earticle

논문검색

Acceleration of Correlation Matrix on Heterogeneous Multi-Core CELL-BE Platform

초록

영어

A parallel implementation of the highly useful and computational intensive cor-relation matrix arithmetic is presented here. A heterogeneous multi-core CELL-BE processor platform has been used for the purpose of accelerating the performance of the kernel. An extensive measures of the performance reflects the massive capability of the CELL-BE processor architecture. In comparison with a Power Processor, the CELL-BE achieves about 500 time better performance, whereas it is showing around 120 times better performance with that of a dual-core Pentium-4 processor.

목차

Abstract
 1. Introduction
 2. Background
  2.1 Correlation Matrix
  2.2 CELL-BE Processor
  2.3 Programming the Cell Processor
 3. Design Strategy on CELL-BE
  3.1 Design Issue in Cell Implementation
 4. Implementation
  4.1 Data Alignment
  4.2 Mean Computation
  4.3 Covariance Matrix Computation
  4.4 Correlation Matrix Computation
 5. FLOP Performance Esti
  5.1 MEAN
  5.2 New Matrix (new ran arr)
  5.3 COVARIANCE MATRIX (ssyrk )
  5.4 STANDARD DEVIATION
  5.5 CORRELATION MATRIX
 6. Performance Result and Comparison
 7. Conclusion
 References

저자정보

  • Manish Kumar Jaiswal Assistant Professor, Faculty of Science & Technology, The ICFAI University, Dehradun, India.

참고문헌

자료제공 : 네이버학술정보

    함께 이용한 논문

      ※ 원문제공기관과의 협약기간이 종료되어 열람이 제한될 수 있습니다.

      0개의 논문이 장바구니에 담겼습니다.