NPACI Panel on Clusters David E Culler Computer

NPACI Panel on Clusters David E. Culler Computer Science Division University of California, Berkeley http: //www. cs. berkeley. edu/~culler 1/29/99 NAPCI Clusters

Clusters have Happened • • • IBM SGI SUN HP Microsoft. . . 1/29/99 NAPCI Clusters 2

Performance + Cost-Performance • April 1998 T 3 E vs Intel Pentium II on NAS Benchmarks http: //www. nersc. gov/research/FTG/pcp T 3 E-900 PII-400 BT 63. 5 62. 2 LU 70. 2 73. 6 SP 44. 7 45. 5 1/29/99 NAPCI Clusters 3

Berkeley (NPACI) NOW • 100 Sun Ultra 2 workstations • Inteligent network interface – proc + mem • Myrinet Network – 160 MB/s per link – 300 ns per hop http: //now. cs. berkeley. edu 1/29/99 NAPCI Clusters 4

Beowolf Consortium • LINUX PCs • Fast ethernet • Basic stand-alone and batch • • • http: //www. beowulf. org/consortium. html http: //www. beowulf. org/ http: //ns. baltzer. nl/cluster. html http: //www. cs. orst. edu/swarm/about. html cluster cookbook 1/29/99 NAPCI Clusters 5

HPVM (NPACE) • • NT PCs Myricom Network Fast Messages LSF start-up http: //www-csag. cs. uiuc. edu/projects/hpvm. html 1/29/99 NAPCI Clusters 6

Berkeley Millennium Business SIMS BMRC Chemistry C. S. E. E. Gigabit Ethernet Biology Astro NERSC M. E. Physics N. E. IEOR C. E. MSME Transport Economy. Math • PC-based Unix and NT Clusters • Departmental and Campus • Shared as Computational Economy http: //www. millennium. berkeley. edu 1/29/99 NAPCI Clusters 7

What you get off the shelf • Go to your favorite web site and order – Dual PII 450 MHz, 1 GB Mem, 36 GB disk – $9, 563 at Dell • 4400 CPU Hours per Quarter to your self! – $1. 20 CPU hours (at 30% over 3 years) • Buy 5? 1/29/99 NAPCI Clusters 8

Three Kinds of Clusters • Throughput Clusters • Availability Clusters • High-Performance Parallel Clusters 1/29/99 NAPCI Clusters 9

Throughput Clusters • Workstation / PC Farms • Provide resource pool for large numbers of sequential jobs • Used widely in industry – toy story on 2000 sparcstations – ultrasparc on 1000 sparcstations • Can also include background on desktops – CONDOR • Application specific front-ends attractive – parametric studies, monte carlo • Fill cracks in parallel clusters • Big farms require infrastructure 1/29/99 NAPCI Clusters 10

Availability Clusters • Use system redundancy to mask faults – all big databases do it • VAX Clusters => IBM sysplex => Wolf Pack Clients Server A 1/29/99 Disk array A NAPCI Clusters Interconnect Server B Disk array B 11

High-Performance Clusters • Utilize modern system area networks and userlevel communication layers to construct general purpose parallel machine from commodity parts 1/29/99 NAPCI Clusters 12

Emerging System Area Networks • Gigabit Ethernet – price dropping, widely deployed • System Area Networks – Myricom – Server. Net – Synfinity • Virtual Interface Architecture – Intel/Microsoft/Compaq std based on univ. research prototypes 1/29/99 NAPCI Clusters 13

MPI Performance 1/29/99 NAPCI Clusters 14

Example: NAS Parallel Benchmarks • Better node performance than the Cray T 3 D • Better scalability than the IBM SP-2 1/29/99 NAPCI Clusters 15

Cluster-Wide Parallel I/O • Sustain 500 MB/s disk bandwidth and 1, 000 MB/s network bandwidth by driving all the disks 1/29/99 NAPCI Clusters 16

Software Base is Growing • Technical software moving to Linux and NT – NAG, matlab, petc, . . . • Cluster prototypes being hardened and packaged • Cookbooks emerging • Few cluster-integrator companies 1/29/99 NAPCI Clusters 17

Summary of the State of the Art • Cluster designs are emerging in many areas – throughput, availability, parallel computing – technology is advancing • Still immature software base – strong ties to free software movement • Many small clusters by spit and baling wire • Large clusters require engineering – commercial components improving • Rapid pace of change presents sys. Admin challenge – not unlike the desktop problem • Management tools badly needed 1/29/99 NAPCI Clusters 18

What Does it Mean for NPACI • Where do clusters fit with computational science and engineering needs? • Cycles vs Software vs Administration vs Expertise? • What role should the center take? • What role should partner sites have? 1/29/99 NAPCI Clusters 19