HPC Projects in Thailand

1
HPC Projects in Thailand

• Putchong Uthayopas

2
Grid and HPC History

• 1994 NECTEC put Thailand's first vectorized
  multiprocessor supercomputer, the Cray EL98
• Computational Chemistry
• Computational Physics
• 1996 Thai Meteorological Department (TMD) use
  IBMSP
• NWP (Numerical Weather Prediction) project
• 1997 the Computational Science and Engineering
  Programme Consortium was established as a
  cooperative effort among Thai universities and
  the National Science and Technology Development
  Agency (NSTDA).
• 1999 Kasetsart University built 72 nodes Beowulf
  Cluster Pirun
• Academic and research applicaitons
• Start of the booming of Beowulf Cluster in
  Thailand
• 2001 Thailand transition into Grid Era by the
  start of ThaiGrid Project between KMITNB and KU
  later many universities join AIT, KMUTT,
  Burapa, Chula

3
Grid Computing Development2000-2005
2000
2001
2002
2003
2004
2005
ThaiGrid project Start
National Grid project approved
PRAGMA participation by NECTEC/KU
Virtual Screening On Grid
Linking KMITNB-KU
National Grid project started
ThaiGrid link more than 100 processors from 5
universities
Forming National Grid project proposal
ThaiGrid has 5 members
ThaiGrid Join SCGlobal2004
4
Grid Computing Development2006-2008
2006
2007
2008
2009
2010
2011
10TF cluster at BIOTEC online
ThaiGrid has 22 members
SILA Cluster (TERAFLOP) on line
ThaiGrid reach 1000 processors core
Thailand has a First SC exhibit at RENO SC2007
Tera Server The First Teraflop machine went
on-line given 2.5 Rmax and 4.5 Rpeak on HPL
benchmark
ThaiGrid forming the first function globus grid
of 14 universities
Thai National grid Center start working in
temporary office
5
Introduction to ThaiGrid

• A National Project under Software Industry
  Promotion Agency (Public Organization) , Ministry
  of Information and Communication Technology
• Started in 2005 from 14 member organizations
• Expanded to 22 organizations in 2008

ThaiGrid an Enabling Platform for Innovation
6
Thai Grid Infrastructure
19 sites About 1000 CPU core.
1 Gbps
1 Gbps
2.5 Gbps
155 Mbps
2.5 Gbps
1 Gbps
1 Gbps
7
TERA Cluster

• 1 Frontend (HP ProLiant DL360 G5 Server)? and 192
  computer nodes
• Intel Xeon 3.2 GHz (Dual core, Dual processor)?
• Memory 4 GB (8GB for Frontend infiniband
  nodes)?
• 70x4 GB SCSI HDD (RAID1)?
• 4 Storage Servers
• Lustre file system for TERA cluster's storage
• Attached with Smart Array P400i Controller for
  5TB space

8
ThaiGrid Usage

• ThaiGrid provides about 290 years of computing
  time for members
• 9 years on the grid
• 280 years on tera
• 41 projects from 8 areas are being support on
  Teraflop machine
• More small projects on each machines

9
Medicinal Herb Research

• Partner
• Cheminormetics Center, Kasetsart Univesity (Chak
  Sangma and team)
• Objective
• Using 3D-molecular databse and virtual screening
  to verify the traditional medicinal herb
• Benefit
• Scientific proof of the ancient traditional drug
• Benefit poor people that still rely on the drug
  from medicinal herb
• Potential benefit for local pharmaceutical
  industry

Virtual Screening Infrastructure
Lab Test
10
NanoGrid
Computing Resources
Computing Resources
AccelRys Materials Studio
2
MS-Gateway
Materials Studio Gateway
3
ThaiGrid
1
Cluster Scheduler
Cluster Scheduler
MS-Gateway
Grid Infrastructure

• Objective
• Platform that support computational Nano science
  research
• Technology used
• AccelRys Materials Studio
• Cluster Scheduler Sun Grid Engine and Torque

