Headline in Arial Bold 30pt. Visualisation using the Grid Jeff Adie Principal Systems Engineer, SAPK July 2008

Similar documents
Making Supercomputing More Available and Accessible Windows HPC Server 2008 R2 Beta 2 Microsoft High Performance Computing April, 2010

Headline in Arial Bold 30pt. SGI Altix XE Server ANSYS Microsoft Windows Compute Cluster Server 2003

HPC and IT Issues Session Agenda. Deployment of Simulation (Trends and Issues Impacting IT) Mapping HPC to Performance (Scaling, Technology Advances)

WELCOME! TODAY S WEBINAR: RECIPES FOR PRODUCT DESIGN & AEC WORKSTATION SUCCESS. Mike Leach. July 25, Senior Workstation Technologist Lenovo

HPC Solution. Technology for a New Era in Computing

FEMAP/NX NASTRAN PERFORMANCE TUNING

A Breakthrough in Non-Volatile Memory Technology FUJITSU LIMITED

The GPU-Cluster. Sandra Wienke Rechen- und Kommunikationszentrum (RZ) Fotos: Christian Iwainsky

Managing CAE Simulation Workloads in Cluster Environments

Distributed Virtual Reality Computation

HPC Architectures. Types of resource currently in use

The rcuda middleware and applications

RWTH GPU-Cluster. Sandra Wienke March Rechen- und Kommunikationszentrum (RZ) Fotos: Christian Iwainsky

Erkenntnisse aus aktuellen Performance- Messungen mit LS-DYNA

Faster Innovation - Accelerating SIMULIA Abaqus Simulations with NVIDIA GPUs. Baskar Rajagopalan Accelerated Computing, NVIDIA

Pedraforca: a First ARM + GPU Cluster for HPC

Intel Workstation Technology

Solving Large Complex Problems. Efficient and Smart Solutions for Large Models

MSC Nastran Explicit Nonlinear (SOL 700) on Advanced SGI Architectures

LBRN - HPC systems : CCT, LSU

REAL PERFORMANCE RESULTS WITH VMWARE HORIZON AND VIEWPLANNER

IBM Power Systems HPC Cluster

High Performance Computing with Accelerators

CMPE 665:Multiple Processor Systems CUDA-AWARE MPI VIGNESH GOVINDARAJULU KOTHANDAPANI RANJITH MURUGESAN

WELCOME! LIVE with ROBERT GREEN:

General Purpose GPU Computing in Partial Wave Analysis

Sun Lustre Storage System Simplifying and Accelerating Lustre Deployments

RZG Visualisation Infrastructure

Illinois Proposal Considerations Greg Bauer

GPGPU Applications. for Hydrological and Atmospheric Simulations. and Visualizations on the Web. Ibrahim Demir

Threading Hardware in G80

ANSYS Improvements to Engineering Productivity with HPC and GPU-Accelerated Simulation

CSE 591: GPU Programming. Introduction. Entertainment Graphics: Virtual Realism for the Masses. Computer games need to have: Klaus Mueller

HPC Middle East. KFUPM HPC Workshop April Mohamed Mekias HPC Solutions Consultant. Agenda

CS8803SC Software and Hardware Cooperative Computing GPGPU. Prof. Hyesoon Kim School of Computer Science Georgia Institute of Technology

OCTOPUS Performance Benchmark and Profiling. June 2015

designed. engineered. results. Parallel DMF

Fujitsu VDI / vgpu Virtualization

LS-DYNA Best-Practices: Networking, MPI and Parallel File System Effect on LS-DYNA Performance

Johannes Günther, Senior Graphics Software Engineer. Intel Data Center Group, HPC Visualization

Introduction to Visualization on Stampede

Maximizing Six-Core AMD Opteron Processor Performance with RHEL

FUJITSU Server PRIMERGY CX400 M4 Workload-specific power in a modular form factor. 0 Copyright 2018 FUJITSU LIMITED

