Performance Evaluation of Low Level Multithreaded BLAS Kernels on Intel Processor Based cc-NUMA Systems

摘要

Parallel implementation of the BLAS library for sparse matrix algorithms in computational linear algebra is a critical problem, especially on the shared memory architectures with finite memory bandwidth. In this study, we evaluate the performance of the cc-NUMA systems using low level multithreaded BLAS kernels. The performance of both the compiler and the systems are evaluated on two Intel processor based architectures, NEC TX7/AzusA and IBM xSeries 440.