Cray SV1 Application Performance

11/16/00

Click here to view HTML slides

or

Click here to view slides in Acrobat

Table of Contents

Cray SV1 Application Performance

Goals

OSC Systems

Outline

SV1 Architecture Highlights

SV1 Architecture

The SV1 Data Cache

The Multi-Stream Processor

The Multi-Stream Processor

Memory Structure

Single Processor Performance Issues

Vector Code Performance

Cache Considerations

Memory Bank Conflicts

Example: Vector Code

In-Cache vs Out-of-Cache

Example: Laplace Equation Solver

SSP Performance: Vector Version

SSP Memory Bandwidth: Vector

SSP Performance: Gaussian98

Gaussian98

Gaussian98

Single Processor Performance Issues Lessons Learned

Multi-Processor Performance Issues

The Multi-Stream Processor

Laplace Equation Solver - Streamed

Computational and Memory Performance

Nonlinear Wave Equation Solver

Code Version 2 - Streamed

Code Version 3 - Streamed

Computational Performance

LS-DYNA

Configurations Examined

Elapsed Time

LS-DYNA Observations

MPI Performance: QCDMPI

QCDMPI

QCDMPI

Lessons Learned

Conclusions

Additional Information

Author: Jim Giuliani
Email: jimg@osc.edu