HPC Current Development in Indonesia SITE Basic Nodes: 80 node 2 processors per node, 4 cores per processors Dual Intel Xeon E5-2609 2,4 GHz 8 GB RAM DDR3-1600 500 GB HD SATA Dual

HPC Current Development

in Indonesia

Dr. Bens PardameanBina Nusantara University

Indonesia

HPC FacilitiesEducational & Research Institutionsin Indonesia

CIBINONG SITEBasic Nodes:80 node2 processors per node, 4 cores per processorsDual Intel Xeon E5-2609 2,4 GHz8 GB RAM DDR3-1600500 GB HD SATADual Gigabit interconnectionLinux (CentOS)

GPU Node:16 node2 processors per node, 4 cores per processorsDual Intel Xeon E5-2609 2,4 GHz8 GB RAM DDR3-1600500 GB HD SATADual Gigabit interconnectionNVIDIA Tesla M2075 GPGPULinux (CentOS

High Memory Node:8 Node2 processors per node, 8 cores per processorsDual Intel Xeon E5-2640 2 GHz256 GB RAM DDR3-16002 x 300 GB HD SASDual Gigabit interconnectionLinux (CentOS)

Master Node:2 node2 processors per node, 8 cores per processorsDual Intel Xeon E5-2650 2,0 GHz128 GB RAM DDR3-160024 TB HD SATA (Raw), RAID 5Dual 10 Gigabit interconnectionLinux (CentOS)

BANDUNG SITEBasic Nodes:34 node2 processors per node, 4 cores per processorsDual Intel Xeon E5-2609 2,4 GHz8 GB RAM DDR3-1600500 GB HD SATADual Gigabit interconnectionLinux (CentOS)

GPU Node:4 node2 processors per node, 4 cores per processorsDual Intel Xeon E5-2609 2,4 GHz8 GB RAM DDR3-1600500 GB HD SATADual Gigabit interconnectionNVIDIA Tesla M2075 GPGPULinux (CentOS)

Master Node:2 node2 processors per node, 8cores per processorsDual Intel Xeon E5-2650 2,0 GHz128 GB RAM DDR3-160024 TB HD SATA (Raw), RAID 5Dual 10 Gigabit interconnectionLinux (CentOS)

Beowulf Cluster Resource Manager: Sun Grid Engine Operating System: Linux 64bit Memory: 12.8 TB Interconnection: Gigabit LAN 320 Core Processors

BNU Bioinformatics & Data Science Research Center

Server:HPE DL380 Gen9 (Intel® Xeon® E5-2630)HPE DL360 Gen9 (Intel® Xeon® E5-2609)HPE ML350 Gen9 (Intel® Xeon® E5-2620) HP ProLiant ML350p Gen8 (Intel® Xeon® E5-2620)

GPU:NVIDIA Tesla P100NVIDIA Tesla P40NVIDIA Tesla P4NVIDIA Tesla K40NVIDIA Tesla K20NVIDIA GeForce GTX Titan X (3 units)

BNU School of Computer Science

GPU:NVIDIA Titan XPNVIDIA GeForce GTX 1050

Gadjah Mada University

GPU:NVIDIA Titan XPNVIDIA GeForce GTX 1080 Ti

University of Indonesia

GPU:NVIDIA GeForce GTX Titan X

Atma Jaya University Yogyakarta

GPU:NVIDIA GeForce GTX 1080 Ti (5 units)

Bandung Institute of Technology

Front end : 1 node with Intel Xeon 8 core, RAM 12 GB, HDD 4 TBCompute nodes: 2 nodes with AMD 32 core, 12 GB of RAMGPU Compute nodes: 4 nodes with Intel core i7, RAM 6 GB, GPU NVIDIA 950GTXLogin Node: 1 buah komputer Intel Xeon 12 core, RAM 12 GB

BNUResearchNetworks

NEAR-FUTURE PARTNERS TO HAVE HPC SYSTEM

NATIONWIDE UNIVERSITY NETWORK INDONESIA (NUNI)

INTERNATIONAL PARTNERS

USA USA

The Usageand Future Development of HPC in Indonesia

HPC inICABIOGRAD

(Indonesian Center for Agricultural Biotechnology and Genetic Resources Research and Development)

Indonesian Center for Agricultural Biotechnology and Genetic Resources Research and Development (ICABIOGRAD) is a technical operative unit of research and development under the authority of the Indonesian Agency for Agricultural Research and Development (IAARD). ICABIOGRAD has a task and mandate to perform research on biotechnology and agricultural genetic resources.

