Greetings,
We have received an interesting suggestion on how to improve the
working set algorithm: the fill phase goal can be set to the maximum of
(a) twice the cache size and
(b) working set size (WSS).
Note that (a) is the current fill goal. The working set is currently
frozen regardless of the phase schedule.
Extending the goal will not have any effect on caches at least WSS/2 in
size, of course. One can hope that most caches meet that criteria. For
caches that are smaller than WSS/2, the duration of the fill phase will
be increased.
The new goal would guarantee that WS is not expanding after the fill
phase is over. Thus, one can hope that the experiment would reach the
steady state sooner. Consequently, it may be possible to shorten the
"top" phases by, say, 30-60 minutes.
Comments? Objections?
Alex.
This archive was generated by hypermail 2b29 : Tue Jul 10 2001 - 12:00:14 MDT