High Performance Computing at University of Hawaii

High Performance Computing at University of Hawaii Susan T. Brown, Ph.D. Director, HPC Outreach Office of the CIO, ITS Overview What IS HPC?  The HPC Community   Across the Islands  Relationship with MHPCC Programs  Stay in touch  High Performance Computing (HPC)       What currently constitutes HPC? Clusters? Parallel? How many CPUs? Grid computing? Vector computing? The UH HPC Community Maui High Performance Computing Center (MHPCC) History in Brief MHPCC is an Air Force Distributed Center for HPC  Hires out for administrative management from University of Hawai’i through RCUH  Faculty, staff, and students access and use facility through and Educational Partnership Agreement (EPA).  MHPCC Equipment  Squall       No DoD related data or codes No DoD work to be performed on Squall 32-processor cluster SSH to use National Agency Check (NAC) not required Phasing out within a month 10 P4 nodes, 320 processors SSH to use National Agency Check (NAC) not required Came online March 2007 5,120 processor Dell PowerEdge NAC required for account Accessed via Kerberos and SecureID Number 11 on the Top 500 List Note: Other machines are located at MHPCC and are used only by the Air Force, i.e., not available for UH use.  Hurricane      Jaws     Details on equipment can be found at http://www.mhpcc.hpc.mil/doc/Doc.html UH MHPCC users  Dr. Duane Stevens, Heleakala Weather Center, WRF Modeling   Guangxia Cao, Grad Student Dr. Kevin Roe, MHPCC staff collaborator   Dr. Maqs Alam, Biochemistry, Advanced Studies in Genomics, Proteomics and Bioinformatics Dr. Neil Frazer, Oceanography  Eva-Marie Nosal, Grad Student, The Kohola Project Grad Student, Yuewei Liu Kim Kido, Undergrad Student, FEMWATER model Grad Student, Tom Fedenczuk, Analysis of Data from Mud Volcanoes!    Dr. Albert Kim, Civil and Environmental Engineering, The PAM Group    Dr. Chitteranjan Ray, Civil and Environmental Engineering Dr. Patricia Fryer, Geology & Geophysics The UH HPC Community Maui Community College  Offers ECET (Electronic and Computer Engineering Technology) Program  High Performance Computing Speciality  Trains systems operators and center support staff  Currently about 50 enrolled  18-node Dell cluster for teaching purposes (cluster sysadmin and parallel programming) The UH HPC Community Dr. Sevki Erdogan, University of Hawaii, Hilo (UHH)  System Architecture classes  Processing of Satellite Data  Huinalu  Linux Supercluster  128 nodes, 254 processors  Migrated from MHPCC  Used by Manoa researchers and Hilo students The UH HPC Community UH Manoa  Department of Meteorology  Dr. Yi-Leng Chen   Weather modeling specializing in terrain effects on islands, especially Hawaiian islands and Taiwan Has a 20-node Dell cluster, uses NCAR, local Dell cluster, is in acquisition phase for larger machine. Daily Weather Data  Dr. Vaughan Philips    Weather modeling Has 64-node Dell cluster on campus Cyclone modeling Environmental studies with IPRC  Dr. Tim Li    IPRC – International Pacific Research Center    Has Several Clusters: Cray SV1 32 processors, Altix 32 node cluster, Linux 64 nodes 4 main themes; many researchers Effects of Global Warming on Cyclones in the Pacific Electrical Engineering  Dr. David Y.Y. Yun   LIPS 240-node computer Assimilates data from MRI scans to get a total picture of the brain for analysis Mechanical Engineering  Dr. Marcelo Kobayashi   Uses Genetic Algorithm for wing design Design of Micro UAVs Long-Range Lightning Detection for TPARC N Pac Storm.mov S Businger, A Pessi, K Squires – University of Hawaii K Cummins, N Demetriades – Vaisala N Pac Storm.mov Dr. Kwok Fai Cheung, ENDEAVOR Project Ocean and Resources Engineering 3rd Generation Spectral Wave Models ENDEAVOR Website Global (WW3) Regional (WW3) Local (SWAN) Normal Tradewinds Extreme Swell Ship Motions Models Location Specific Wave Spectra