From: Michael J Mulhearn
Date: Tue, 28 Feb 2006 19:19:33 -0500 (EST)
To: Hal Evans

Hello!

Our last overnight test with 1600 starts is equivalent to some number of decades of running the system.

The number of *coarse* descews needed to force the system into a working state breaks down as follows:

total init count summary:
 -->  1 times: 1150 84.99%
 -->  2 times: 168 12.41%
 -->  3 times: 29 2.14%
 -->  4 times: 5 0.36%
 -->  5 times: 0 0%
 -->  6 times: 0 0%
 -->  7 times: 1 0.07%
 -->  8 times: 0 0%
 -->  9 times: 0 0%
 -->  10 times: 0 0%
(last bin is overflow!)

So 85% of times it works on first try (sorry, nine in ten was *bit* inflated).  12% on second, and rarely any longer.

cheers,
mike