Advances of parallel computing. Kirill Bogachev May 2016

Understanding Dynamic Parallelism

Delivering Real World 3D Applications with VMware Horizon, Blast Extreme and NVIDIA Grid

Smart Trading with Cray Systems: Making Smarter Models + Better Decisions in Algorithmic Trading

CSE 591/392: GPU Programming. Introduction. Klaus Mueller. Computer Science Department Stony Brook University

High Performance Computing with Fujitsu

PART-I (B) (TECHNICAL SPECIFICATIONS & COMPLIANCE SHEET) Supply and installation of High Performance Computing System

Accelerating Realism with the (NVIDIA Scene Graph)

GPGPU, 1st Meeting Mordechai Butrashvily, CEO GASS

Real - Time Rendering. Graphics pipeline. Michal Červeňanský Juraj Starinský

IBM Power AC922 Server

EE , GPU Programming

Building NVLink for Developers

FUSION1200 Scalable x86 SMP System

(Reaccredited with A Grade by the NAAC) RE-TENDER NOTICE. Advt. No. PU/R/RUSA Fund/Equipment Purchase-1/ Date:

The Stampede is Coming: A New Petascale Resource for the Open Science Community

BEAR User Forum. 24 June 2013

HP GTC Presentation May 2012

in Action Fujitsu High Performance Computing Ecosystem Human Centric Innovation Innovation Flexibility Simplicity

The HP Blade Workstation Solution A new paradigm in workstation computing featuring the HP ProLiant xw460c Blade Workstation

DATARMOR: Comment s'y préparer? Tina Odaka

On-Demand Supercomputing Multiplies the Possibilities

Teaching Cg. This presentation introduces Cg ( C for graphics ) and explains why it would be useful when teaching a computer graphics course.

Allinea Unified Environment

Solaris Engineered Systems

MOHA: Many-Task Computing Framework on Hadoop

VISUALISATION AND ANALYSIS

Maximize automotive simulation productivity with ANSYS HPC and NVIDIA GPUs

Performance Optimizations via Connect-IB and Dynamically Connected Transport Service for Maximum Performance on LS-DYNA

NVIDIA GRID. Ralph Stocker, GRID Sales Specialist, Central Europe

GeoImaging Accelerator Pansharpen Test Results. Executive Summary

N-Body Simulation using CUDA. CSE 633 Fall 2010 Project by Suraj Alungal Balchand Advisor: Dr. Russ Miller State University of New York at Buffalo

ECE 574 Cluster Computing Lecture 16

Enhancing Analysis-Based Design with Quad-Core Intel Xeon Processor-Based Workstations

Stan Posey, CAE Industry Development NVIDIA, Santa Clara, CA, USA

HIGH PERFORMANCE COMPUTING FROM SUN

CUDA Conference. Walter Mundt-Blum March 6th, 2008

Overview of Parallel Computing. Timothy H. Kaiser, PH.D.

To hear the audio, please be sure to dial in: ID#

Leveraging Software-Defined Storage to Meet Today and Tomorrow s Infrastructure Demands

Profiling and Debugging Games on Mobile Platforms

MELLANOX EDR UPDATE & GPUDIRECT MELLANOX SR. SE 정연구

Turbostream: A CFD solver for manycore

Leveraging LS-DYNA Explicit, Implicit, Hybrid Technologies with SGI hardware and d3view Web Portal software

HPC and IT Issues Session Agenda. Deployment of Simulation (Trends and Issues Impacting IT) Mapping HPC to Performance (Scaling, Technology Advances)

Oncilla - a Managed GAS Runtime for Accelerating Data Warehousing Queries

Delivering Transformational User Experience with Blast Extreme Adaptive Transport and NVIDIA GRID.

Windows Compute Cluster Server 2003 allows MATLAB users to quickly and easily get up and running with distributed computing tools.