Computational Chemistry John Head, Department of Chemistry, UHM Goal: use first principle quantum mechanical methods to predict the electronic structure properties of new materials Example project: design of optoelectronic devices using OH and CH3 passivated silicon clusters. Global optimization methods were developed to identify the lowest energy (most stable) structures for passivated silicon nanoclusters. Si10(OH)16 Si10(CH3)16 Theoretical Investigation of Low Energy SixLy Clusters L=H, CH3, OH and F Yukihide Shiraishi1, David Robinson2, Yingbin Ge3 and John D. Head4 1.Department of Material Science and Environmental Engineering, Tokyo University of Science, Yamaguchi, SanyoOnoda-shi, Yamaguchi, 756-0884, Japan 2. Department of Chemistry, University of Tulsa, OK. 3. Ames Laboratory, Iowa State University, IA. 4. Department of Chemistry, University of Hawaii, 2545 The Mall, Honolulu, HI 96822 Si10(CH3)16 U structure Si10(OH)16 A structure L2Si=SiL2 versus LSi-SiL3 Low energy Si10L16 clusters A B C •Unlike bulk Si, which is an indirect gap semiconductor, porous Si and Si nanoparticles exhibit a bright photoluminescence which may eventually find use in a Si based optoelectronic device [1,2]. •This poster investigates the preferred geometric arrangement for Si atoms for a Si cluster passivated by different ligands L. •Previous calculations often assume the atoms in the Six core form a bulk Si lattice like structure [3,4]. •Our previous global optimizations calculation have found the Six core to be bulk Si like in SixHy clusters with enough H atoms to fully passivate the cluster [5,6]. •However, we recently found the lowest energy SixFy clusters to be very different to the structure found for SixHy [7] •In this poster we discuss preliminary low energy structures we have found for Si2L4 and Si10L16 with L=H,CH3,OH and F. H and CH3 have similar electronegativities and were expected to have similar low energy structures. The OH electronegativity is intermediate between H and F and it is not obvious which structure would be preferred. • Geometry optimizations were performed using B3LYP/631G(d) calculations using PC GAMESS [8,9]. • Low energy Si10(CH3)16 and Si10(OH)16 clusters were generated by replacing the L atoms in the 3 lowest energy Si10H16 (U,V,W) and Si10F16 (A,B,C) structures. • Several low energy Si10H16 [6] and Si10F16 [7] structures were determined using a genetic algorithm (GA) based global optimization strategy [5]. • Calculation consistency checks were by performed using L2Si=SiL2 and LSi-SiL3 where L is only treated as a nonbridging ligand. MOTIVATION U Table of L2Si=SiL2 and LSi-SiL3 relative energies (kcal/mol) B3LYP/ B3LYP/ MP2/ 6-31G(d) LAN2DZ LAN2DZ H2Si=SiH2 0 0 0 HSi-SiH3 5.2 12.1 10.5 (CH3)2Si=Si(CH3)2 0 0 0 CH3Si-Si(CH3)3 5.3 2.5 8.6 (HO)2Si=Si(OH)2 8.7 7.9 13.1 HOSi-Si(OH)3 0 0 0 F2Si=SiF2 7.3 12.7 8.1 FSi-SiF3 0 0 0 V W Figure showing different Si10L16 cluster types where L=F. Structure U has a Si10 core like bulk Si and corresponds to the global minimum structure previously found for Si10H16 [5]. Structure A is the global minimum structure previously found for Si10F16 [7]. Table of Si10L16 ranking and relative energies (kcal/mol) Si10H16 Si10(CH3)16 Si10(OH)16 Si10F16 COMPUTATIONAL METHOD •Hydrogen and methyl favour a L2Si=SiL2 structure •Fluorine and hydroxide favour a LSi-SiL3 structure. . REFERENCES 1. L. T. Canham, Appl. Phys. Lett., 57, 1046 (1990). 2. W. L. Wilson, P. F. Szajowski, L. E. Brus, Science, 262, 1242 (1993). 3. Z. Zhou, R. A. Friesner, L. E. Brus, J. Am. Chem. Soc., 125, 15599 (2003). 4. E. Degoli et al., Phys. Rev. B, 69, 155411 (2004). 5. Y. Ge, J. D. Head, J. Phys. Chem. B, 108, 6025 (2004). 6. Y. Ge, J. D. Head, Chem. Phys. Lett., 398, 107 (2004). 7. Y. Ge, J. D. Head, Mol. Phys., 103, 1035 (2005). 8. M. W. Schmidt et al., J. Comput. Chem., 14, 1347 (1993). 