What about trying to improve and tweak the code which is used to create these benchmarks?
http://shootout.alioth.debian.org
Many Tcl'ers already did a great job, but there are still some scripts missing and some others might have room for optimisations.
The disciplines:
DKF: Before taking any of this too seriously, do remember to read http://www.cis.udel.edu/~silber/470STUFF/article.pdf (also linked from the FAQ of the shootout site).
hat0 where something like this becomes useful is in public relations — being able to have a strong showing in code size, performance, memory use, etc., or showing some good performance with a piece of particularly elegant code, this will draw some good, positive attention to this fine language.
TV If I were a student assistant (which I was in electronics for various student kinds) and I was to deal with this, or if some graduation student from CE would be co-managed by me I'd kindly wish to remind such a person of what was a common idea at the faculty (of EE) where I worked at the time: when people talk about an optimal system (control systems it was about) they should be asked what optimal means in the case at hand, in other words: what is being optimized.
2010-1-11 pmarin. Tcl and many other languages have been dropped from the current benchmark. Why?
Anyway, the last benchmark with Tcl can be found here: [L1 ]
2012-01-13 akm Amusingly the current gcc C entry uses the Tcl regexp engine :-) [L2 ]