In simulation studies of parallel processors, it is useful to consider
the following abstraction of a parallel program. A job is partitioned
into n processes, whose running times are proportional to i.i.d. rand
om variables X(1),..., X(n). As a measure of performance we consider t
he normalized job completion time S = max{X(i)}/Sigma(i=1)(n) X(i). We
consider a simple approximation to the expected value of S, valid asy
mptotically whenever the X(i)'s are bounded, and assess its accuracy a
s a function of n both theoretically and experimentally. The approxima
tion is easy to compute and involves only the mean, variance, and maxi
mum of X(i). We also give asymptotic series for the mean and standard
deviation of S when the X(i) have a uniform distribution. In this case
, the standard deviation is asympotically negligible compared to the m
ean.