Silicon Graphics, Inc.
Presented by:
SGI Proprietary
SGI in the Environmental Sciences
CJ Suchyta, III, Ph.D.SGI Benchmarking Specialist
SGI Proprietary
The legal Page…
This presentation contains forward-looking statements regarding SGI technologies and third-party technologies that are subject to risks and uncertainties. These risks and uncertainties could cause actual results to differ materially from those described in such statements. The viewer is cautioned not to rely unduly on these forward-looking statements, which are not a guarantee of future or current performance. Such risks and uncertainties include long-term program commitments, the performance of third parties, the sustained performance of current and future products, financing risks, the ability to integrate and support a complex technology solution involving multiple providers and users, and other risks detailed from time to time in the company's most recent SEC reports, including its reports on From 10-K and Form 10-Q
©2004 Silicon Graphics, Inc. All rights reserved. Silicon Graphics, SGI, IRIX, XFS, Onyx, Reality Center, OpenGL, the SGI logo and the SGI cube are registered trademarks and Silicon Graphics Prism, Onyx4, CXFS, NUMAflex, OpenGL Performer, OpenGL Volumizer, OpenGL Vizserver, OpenGL Multipipe, SGI ProPack and The Source of Innovation and Discovery are trademarks of Silicon Graphics, Inc., in the U.S. and/or other countries worldwide. Linux is a registered trademark of Linus Torvalds in several countries. UNIX is a registered trademark of The Open Group in the U.S. and other countries. Intel and Itanium are trademarks or registered trademarks of Intel Corporation or its subsidiaries in the United States and other countries. Linux penguin logo created by Larry Ewing. All other trademarks mentioned herein are the property of their respective owners. (10/04)
SGI Proprietary
MM5 3.6.3 - Standard benchmark, 2004
Higheris
Better
Source: http://www.mmm.ucar.edu/mm5/mpp/helpdesk/20040304a.html
SGI Proprietary
MM5 3.6.3 t3a benchmark
Higheris
Better
Source: http://www.mmm.ucar.edu/mm5/mpp/helpdesk/20040304a.html
Altix version was compiled with the Intel 8.1.007 (beta) ifort Fortrancompiler and the Intel 8.1.010 (beta) icc C compiler, and linked with SGI's MPI from MPT 1.10.
SGI Proprietary
WRF performance
• Version 2, EM core• 425x300x35 grid, standard benchmark problem• Data from latest web page shows SGI Altix has fastest
performance• Code scales well to 1024p on standard size benchmark!
SGI Proprietary
WRF 2.0.2 - Scalability & Performance on Altix 3000
Source: http://box.mmm.ucar.edu/wrf/bench
SGI Proprietary
WRF v2.0.3 EM core
• Standard benchmark job, 425x300x35 grid
• Scales to 1024p, best total performance of all machines (compare to previous slide)
050
100150200250300350400
0 200 400 600 800 1000 1200# processors
Gflo
ps
Altix with RSL-Lite Altix with RSL
SGI Proprietary
POP 1.4.3 on various systems. 1 degree problem
0
50
100
150
200
250
4 8 16 32 64 128 256 512 960
Sim
ulat
ed Y
ears
per
Com
pute
Day
ES40XD1
p690AltixX1 MSP
SGI Proprietary
POP 1.4.3(Optimization by reducing the # of synchronizations)
Scalability on Altix using Intel Compiler and SGI MPT library
Increasing Message length and reducing synchronization frequency improved the performance by 20% at 256 CPU
SGI Proprietary
CAM 3.0
CAM 3.0_p1 T85 Simulation Speed
0
2
4
6
8
10
12
16 32 64 128
NCPUs
Sim
yea
rs/w
allc
lock
day
Altix 3700 BX2 1.6GHz/9MBL3
SGI Proprietary
CCSM 3.0 T85_gx1v3 on the SGI Altix 3700 BX2
CCSM3 T85_gx1v3 Load Balancing Experiments
0
1
2
3
4
5
6
7
8
9
10
0 50 100 150 200 250 300
Number of Processors
Sim
ulat
ed Y
ears
per
Wal
lclo
ck D
ay
SGI Altix 3700 BX2, 9MBL3/1.6GHz SGI Altix 3700 BX2, 6MBL3/1.6GHz
SGI Proprietary
CCSM 3.0 on various systems
Courtesy of G Carr, Jr. (NCAR)
SGI Proprietary
IFS - Scalability on SGI Altix 3000
Itanium2 @1.5GHz is 1.3x faster than Itanium @1.3GHz because of the larger cache and higher clock rate
Source: Roland Richter, SGI, May 2004
SGI Proprietary
Climate and Weather Model Performance
•SGI MPT library takes advantage of shared memory to provide very low latency, high bandwidth MPI communication.
– MPI codes perform very well•Globally addressable memory allows additional optimizations that greatly improve scalability and performance.
•Large shared memory allows easy post-processing and analysis of large, high resolution data sets.