[Chandler-dev] Maximum acceptable std dev for perf test?

[Heikki Toivonen]

I did some checks on how variable our performance test really are (from
current results today):

min std dev  0.00%
max std dev 78.57%
median       5.98%

In other words, about half of the performance tests actually have less
than 6% standard deviation. Unfortunately, some platforms are worse than
others; on Windows only one test has a standard deviation under 5%.

What is the threshold standard deviation where the test becomes
meaningless to you?

-- 
  Heikki Toivonen


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 254 bytes
Desc: OpenPGP digital signature
Url : http://lists.osafoundation.org/pipermail/chandler-dev/attachments/20070417/564c7a0e/signature.pgp