Note on a problem with Reed and Long's FBR results

  • Authors:
  • John T. Robinson;Murthy V. Devarakonda

  • Affiliations:
  • IBM T. J. Watson Research Center, P. O. Box 704, Yorktown Heights, NY;IBM T. J. Watson Research Center, P. O. Box 704, Yorktown Heights, NY

  • Venue:
  • ACM SIGOPS Operating Systems Review
  • Year:
  • 1997

Quantified Score

Hi-index 0.00

Visualization

Abstract

This short note discusses a problem we found in a recent article in Operating Systems Review [1]. The problem is related to apparent inconsistencies in results presented for a data cache algorithm, known as FBR (frequency based replacement), which we first described in [2]. A property of this algorithm is that cache miss ratios obtained using FBR can be related to LRU cache miss ratios for any given trace. Application of this property to the results in [1] reveals apparent inconsistencies. These inconsistencies could be the result of errors either in implementing the algorithm or in recording the results. This note also highlights this property of FBR, which should help contribute to its understanding.