[Linux-disciples] dma_timer_expiry problems

[Stephen R Laniel]

On Wed, Jun 30, 2004 at 06:05:35PM -0400, Adam Rosi-Kessel wrote:
> Not entirely predictable, but doesn't it seem at least a little suspicous
> that it seems to be on the :00 :20 and :40?  Maybe someone who knows
> probability could determine whether or not this could just be totally
> random because it's a small sample.

The times actually don't fall exactly on the :00, :20, and
:40. Assuming they do, though, the probability that any one
expiry error would fall on those three times is 3/60,
assuming the errors fell at random. Assuming those errors
are independent of one another, the probability that 13 of
them would fall on those three times should be something
like (3/60)^13 = approximately 1/(80 quadrillion). Which is
not a large number. A pretty small one, actually.

-- 
``Putting together Justice Thomas' opinion in Hamdi with
  his vote in ACLU v. Ashcroft, we may infer that the
  President can throw any citizen in a military prison
  indefinitely, but that the citizen has the right to view
  pornography while there."
 -Jack Balkin, http://shorl.com/bograsabrorulu