2017 IXPUG US Annual Meeting at TACC
Experts from around the world are invited to the Texas Advanced Computing Center (TACC) in Austin, TX for the
IXPUG 2017 Annual US Meeting (IXPUG2017).
Share experiences with Xeon Phi-based systems, and learn how to optimize software for manycore machines.
The University of Texas at Austin, JJ Pickle Research Campus, Austin, TX USA
SURVEY!
Did you attend this event? Please fill out the event survey.
Keynote Speakers:
James R. Reinders
Nishanth Dandapanthu (Dell EMC)
Sameer Shende (ParaTools, Inc., U of Oregon)
Intel Roadmap/Technologies Speakers:
Barry Davis (Intel)
Day 1: Tuesday, Sept. 26, 2017 - IXPUG Annual Fall Conference
|
Tuesday zoom web link to sessions: Join from PC, Mac, Linux, iOS or Android: https://zoom.us/j/228599669 Or iPhone one-tap : |
Day 1: Tuesday, Sept. 26, 2017 - IXPUG Annual Fall Conference
| Start | End | Title | Author(s) | |
| 8:00 | 8:30 | Registration & Continental Breakfast | ||
| Opening Session [Chair: Lisa Smith] | ||||
| 8:30 | 8:35 | Welcome from TACC | Dan Stanzione | |
| 8:35 | 8:45 | Welcome from IXPUG Leadership Team | David Martin | |
| 8:45 | 9:45 | Keynote1: "Supercomputing" is the best description of the future of HPC | James Reinders (Intel Corporation, retired 2016), Parallel Programming and HPC Enthusiast (and Expert) | Video_k1 |
| 9:45 | 10:30 | Center Updates (~15 minutes each) | Video | |
| 10:30 | 11:00 | Coffee Break | ||
| Session 1 [Chair:David Martin] | ||||
| 11:00 | 12:00 |
11 Advancing MPI Libraries to the Many-core Era: Designs and Evaluations with MVAPICH2 9 Performance of PGAS Models on KNL: A Comprehensive Study with MVAPICH2-X |
Sourav Chakraborty, Mohammadreza Bayatpour, Hari Subramoni and Dhabaleswar Panda Jahanzeb Maqbool Hashmi, Mingzhe Li, Hari Subramoni and Dhabaleswar Panda |
Video |
| 12:00 | 1:00 | Lunch | ||
| Session 2 [Chair:Lars Koesterke] | ||||
| 1:00 | 2:30 |
13 Improving the Performance of the MILC Code on Intel Knights Landing, An Overview 14 Performance Portability of the Wilson Dslash Operator for Lattice QCD |
Adetokunbo Adedoyin
Douglas Doerfler, Karthik Raman and Ruizi Li
Balint Joo, Thorsten Kurth, Jack Deslippe and Kate Clark |
video4 video13 video14 |
| 2:30 | 3:00 | Coffee Break | ||
| Session 3 [Chair: Richard Gerber] | ||||
| 3:00 | 4:30 | 5 Early Results of Deep Learning on the Stampede2 Supercomputer |
Zhao Zhang, Weijia Xu, Niall Gaffney and Daniel Stanzione Sandra Philpott Kent Milfeld |
video5 video12 video21 |
| Evening | 4:30-6:00 | TACC Tour and Reception | Tommy Minyard and Melyssa Fratkin |
Day 2: Wednesday, Sept. 27, 2017 - IXPUG Annual Fall Conference
|
Wednesday zoom web link to sessions: Join from PC, Mac, Linux, iOS or Android: https://zoom.us/j/967398855 Or iPhone one-tap : |
Day 2: Wednesday, Sept. 27, 2017 - IXPUG Annual Fall Conference
| Start | End | Title | Author(s) | |
| 8:00 | 8:30 | Registration & Continental Breakfast | ||
| Session [Chair: Lisa Smith] | ||||
| 8:30 | 9:30 | Keynote: Pursuit in the simplification of HPC and Deep Learning | Nishanth Dandapanthu (Dell EMC) | video_k2 |
| 9:30 | 10:30 | Intel Compiler and Tools Updates | James Tullos (Intel) | |
| 10:30 | 11:00 | Break | ||
| Lightning Talk Session 4 [Chair: Clayton Hughes] | ||||
| 11:00 | 12:00 |
10 Reducing OS noise using offload driver on Intel Xeon Phi x200 Processor 8 A Comparative Evaluation of Xeon Phi Platforms Based on a Hodgkin-Huxley Neuron Simulator 22 Using an Interactive Tool to adapt codes to KNL 7 In Situ Visualization on Stampede2 - an IXPUG Workshop Report
Feedback on IXPUG Working Groups |
Grzegorz Andrejczuk and Jarek Kogut
George Chatzikonstantis, Diego Jimenez, Esteban Meneses, Christos Strydis, Harry Sidiropoulos and Dimitrios Soudris Lars Koesterke and Ritu Arora Paul Navratil and Jim Jeffers
|
video10 video8 video22 video7 |
| 12:00 | 1:00 | Lunch | ||
| Session 5 [Chair: John Pennycook] | ||||
| 1:00 | 2:30 | 1 MPI/OpenMP parallelization of the Hartree-Fock method for the second generation of Intel Xeon Phi processor
16 Performance optimization of WEST and Qbox on Intel Knights Landing 18 Rapid evaluation of scalar and vector fields of molecular charge density properties on KNL |
