Hi,  I'm hoping someone can help me.  I'm trying to prove that the first

50 outputs from a given test is statistically similar to the entire
test.

[For those of you familiar with computer architecture, I'm actually
trying to prove that the behavior of the first 50 encounters of a branch

(taken or not taken) is statistically similar to the overall behavior of

a branch throughout an entire SPEC int benchmark]

I have two numbers for each branch encountered:  The percentage taken
for the first 50 encounters, then the percentage taken overall.    I've
tested hundreds of branches and have the results in tabular form.

I know how to determine the error for a particular branch:

| actual - test50 | / actual  (where test50 is the outcome of the first
50 branches)

but then how to I combine all of the errors?  Can I just average them?

Next, I need to some way to determine the statistical significance of
the error.  I mean, if I end up with an average error of 5%, how do I
prove that that is "OK" or "NOT OK".  There should be some sort of
cutoff point.

Sorry, but it's been years since I took statistics and I didn't save any

of my old textbooks.  Any help would be greatly appreciated.

--Kim



===========================================================================
This list is open to everyone.  Occasionally, less thoughtful
people send inappropriate messages.  Please DO NOT COMPLAIN TO
THE POSTMASTER about these messages because the postmaster has no
way of controlling them, and excessive complaints will result in
termination of the list.

For information about this list, including information about the
problem of inappropriate messages and information about how to
unsubscribe, please see the web page at
http://jse.stat.ncsu.edu/
===========================================================================

Reply via email to