I'm not sure why but the Tally Table method appears (from this angle... can't tell really) to be Cross Joining with the data which, of course, is going to make it terribly slow. The question is... why? Again... can't tell from here and it's not happening that way with any of the (apparently) same tests I've run on 2005 or 2008. At this point, I'm not sure what the two of us are doing differently.

I've also lost track a bit of what we've done. The difference may be in the Tally Table itself (post the CREATE TABLE statement and all indexes you have on it just so I can have a look, please?). This type of stuff is difficult to TS remotely.

If you want (don't worry about the CLR side... we know that's good), post all of the code YOU are using to setup the test, the function, and the Tally Table and I'll try that on my machine to see if I can duplicate the problem you're having. I have both a 2k5 and 2k8 machine at home to test on now.

And, yes... I realize that much of the code may be what's already been posted and much of it may be mine. Since I'm trying to duplicate your problem, I wanted you to post the actual code you used so I can try and find a difference.

Thanks, Pavel.