High Performance Computing (HPC) was deployed as a high-speed and high-capacity computing for solving complex computational problems in genome-wide association studies (GWAS) for agriculture genetic research.

Many problems have been solved with HPC techniques that were impossible to solve with individual workstations or personal computers.

Dell Precision

7500

DELL PowerEdge

R610 (3 units)

EMC Isilon iQ2000

(3 units –capacity 36 TB)

User (Remote)

Illumina Systems

NVIDIA - BNU AI R&D CENTER

• NVIDIA - BNU AI R&D Center aims to accelerate research and development of AI in Indonesia. To achieve this goal, AI R&D Center opens up collaboration in Artificial Intelligence (AI) research between academics and industries.

• AI R&D Center also holds numerous AI workshops to accelerate the spread of AI knowledge in Indonesia. These workshops are run in collaboration with NVIDIA Deep Learning Institute.

.

BNU HPC RESEARCH NETWORKS

BENCHMARKING with HPC at UNIVERSITY of SOUTHERN

CALIFORNIA• Dell X2200 with 10 Gigabit of Backbone, 4-core/dual proc 2.3GHz, 16GB memory• Dell PE1950 with 10 Gigabit of Backbone, 4-core/dual proc 2.3GHz, 12GB memory• Dell PE1950 with 10 Gigabit of Backbone, 4-core/dual proc 2.3GHz, 12GB memory• Oracle X2200 with 10 Gigabit of Backbone, 4-core/dual proc 2.3GHz, 12GB memory• IBM DX340 with 10 Gigabit of Backbone, 4-core/dual proc 2.3GHz, 16GB memory • IBM DX360 with 10 Gigabit of Backbone, 6-core/dual proc 3.0GHz, 24GB memory• HP SL160 with 10 Gigabit of Backbone, 6-core/dual proc 3.0GHz, 24GB memory• HP DL165 with 10 Gigabit of Backbone, 12-core/dual proc 2.3GHz, 48GB memory• Dell R910 with 10 Gigabit of Backbone, 10-core/quad proc 2.0GHz, 1TB memory• HP SL250S 56.6 Gigabit of Backbone, 8-core/dual proc 2.4GHz, 64GB memory dual K20 NVIDIA GPUs• HP SL230S 56.6 Gigabit of Backbone, 8-core/dual proc 2.4GHz, 128GB memory

https://hpcc.usc.edu/support/infrastructure/hpcc-computing-resource-overview/

BNU HPC DEVELOPMENT ROADMAP

• Upgrade HPE DL380 with 2x GPUs Tesla V100 to be used for Deep Learning

2020 2022 2024 2026

Single GPU Node HPC for Deep Learning

Multi GPU Nodes HPC System for Distributed

Computing

• Add 4 HPE Apollo 6500 (can contain 8 GPUs) or better

• Form a distributed system with 5 CPU nodes and 3 GPU nodes

• Add 35 GPUs Tesla V100 or better

• Form a distributed system with 8 GPU nodes

• Add 4 NVIDIA DGX-1 with Tesla V100 or better

• Form a distributed system with 12 GPU nodes

Multi GPU Nodes HPC System for Distributed Deep Learning

BNU NETWORKS HPC DEVELOPMENT ROADMAP

Help 4 universities to get a GPU node:• Satya Wacana Christian

University, Salatiga• Indonesia Islamic

University, Jogjakarta• Hasanuddin University,

Makassar• Mulawarman University,

Samarinda

2020 2022 2024 2026

Single GPU Node in 4 Universities

Single GPU Node in All Partner Universities

• Help all partner universities to get at least 1 GPU node

• Help all partner universities to get more GPU nodes

• Form a distributed system with all GPU nodes of partner universities

Nationwide Multi GPU Nodes HPC System

Thank You