Creating High Performance Clusters for Embedded Use

Engineers can be significantly more productive when ANSYS Mechanical runs on CPUs with a high core count. Executive Summary

Linux Compute Cluster in the German Automotive Industry

Dell DVS. Enabling user productivity and efficiency in the Virtual Era. Dennis Larsen & Henrik Christensen. End User Computing

Welcome! Roberto Mucci SuperComputing Applications and Innovation Department

Whiz-Bang Graphics and Media Performance for Java Platform, Micro Edition (JavaME)

MERCED CLUSTER BASICS Multi-Environment Research Computer for Exploration and Discovery A Centerpiece for Computational Science at UC Merced

ANSYS HPC. Technology Leadership. Barbara Hutchings ANSYS, Inc. September 20, 2011

HPC Current Development in Indonesia. Dr. Bens Pardamean Bina Nusantara University Indonesia

Transcription:

Headline in Arial Bold 30pt Visualisation using the Grid Jeff Adie Principal Systems Engineer, SAPK July 2008

Agenda Visualisation Today User Trends Technology Trends Grid Viz Nodes Software Ecosystem Remote Visualisation Computational Steering Q & A 17-Sep-08 Slide 2

The Visualization Landscape Textures Pixel Fill Rate Shading Visualization Gaming/Movies Model Size Polygon Count Compute Power Sci Viz Image courtesy 20 th Century Fox 17-Sep-08 Slide 3

Market Forces Fragmented Market LAN / WAN Clusters Sci Viz Fragmented Market Visualization GPU GP/GPU Gaming/Movies Complex Workflows 17-Sep-08 Slide 4

17-Sep-08 Slide 5

Visualization User Trends Increasingly Complex Data Sets Scope: Larger Geographic and Physical Models Precision: More polygons per model Depth: More complex interactions Need for Speed Run multiple iterations of a problem Time to solution remains critical, even with very large models Ability to interact in real-time with very large models Display & Interactivity Requirements Very large display environments Computational Steering and Interactivity Multiple remote displays Remote group collaboration 17-Sep-08 Slide 6

Example: Manufacturing CAE 17-Sep-08 Slide 7

Example: NASTRAN job surge> ls -l total 163260624 -rw-r--r-- 1 jeffadie users 1558 2008-08-21 17:08 INPUT2_with_ACMS.dat -rwxr-xr-x 1 jeffadie users 249 2008-08-21 17:20 ana64 -rw-r--r-- 1 jeffadie users 484103 2008-08-21 19:39 input2_64core.f04 -rw-r--r-- 1 jeffadie users 107970679 2008-08-21 19:39 input2_64core.f06 -rw-r--r-- 1 jeffadie users 70541 2008-08-21 19:39 input2_64core.log -rw-r--r-- 1 jeffadie users 83535079275 2008-08-21 19:39 input2_64core.pch -rwx------ 1 jeffadie users 0 2008-08-21 17:19 mkl.cfg -rwxr-xr-x 1 jeffadie users 435 2008-08-21 17:14 run64 surge> 17-Sep-08 Slide 8

Example: NASTRAN Job Input File : 41MB Output File: 83.5GB Network (Singapore USA): 1Mb/s measured Upload model: 410s ~ 7 minutes Solution (1000 steps, 64 cores): ~ 1 hour Download results: 835,000s ~ 232 hours ~ 10 days NO WAY TO REASONABLY GET RESULTS 17-Sep-08 Slide 9

Visualization Technology Trends Increasing COTS GPU Performance Price/Performance Advantage Use as Application Accelerators Cluster Architectures Distributed Compute & Render Expandability High Speed Data I/O Necessary to handle large data transfers Resource Optimization Identify and utilize available compute & render resources Grid computing and Virtualization Smart Workflow Management Increase user and system productivity 17-Sep-08 Slide 10

