Empirically comparing strict vs. lazy evaluation

Hello Haskellers!

I wonder if you know of benchmarks that attempt to compare,
empirically, lazy vs. eager evaluation.  Pointers to papers and/or
code would be most appreciated.

Our group (at UMD) is working on a paper that develops some technology
for lazy programs, and we would like to choose benchmarks for
evaluating it that span the gamut of computations that perform well
when done lazily, vs. those that do not perform as well when evaluated
lazily. Then we want to look at the best case and worst case for our
technology WRT laziness, and see how it stacks up.

Thanks in advance for any ideas you have!

Kris Micinski