NOTUR notur

NOTUR

www.notur.org

The NorwegianHigh-Performance Computational

Infrastructure

NOTUR

The Consortium:NTNUUiOUiBUiT

DNMISINTEFStatoil

Ceetron ASA/ViewTech ASA(Telenor)

….in response to

The Principal Mission

defined by the Research Council to be:• to offer a cost effective national infrastructure for high

performance computing for scientists at Norwegian universities and colleges;

• to provide the Norwegian Meteorological Institute with the required high performance computing facilities for operational forecasting and research;

• to ensure that other participants who contribute to the funding of the program can utilise the facilities.

National High Performance Computational Environment

A nationally accessible high-performance computational environment will give users a competitive edge.

TRONDHEIM

OSLO

BERGEN

Tromsø

National Roles

NTNU: Large-scale parallel computation.

UiO: Data intensive computation.

UiB: Affordable parallel computing.

UiT: CPU-intensive sequential simulations.

(Serve local needs)

High Operational Quality

Round the clock continuous operation will provide high availability and stability of the resource.

Operations Metacenter

Common framework for the user interface, including e.g.

• homogeneous file system• homogeneous queuing system• uniform application- and log-in

environmentAvoid duplicate efforts through e.g.

• common training programs• local user support, but with a single

contact point

Integration

Integrated collaboration between universities, research institutes and industry will accelerate the innovation process.

Feedback

Implementation

Decision

Solution

Modelling

Problem formulation

Technology Push and Application Pull

Multi-disciplinary collaboration between mathematicians, computer scientists and researchers within science and engineering will stimulate innovative research and development. A

CM

Applications

Tec

hnol

ogie

s

Projects

CSE

Scientific Visualisation

The national high-performance computational environment will be extended to include real-time and large-scale visualisation technologies.

International Collaboration

Links to international partners will be important for the national HPC environment.

Project Organisation

M anagem entTeam

O p e ratio nsU iO

O p e ratio nsU iB

O p e ratio nsU iT ø

O p e ratio nsN T N U

P roc u rem e n t andIn s talla tion

E m ergingT ec h no lo g ies

T ec h n. T ran s ferP ro jec ts

O u trea ch a ndD is sem in a tion

Project Manager

Steering Com m ittee

NTNUv/U niversity D irector

Resource allocationC om m ittee

Research Councilv/Program m e C om m ittee

for H PC

HPC Metacenter

HPC Metacenter

The goal of the Metacenter is

to establish a uniform, and

as seamless as possible,

access to the Resources and

Support infrastructure in the

NOTUR network

• Quota application• Regulations of use• Training• Computer resources• Application management• User support

Emerging Technologies

Enable the Norwegian research and engineering community to take advantage of early use of emerging technologies in HPC.

• Technology watch (new technologies, new demands)

• Computational GRID concepts• Support the Metacenter

(operations, user interface)• National dimension

Technology Transfer Projects

The goal of the TTPs is to

strengthen the competence and

usage of HPC.

In addition the project shall

accelerate the accessability of

effective HPC-tools and

mechanisms amongst

HPC-users in Norway.

• Scalable algorithms• Scientific visualisation• Climate and Environment• Industry and Energy• Nature Science

Outreach and Dissemination

Acceleration of the awareness and take up of HPC among a broader field of application will extend the impact of HPC on the Norwegian society. Introduce HPC to at least one

new research area every year

Budget (Total 168 MNOK)

Expenditures Sources

Progress Plan

Q4, 2000: First installation, Trondheim. Start Metacenter, Emerging technologies, Outreach and dissemination.

Q1-Q2, 2001: First installations at UiO and UiT. Start TTPs.

Q3, 2001: Installation at UiB.

Q2-Q4, 2002: Equipment upgrades at NTNU, UiO, UiT.

Q4, 2003: Project termination, option for extension.

HPC resources (Oct-2001)

Centre System Type CPUs Mflops prcpu/

Total Gflops

Totalmemory

(GB)

Diskcapacity

(GB)

NTNU SGI Origin SMP 220 800/176 220 1700

NTNU Cray T3E MMP 96 600/58 13 237

UiO HP SuperDome SMP 44 2200/97 88 2500

UiO HP N4000 SMP 8 1760/14 6 300

UiT HP SuperDome SMP 32 2200/71 32 1000

UiB

Possible Areas of Nordic Cooperation

• User group, operational experience• Technical problems data base• User FAQs data base • Short courses, training programs for technical

staff