My papers, but you know what they say:
D. P. Miranker. ``TREAT:A Better Match Algorithm for AI Production Systems''. In Proceedings of the National Conference on Artificial Intelligence, pages 42--47. American Association for Artificial Intelligence, August 1987.
D.P. Miranker and B. J. Lofaso, "The Organization and Performance of a TREAT Based Production System Compiler", IEEE Trans. on Knowledge and Data Engineering, March, 1991, 3-10
Independent duplication of results:
Y.-W. Wang and E. Hanson. ``A Performance Comparison of the Rete and TREAT Algorithms for Testing Database Rule Conditions''. In Proceedings of the Eighth International Conference on Data Engineering, 1992. Contains very detailed cost functions and graphs.
J.-P. Haton and F. Charpillet. ``X-TRA: An Inference Engine Integrating an ATMS and Two Rule TREAT or RETE Compilation Modes''. In Proceedings of the 9th International Conference on Expert System, Avignon, France, pages 617--630, May 1989. Much like my work, except has the added credibility that I didn't do it.
P. Nayak, A. Gupta, and P. Rosenbloom. ``Comparison of the Rete and Treat Production Matchers for Soar (A Summary)''. In Proceedings of the National Conference on Artificial Intelligence, pages 693--698. AAAI, August 1988. Only people to ever find in favor of RETE. A good lesson on how to exploit statistics. See if you can find the flaw in the paper.
In Japanese:
"Comparison of Two Matching Algorithms, RETE and TREAT", and "Concurrent Production System ARCH-3", where TREAT is implemented. Both papers were presented at the 37th annual convention of Information Processing Society of Japan, 1988.
If you have know of a paper comparing TREAT vs RETE that is not in this list, please let me know. (miranker@cs.utexas.edu)