Implementing Visualization on Clusters Leverage Cluster Architecture and Advantages Clusters dominate mindshare and marketshare Scalability, Costs, Selection, Performance, Open Source Reduce Cluster Complexity Integration, Q/A, Time-to-Productivity Leverage Graphics (Gaming) Technology Provide Textures, Pixel Fill Rate, Shading Graphics Cards don t directly address Visual Supercomputing Demands The Challenge of Visualization on Clusters Scaling Visualization Applications Maintaining Data I/O Throughput Software Stack Complexity Resource Management Maintaining Interactivity in a Batch Environment 17-Sep-08 Slide 11

SGI Virtu VN200 Visualization Node 8 FB-DIMMs 8 to 32 GB RAM DDR 4x IB interconnect Two (2.5 ) SATA Drives NVIDIA Quadro FX High Performance GPU Two (2) Intel Quad core Xeon CPUs Redundant (2) Power Supplies 17-Sep-08 Slide 12

SYSTEM MANAGEMENT STORAGE MANAGEMENT FILE SYSTEM COMPILERS LIBRARIES PROFILERS, DEBUGGERS Software Ecosystem JOB SCHEDULER LINUX OPERATING SYSTEM 17-Sep-08 Slide 13

REMOTE VISUALIZATION SYSTEM MANAGEMENT STORAGE MANAGEMENT FILE SYSTEM COMPILERS LIBRARIES PROFILERS, DEBUGGERS GRAPHICS TOOLKITS Software Ecosystem WORKFLOW ORCHESTRATION BATCH SCHEDULER INTERACTIVE SCHEDULER JOB SCHEDULER LINUX OPERATING SYSTEM 17-Sep-08 Slide 14

The Market Demand for Grid Visualization Global workflow is reality in today s business Visualization must fit into the global workflow model Greater demands on security, data sharing, and resource management Reduce costs for applications and client systems Grid virtualization Laptop/desktop limitations Without Remote Grid Visualization, the only method to share data is to physically copy the data to a remote location, with duplicate applications and processing power. 17-Sep-08 Slide 15

Remote Grid Visualization: Key Concepts Traditional Approach: Data Paradigm Download DATA to individual graphics workstations Multiple copies of the model/data across users Large data transmissions across networks, iteratively Store Compute Manage Render / Interact 2D, 3D, 4D non-visual model Compute Nodes Admin Service Node 2D, 3D, 4D visual model Desktop workstation Batch Interactive 17-Sep-08 Slide 16

Remote Grid Visualization The Silicon Graphics Difference Visual Information Paradigm Securely share visual information to all users no data transfer Real-time iterative analysis and updates Single data image secure, simpler management, consistent Store Compute Manage Interact 2D, 3D, 4D models Compute & Visualization Nodes Visualization & Admin Node Visualization Display System or Thin Client Batch Interactive 17-Sep-08 Slide 17

Visualisation across the Grid The Power of a Reality Center The Power of Fusion The Power of Many Applications The Power of Many Devices 17-Sep-08 Slide 18

Benefits of Grid Visualisation Remove the physical link between users, their data, and the visualization system Solve much more difficult problems than possible with desktop workstations but do it from your desktop Ability to reserve HPC resources to enable interactive applications Connect experts with users and their data, independent of their locations Enable computational steering from anywhere in the world 17-Sep-08 Slide 19

SGI High Performance Visualization Computational Steering Traditional Methods Interactive Discovery Prepare Prepare/Analysis Submit Jobs Analysis Simulation Interactively change the simulation Faster Time to Insight 17-Sep-08 Slide 20

Summary Remote Grid Visualisation benefits: Data size One copy of the data is maintained in the database. User has fast access to the data. Sharing data Parallel file systems allows multiply users to analysis the database. Model size Distributed rendering allows larger models to be visualized. Frame buffer limit of the GPU is no longer a limiting factor. Faster computation Render nodes are integrated in the Grid and can be used to for MPI jobs when not running visualization. License consolidation For end user application licenses can be shared on the Grid as opposed to each user having their own license. 17-Sep-08 Slide 21

Thank You! 17-Sep-08 Slide 22