1. David Gleich, Ahmed Sameh, Ananth Grama, et al.
High Performance Computing: Enabling Complex Decisions Making in the Information Era
David Gleich, Ahmed Sameh, Ananth Grama, et al.

2. Overview High Performance Computing Research@Purdue Scalable Analytics
Science of Information
Multidisciplinary Education Programs
Computational Science and Engineering
Computational Life Sciences
Data Sciences (forthcoming)

3. HPC Research@Purdue From platforms to applications and services
Purdue has among the largest academic supercomputing installations (Conte with 78K cores and close to 1.3 PF peak)
These and other clusters (Carter and others) are available to researchers on campus, as well as for other projects.

4. HPC Systems Research
Systems research focuses on programming models, language support, runtime systems, fault tolerance, performance monitoring and optimization, and power consumption.
Among the largest and most active programming languages group, worldwide, investigating issues of language support, transactions, asynchrony, etc.
Systems research spanning multiple schools focusing on issues of resource utilization, fault tolerance, application performance, and power considerations.

5. HPC Applications

6. HPC@Purdue: Simulation and Design
PRISM: Prediction of Reliability, Integrity, and Survivability of Microsystems.
An NNSA Center of the Department of Emergy

7. HPC@Purdue: Integrating Simulations and Experiments
George E Brown Jr. Network for Earthquake Engineering Simulation

8. HPC@Purdue: Simulation Hubs
Network for Computational Nanotechnology

9. HPC@Purdue: Sensing and Control
Modeling and real-time control of wind turbines.

10. HPC@Purdue: Structural Health Monitoring
Testing and simulation of large structures.

11. HPC@Purdue: Predictive Design
Purdue Advanced Reactor Core Simulator 

12. HPC@Purdue: From Simulation to Validation
Nanoscale Systems and Devices

13. HPC Algorithms and Software

14. Solvers and Preconditioners

15. Spike Performance Significantly more scalable than state of the art
Excellent FLOP counts on conventional deep memory hierarchies
Polyalgorithm that accomodates a variety of communication and processor granularities
Available as open-source software.

16. Atomistic Modeling: PuReMD
Purdue Reactive MD software is a unique open-source software package for ReaxFF simulations.
Integrates a number of novel data structures, sparse solvers, and numerical optimizations.
Available for CPU, GPU, clusters, and GPU clusters.
Is currently used at over 200 labs worldwide to simulate systems ranging from lipid bilayers to RDX.

17. Comparison with LAMMPS-Reax Qeq solver performance
Time per time-step comparison
Qeq solver performance
Memory foot-print
different QEq formulations
similar results
LAMMPS:
CG / no preconditioner

18. Bulk Water: 6540 atoms in a 40x40x40 A3 box / core
PuReMD: Weak Scaling
Bulk Water: 6540 atoms in a 40x40x40 A3 box / core

19. Si/Ge/Si Nanoscale bar Simulations
Key Result:
When Ge section is roughly square shaped, it has almost uniaxial strain!
W = nm Si Ge
average transverse Ge strain
average strains for Si&Ge in each dimension
Simple strain model derived from MD results

20. Water-Silica Interface
Motivation
a-SiO2: widely used in nano-electronic devices
also used in devices for in-vivo screening
understanding interaction with water: critical for reliability of devices
A Reactive Simulation of the Silica-Water Interface Fogarty et al., Journal of Chemical Physics 132, (2010)

21. Water-Silica Interface
Key Result
Silica surface hydroxylation as evidenced by experiments is observed.
Proposed reaction:
H2O + 2Si + O  2SiOH
Silica
Water Si O H

27. Education Programs: The Computational Science and Engineering Program
CSE is one of the top-ranked academic programs at Purdue
Initiated in 1994, it has grown significantly over the past 20 years!

28.
One of the largest programs world-wide with over 150 affiliated faculty and 160 graduate students
CSE faculty contribute in leadership roles to some of the largest research efforts on campus (DoE PRISM, CSoI NSF STC, NEES), along with many other interdisciplinary efforts
CSE faculty have been supported by two US Department of Education GAANN grants over the past eight years.
CSE students provide critical expertise to a broad set of projects

29. Education Programs: Data Science
To be initiated in Fall 2014, this program is patterned after the CSE program.
It is a multidisciplinary program that requires students to take two “bridge” courses and two disciplinary courses focused on data science.
Other requirements include seminar/ independent study, multidisciplinary PhD committee.

30. Related Centers: Center for Science of Information.
A Science and Technology Center of the National Science Foundation.

31. Thank You.