Vladimir Mironov, Yuri Alexeev, Kristopher Keipert, Michael Dmello, Alexander Moskovsky and Mark Gordon
Huihuo Zheng, Christopher Knight, Giulia Galli, Marco Govoni and Francois Gygi Alvaro Vazquez Mayagoitia, Raymundo Hernandez-Esparza and Jorge Garza |
video1 video16 video18 |
| 2:30 | 3:00 | Break | ||
| Session 6 [Chair:Kent Milfeld] | ||||
| 3:00 | 4:00 | Intel Roadmap/Technologies | Barry Davis (Intel Corporation) | |
| 4:00 | 4:45 | Panel: Where are we going next? | James Reinders, Tommy Minyard (TACC), Nishanth Dandapanthu (Dell EMC), Richard Gerber (NERSC), David Martin (ANL), Sameer Shende (ParaTools, UofOregon) |
|
| Evening | 6:00 | TEXAS BBQ Dinner at County Line On The Lake | Meet at Hyatt House for bus ride (6:00). |
Day 3: Thursday, Sept. 28, 2017 - IXPUG Annual Fall Conference
|
Thursday zoom web link to sessions: Join from PC, Mac, Linux, iOS or Android: https://zoom.us/j/967398855 Or iPhone one-tap : |
Day 3: Thursday, Sept. 28, 2017 - IXPUG Annual Fall Conference
| Start | End | Title | Author(s) | |
| 8:00 | 8:30 | Registration & Continental Breakfast | ||
| Opening Session [Chair: Lisa Smith] | ||||
| 8:30 | 9:30 | Keynote: Portable Performance Tools to Observe, Optimize, and Scale Application | Sameer Shende (ParaTools, Inc. U. of Oregon) | video_k3 |
| Session 7 [Chair: John Cazes] | ||||
| 9:30 | 10:30 |
2 Harnessing the Intel Xeon Phi x200 Processor for Earthquake Simulations |
Alexander Breuer, Yifeng Cui, Alexander Heinecke, Josh Tobin and Charles Yount Timbwaoga Ouermi, Martin Berzins and Robert M. Kirby |
video2 video20 |
| 10:30 | 10:45 | Break | ||
| Session 8 [Chair: Doug Doerfler ] --- 26min presentations | ||||
| 10:45 | 12:30 |
15 WARP3D Implementation of MKL Cluster Pardiso Solver 17 Scaling and optimization results of the real-space DFT solver PARSEC on Haswell and KNL systems |
Jeremy Nicklas, Karen Tomko, Robert Dodds and Kevin Manalo Kevin Gott, Charles Lena, Ariel Biller, Josh Neitzel, Kai-Hsin Liou, Jack Deslippe and James R Chelikowsky
Phillip Romero Ivo Kabadshow and Andreas Beckmann |
video15 video17 video6 video19 |
| 12:30 | 1:00 | Lunch | ||
| TAU Tutorial | Sameer Shende | |||
| 1:00 | 3:00 | TAU | The complex nature of HPC platforms and the application development environment, combining multiple languages, programming paradigms, hardware, and compilers, make effective performance engineering a challenging task. To meet the needs of computational scientists in performance engineering their codes, we present a tutorial with hands-on sessions on the TAU Performance System. TAU is a powerful profiling and tracing toolkit that covers multiple aspects of performance instrumentation, measurement, and analysis. After describing and demonstrating how performance data is collected using TAU’s automated instrumentation, the workshop will present ways to analyze the performance data collected and to drill down to find performance bottlenecks. Topics will cover generating performance profiles and traces with memory and system load utilization metrics, I/O, communication, and hardware performance counter data using PAPI. The workshop will cover instrumentation of Hybrid MPI and OpenMP codes on the Stampede 2 Intel Xeon Phi (KNL) platform at TACC. |
video_tau |
| KNL Tutorial ROOM ACB 1.104 | Lars Koesterke, Todd Evans, Kent Milfeld | |||
| 1:00 | 2:30 | KNL Tutorial | [Kicking the tires-- Hardware Experiences] 15 min Lect, 35min Lab LECTURE 1: Overview of the Stampede2 KNL system and programming environment. Lab1: Login on to a Stampede2 KNL compute node interactively (compile & OpenMP/MPI/hybrid execution). Hardware Experiences: run some basic benchmarks|utilities that illustrate the numa node configurations, DDR and MCDRAM speeds, and cluster modes. (Download your own code, compile and run -- compare to your own system)
|
|
| 2:30 | 2:45 | Break | ||
| 2:45 | 4:45 | KNL Tutorial (cont.) | [Many-core Affinity] 30min Lect, 30 min Lab
LECTURE3: Learn the basics about the kernel (affinity) map, and how to control affinity (the map) through environment variables.
