Carl,

>Thread started: Does violating 3rd normal form make queries faster?

In certain situations, yes denormalizing can increase teh performance of data retreival. That is the exact reason that Chris Date discusses for denormalization. Date's Premiss is, "In a sense, the normalization discipline can be regarded as optimizing for update at the expense of retreival - a fully normalized database tends to require less processing on update but more on retreival." Relational Database Selected Writings, C. J. Date Addison-Wesley 1986.

>What is a "two table relation"? I am guessing USE a, USE b, SET RELATION...

Yes.

>Why would the location of the tables make a difference? are you saying that if they are local, one method will be faster? I would think that local tables would speed up both processes by a constant factor.

The difference in table access woudl be more pronounced on server resident tables meanign that my tests looked at only a best case scenario.

>What is "the query has to run" and "processing the records"? To me they are the same, so I have no idea what you mean.

First one needs to get teh records then one needs to process them. My test was on how fast the records could be processed.