The number of observations made by the sampling program
over a given period of time is a function of the elapsed time required
by the program being measured (t) and the average time that the
s
sampling program waits before making its next observation (t). The
a
ratio ts/t determines the number of random points generated within
a
the interval 0 to t . In order to determine a suitable value for t ,
s a
i0 replications of a FORTRAN G compile were run under conditions of
minimal paging, low paging, medium paging and high paging with t being
a
set at .05, .1, .15, .2 and .4 seconds. Based on the results a value
of t = .15 sec. was selected as the minimum value which would not
a
distort elapsed time and CPU time of the program being measured. With
an average time of approximately .02 sec. for the sampling program to
complete an observation this meant that an observation could be made in
approximately the same time as the minimum average page transport time
for a 2314, therefore page queue measurements were unlikely to change
significantly during the course of an observation.
The number of observations made by the sampling programover a given period of time is a function of the elapsed time requiredby the program being measured (t) and the average time that thessampling program waits before making its next observation (t). Thearatio ts/t determines the number of random points generated withinathe interval 0 to t . In order to determine a suitable value for t ,s ai0 replications of a FORTRAN G compile were run under conditions ofminimal paging, low paging, medium paging and high paging with t beingaset at .05, .1, .15, .2 and .4 seconds. Based on the results a valueof t = .15 sec. was selected as the minimum value which would notadistort elapsed time and CPU time of the program being measured. Withan average time of approximately .02 sec. for the sampling program tocomplete an observation this meant that an observation could be made inapproximately the same time as the minimum average page transport timefor a 2314, therefore page queue measurements were unlikely to changesignificantly during the course of an observation.
การแปล กรุณารอสักครู่..