LAB3: Evaluate and explore simple and explicit settings with the AMASK tool.
[Tools-- VTune] 30min Lecture, 30 min Lab
LECTURE4: KNL-centric hands-on experience with VTune.
LAB4: A few examples |
| Abstract Submission Deadline | Aug 6 -> Aug 13 | 2017 |
| Abstracts Reviewed by IXPUG Committee | Aug 16-22 | 2017 |
| Acceptance Notification | Aug 15 ->Aug 23 | 2017 |
| Preliminary Agenda Posted to IXPUG Website | August 25 | 2017 |
| Registration Deadline | extended Sept 25 | 2017 |
| Agenda Finalized. -- Registration Deadline -- | September 15 | 2017 |
| Final Presentations Due from Speakers | September 19 | 2017 |
| IXPUG Sessions | September 26-28 | 2017 |
Call for Presentations:
IXPUG welcomes submissions on innovative work from KNL users in academia, industry and government labs, describing original discoveries and experiences that will promote and prescribe efficient use of manycore and multicore systems. The authors of the best scored abstracts and draft presentations will be selected for a full 30 minute presentation; others may be offered an opportunity to present shorter Lightning Talks.
Submission Guidelines:
A short Extended Abstract and Draft Presentation should be submitted by Sunday, August 6th 13th. The suggested organization of the extended abstract is: short abstract summarizing the work, benefits and accomplishments; an introduction with an objective; accomplishments (analysis, optimization, algorithm/software design, tool design, experience, new concepts, etc.) with results; and a summary. An abstract submission must be a PDF file, in a convenient text format; two or three pages should be sufficient. The Draft Presentation does not need to be complete by this date. It should reflect the overall intent of the presentation and contain placeholders for the remaining content to be completed by the Final Presentation. Presentations describing application results and work on KNL-specific features (e.g. use of MCDRAM, multi-node messaging (MPI) performance and configurations, and new performance tools exploitation) will be prioritized.
For presentation format, please use the IXPUG presentation template; submit Abstract (pdf) and Presentation (pptx or pdf) through EasyChair. (Please indicate Lightning Talk at the top of your abstract if you only intend to submit a short presentation.)
Topics of interest are (but not limited to):
- Vectorization: SIMD operations and directives, data layout
- Memory: DDR/MCDRAM partitioning, memory affinity, prefetching, latency, streams, etc.
- Communication: MPI inter-/intra-node performance, scaling and tuning for Omni-Path/IB
- IO:Local disk vs global (Lustre, etc.)
- Thread and Process Management: Affinity, resource sharing in SMTs (simultaneous multi-threading) and Tiles
- System Management: Memory/Cluster Modes, Large Pages, Node Stats, XPPSL, OpenHPC
- Hybrid Computing: MPI Process/Thread Partitioning, On-node/Off-node Scaling
- Programming Models: OpenMP, TBB, Cilk, hStreams, MPI, --others
- Algorithms and Methods:Application Scaling and Vectorizable Algorithms
- Tools: Benchmarking, Profiling, Performance Analysis, Affinity
- Visualization: Software Performance, Algorithms, Methods
- Deep Learning:Application to speech, image, bioinformatics, natural language, ...
Travel and Hotel Information:
- Transportation, Parking, etc.: Information on how to get to the venue can be found here.
- Lodging: Special conference rates will be available at the Hyatt House hotel (shuttles available). Use this site for finding other accommodations.
| Hotel | Hotel Address/Phone | Conference Rate | Cut-off Date |
|
Hyatt House Arboretum |
10001 N Capital of TX Hwy Phone: 512 342-8080 |
$119 |
August 26,2017 |
IXPUG2017 Program Committee:
| Kent | Milfeld (co-Chair) | Texas Advanced Computing Center (TACC) |
| Richard | Gerber | NERSC/Lawrence Berkeley National Laboratory |
| Gilles | Civario | Dell Inc. (DELL) |
| Douglas | Doerfler | NERSC/Lawrence Berkeley National Laboratory |
| Helen | He | NERSC/Lawrence Berkeley National Laboratory |
| Clayton | Hughes | Sandia National Laboratories |
| Juha | Jaykka | University of Cambridge |
| Michael | Klemm | Intel Corporation |
| Lars | Koesterke (co-Chair) | Texas Advanced Computing Center (TACC) |
| David | Martin | Argonne National Laboratory |
| Hai Ah | Nam | Los Alamos National Laboratory |
| John | Pennycook | Intel Corporation |
| Thomas | Steinke | Zuse Institute Berlin |
| Estela | Suarez | Forschungszentrum Juelich |
| Sameer | Shende | ParaTools, Inc.,/University of Oregon |
| Jerome | Vienne | Texas Advanced Computing Center (TACC) |