11
Potential Area
Initiate more projects with clear focus and more
impact
12
ThaiGrid Achievement

• The building of a community of 22 universities.
• Trained more than 2000 people to workforce during
  3 years
• Initiate 2 national competition
• train 500-600 high school students on grid/mpi
• 73 teams, 200 people from 20 schools compete on
  coding MPI to solve N body simulation using MPI
• Winning team make application 10 times faster on
  16 nodes cluster
• One Thai book on using Taverna for bioinformatics
• A lot of seminars, local conference, activities,
  small research project

13
SILA Cluster Ramkhamhaeng University

• Second Teraflop Machine (1TF)
• Usage General research
• System
• Management    Node  (5 )    Sun Fire
  X4200        
• AMD64 Opteron 2.4 GHz
• Two Dual core CPU
•  Memory Ecc Registered 8 GB
•  HardDisk 73 GB
• File Server Node    (2)         Sun Fire X4200
• AMD64 Opteron 2.4 GHz
• Two Dual core CPU
• Memory Ecc Registered 8 GB
• HardDisk 73 GB 2 units
• Computing Node    (16)       Sun Fire X4600 M2
•  AMD64 Opteron 2.6 GHz
• 8 Dual core CPU
• Memory Ecc Registered 32 GB
• HardDisk 73 GB 2 Units
• Storage Sun Storage Tek 6140 Array

14
ECLIPSE Cluster

• Own by BIOTEC, MOST
• Largest HPC system in Thailand
• Physical specification
• 22 nodes of Sun Fire X4600 M2 , 32 cores per node
• 704 cores of AMD Opteron 2.3GHz 3.238 Rpeak
  performance
• 64GB memory per node x 22 nodes 1.4TB memory
  total
• Infiniband links connect all 22 nodes together
• Application
• Serving Bioinformatics Statistical application
  for BIOTEC Partner universities researcher

15
Workshop on HPC Education under JCSSE6th May
14,2009

• How to educate a future generation to understand
  and keep up with rapid changing technology?
• What is the skill set required by research and
  industry?
• What is the right element of HPC that should be
  teach?
• How to create a growth cycle that drive the
  expansion of adoption and practice of HPC?
• How to attract more investment from government
  and industry in this area?

16
Format

• Each speaker will have 15-20 minutes to express
  the idea
• Participants from 7 universities
• Chulalongkorn university
• Kasetsart University
• KMUTT
• KMUTNB
• NECTEC
• Mahidol University
• Thammasart University

17
Some Facts

• HPC teaching is available in most institute
• Mostly a selective course
• HPC research is being practice in all
• HPC application
• Simulation, logistics , geoinformatics,
  bioinformatics, medical system
• HPC system
• Monitoring, collaborative system, portal
• HPC algorithm
• Communication, database, scheduling, data
  partitioning
• Platform
• Self-built
• ThaiGrid machine
• Small test cluster

18
Teaching

• Mostly using Linux
• MPI and Multicore is the main things
• GPU is at the beginning , high potential
• Early interest in using Windows and alternative
  parallel programming
• C, CCR, MPI.NET
• Awareness of energy usage, performance,
  reliability should be added.

19
Common Interest

• Multicore ?
• Multicore forum on Facebook
• GPU consortium ?
• Certification, how to proof to industry that our
  students know about parallel computing

20
Long Term Objective
SAC
21
Summary

• HPC is growing steadily in Thailand
• Most users are still in academics
• HPC teaching is now common in leading institute
  producing a few hundreds people per year
• Limited research and education funding will slow
  the growth of HPC in education
• HPC growth in the future will depends on how well
  the technology is adopted in industry
• Driving more application related research that
  link with local usage and local industry
• Technology transfer program can be useful in
  introducing technology to industrial sector
• Leadership is very much needed to drive the
  growth
• Depend on the transition of ThaiGrid to MOST

22
Thank you