9. A. A. Granovsky, PC GAMESS version 7.0, http://classic.chem.msu.su/gran/gamess/index.html A 6 13.6 • H and CH3 give U as the lowest energy cluster to presumably favor forming more readily Six cores with bulk Si like structures. • F and OH have A as the lowest energy structure preferring to form SiL3 groups over low strain Si rings. D.R. acknowledges summer support from the UH REU program funding by NSF grant CHE03-53251. We are gratefully for the generous supply of computer tiime on the University of Hawaii Dell Cluster. B 5 13.5 2 0.9 1 0.0 CONCLUSION 1 0.0 ACKNOWLEDGEMENTS 3 5 1.4 3 5.5 1.7 Geology and Geophysics  Dr. Eric Gaidos  Jillian Ward, Grad Student  Understanding gene evolution through the study of placozoans  UH/MHPCC Engagement Grant Recipient Molecular Biosciences and Bioengineering  Dr. Gernot Presting  Protein folding, Gene annotation  Uses Dell cluster and MHPCC  Aren Ewing, Engagement Grant Awardee  Phylogenetic Profiling of the Arabidopsis Genome Information & Computer Science  Dr. Henri Casanova  Virtual Compute systems  Networking  Dell 96 node cluster open for general use   Dr. Guylaine Poisson  Bioinformatics Dr. Philip Johnson  Parallel Programming Optimization  Hackystat How Many Clusters on Manoa Campus?  I don’t honestly know, but those I have found to date: Make/size Dell/96 compute nodes w/dual processors Dell/64 nodes w/dual processors Dell/16 nodes/32 processors Department/Person ICS/Henri Cassanova Meteorology/Vaughan Phillips CEE/Albert Kim IPRC/Ron Merrill 32 dual node Linux cluster, 4 Linux quads, 16 dual node linux cluster, 48 processor linux cluster, 32 processor IRIX cluster LIPS 240 Intel Celeron Processors 20 nodes, Bidding for larger machine Will have Dell 10 nodes/8 processors each – as soon as the ink dries! EE/ David Y.Y. Yun Meteorlogy/Yi-Ling Chen ORE/Kwok Fai Cheung INBRE/Steven Siegfried Meteorology/ Steven Businger Dell 20 node cluster 8 node Dell cluster, 10 Mac quadras Current Programs for Faculty/Students    Engagement Grants UH/MHPCC Summer internships DoD Mod Office Joint Opportunities for Minorities (JEOM) summer internships Find out about these by signing up for the email list uh-hpc today or going to www.hawaii.edu/hpc or emailing me at stbrown@hawaii.edu UH/MHPCC Engagement Grants        Started in 2004 Start-up; for student support only Must use MHPCC computational resources Must have UH faculty sponsor 9-month or 12-month Subject matter not restrictive Applications open this month for Sept 2007! Current Programs for Faculty/Students    Engagement Grants UH/MHPCC Summer internships DoD Mod Office Joint Opportunities for Minorities (JEOM) summer internships Find out about these by signing up for the email list uh-hpc today or going to www.hawaii.edu/hpc or emailing me at stbrown@hawaii.edu UH/MHPCC Summer Internships        Started in 2004 About 5 students each year Live in Maui; Work at MHPCC Systems, SciVis, Data Base, Security Systems Students from any UH Campus eligible Term from June 1 - July 31 Applications open in February 08 for Summer 08 Current Programs for Faculty/Students    Engagement Grants UH/MHPCC Summer internships DoD Mod Office Joint Opportunities for Minorities (JEOM) summer internships Find out about these by signing up for the email list uh-hpc today or going to www.hawaii.edu/hpc or emailing me at stbrown@hawaii.edu DoD Mod Office Joint Opportunities for Minorities (JEOM) summer internships New this year!  Located at MHPCC or on a UH campus  Graduate or Undergraduate students  Must be minority  Applications not open yet for Summer 07!  Watch for the small window of opportunity….  Current Programs for Faculty/Students    Engagement Grants UH/MHPCC Summer internships DoD Mod Office Joint Opportunities for Minorities (JEOM) summer internships Find out about these by signing up for the email list uh-hpc today or going to www.hawaii.edu/hpc or emailing me at stbrown@hawaii.edu