate Director for Computing Science CRT Top Floor South
Workshop 1: Library and Tool Readiness for KNL and Beyond (Parallel Session, see below)
Remote broadcast: https://zoom.us/j/370300202
 10:00 Break    
 10:30 Library and Application Usage at HPC Centers    CRT Conference Room 3101
 10:45 Intel math libraries  Ken Craft, Intel  CRT Conference Room 3101
 11:15 PETSc Tobin Isaac,  University of Texas  CRT Conference Room 3101
 11:45 Sparse Solvers Sherry Li, Berkeley Lab  CRT Conference Room 3101
 12:15 Lunch    
 1:30 FFT Libraries  Ken Craft, Intel  CRT Conference Room 3101
 1:45 I/O Libraries Suren Byna, Berkeley Lab; Quincey Koziol, The HDF Group  CRT Conference Room 3101
 2:15 Performance Tools Srinath Vadlaman, Paratools  CRT Conference Room 3101
 2:45 libsxmm: A Library for small matrix-matrix multiplications targeting Intel Architecture (x86) Alexander Heinecke, Intel CRT Conference Room 3101

Workshop 2: Programming Models, Languages, and Runtimes (Parallel Session, see above)
Remote broadcast: Remote broadcast: https://zoom.us/j/506858130

 10:00 Break    
 10:30 hStreams: a heterogeneous streaming library Wojciech Wasko and CJ Newburn, Intel  CRT Top Floor South
 11:00 Data container / tiling abstractions: TiDA Tan Nguyen, Berkeley Lab  CRT Top Floor South
 11:30 MPI-3 RMA (Remote Memory Access)  Jeff Hammond, Intel CRT Top Floor South
 12:00 Lunch    
 1:30 PGAS (UPC++) Yili Zheng, Berkeley Lab  CRT Top Floor South
 2:00 AGAS (HPX) Bryce Lelbach  CRT Top Floor South
End of Workshops: Community Meetings
2:30-3:00 Break    
  Community Meetings    
3:00 NESAP All-Hands   CRT Top Floor South
  Open time for ad-hoc meetings    

Friday, October 2: Community Engagements 

Confirmed meetings:

  • Density Functional Theory (DFT) for Exascale Community Workshop, CRT Top Floor South
  • Porting particle accelerator codes on the Intel Xeon Phi, CRT Conference Room 3101, 9:00-3:30 PDT, Remote broadcast: https://zoom.us/j/370300202

Density Functional Theory for Exascale Community Workshop

8:30 - 9:00 Arrivals and Breakfast

9:00 - 9:20: Jeongnim Kim - Intel - Overview of KNL and how we expect it to effect DFT applications.
9:20 - 9:40: Alvaro Vazquez - ALCF - An overview to ALCF: Current and Upcoming Many-core Technologies to Accelerate Quantum Chemistry Calculations
9:40 - 10:00: Martjin Marsman - Vienna - Preparing VASP for KNL: MPI/openMP parallelization
10:00 - 10:20:  Max Hutchinson - University of Chicago - Performance of accelerators in theory and in practice
10:20 - 10:40 Break
10:40 - 11:00 Francois Gygi - UC Davis - Large-Scale First-Principles Molecular Dynamics using the Qbox Code
11:00 - 11:20 Marco Govoni - University of Chicago - WEST: Scalable Software for Excited State Properties of Materials and Molecules
11:20 - 11:40 Andrea Ferretti - CNR - Yambo and Quantum-ESPRESSO as test-cases for code optimization towards exascale
11:40 - 12:00 Robert Distasio - Cornell - Enabling Large-Scale Hybrid Density Functional Theory Calculations in Condensed-Phase Systems
12:00 - 1:00     Lunch (Cafeteria)
1:00 - 2:00 DFT targeted Math library presentation from Intel’s Alexander Heinecke
2:00 - 2:10       Break
2:10 - 2:30 Lin Lin - UC Berkeley - DGDFT: A Massively Parallel Method for Large Scale Density Functional Theory Calculations
2:30 - 2:50 Lin Wang - Lawrence Berkeley National Lab - GPU calculation of the linear scaling three dimensional fragment method
2:50 - 3:10 Eric Bylaska - PNNL - Development Of Protocols To Identify The Solution Structure And The Incorporation Into Solids Of Environmentally Important Species (U, Zn, Al, Mg, Ca,…): Combining HPC And Synchrotron Light Sources
3:10 - 3:30 Georg Huhs - Barcelona Supercomputing Center - SIESTA - recent developments and future paths
3:30 - 4:00+ Panel/Group/Open Discussion on Common Challenges for DFT on Exascale.


Lawrence Berkeley National Laboratory

Computational Research and Theory Building (CRT)
1 Cyclotron Road, Bldg. 59
Berkeley, CA 94720
Tel: 510.486.4000

Lawrence Berkeley National Laboratory is a multidisciplinary national laboratory located in Berkeley, California on a hillside directly above the campus of the University of California at Berkeley. The site consists of 76 buildings located on 183 acres, which overlook both the campus and the San Francisco Bay. [More Visitor Information.]

Site Access

Beginning Monday, Sep. 28, call 510-387-0524 with urgent questions about site access and other logistics. 

All non-LBNL employees attending must obtain a visitor bus pass or a visitor parking pass in advance.  This applies whether you are taking the shuttle or driving.  Check-in with Security and picture ID is required.  LBNL Security requires the names, citizenship and country of birth of all visitors.  

Please fill out this form to request site access to Berkeley Lab.

The Blackberry Canyon entrance is closest to the CRT Building. The shuttle buses stop at the 65 building or main shuttle bus stop (across the street from CRT).


Parking at LBNL is limited due to a number of construction projects. We recommend car pooling, LBNL shuttle, or taxi. If you drive, a parking space may be reserved.  For those visitors who plan to drive to LBNL, we will reserve general parking near the CRT Building.

Shuttle Information

There is a Lab shuttle bus from various locations around the City of Berkeley, including near the Downtown Berkeley BART station.  If you plan to take the LBNL shuttle, you are required to have an LBNL Visitor Bus Pass  (requires advance notice) and picture ID to show the bus driver. 

Bay Area Rapid Transit Train (BART)

The Downtown Berkeley BART station is two short blocks from the LBNL shuttle line on the corner of Addison and Shuttuck.  The LBNL Blue Shuttle Bus will take you to the Blackberry Canyon entrance and the CRT Building.  The CRT building is across the street from LBNL's main bus stop. [BART Web site]


It is recommended lodging reservations be booked as early as possible. You can book a hotel anywhere in the Bay Area that has convenient access to BART and get to the meeting. 

Hotel Shattuck
2086 Allston Way, Berkeley, CA 94704
Tel: 510.845.7300
Rate: $144 (LBNL rate)
(One short block from the Downtown Berkeley BART station and LBNL shuttle stop)

Hotel Durant
2600 Durant St., Berkeley, CA 94705
Tel: 510.845.8981
Rate: $124 (LBNL rate)

Courtyard Oakland Airport
350 Hegenberger Road, Oakland, CA 94710
Tel: 510.568.7600
Rate: $149 (available rate)

LBNL Guest House (Sold out as of 8/20/2015)
1 Cyclotron Road
Berkeley, CA 94720
Rate: $124

[Additional Lodging]

Meeting Logistics Information

Norma Early
This email address is being protected from spambots. You need JavaScript enabled to view it.
Tel:  510.486.5893
On-Site Tel: 510.387.0524

Event Registration


Event Details

Event Date

Sep 28, 2015 - Oct 02, 2015