Theory grad students read this:
Rainer put up a summary of the numerical methods discussion for globular cluster modeling on the program wiki
if you are doing any sort of collisional dynamics, or numerical modeling of collisional self-gravitating systems, you might find this useful
There is also a nice summary of the different order post-Newtonian approximations here
and some random selected subset of references - for those discussions where a number of papers came up or were requested
Rainer also put up a summary of special purpose hardware status;
specifically the GRAPES and uses of Graphical Processing Units and Field Programmable Gate Arrays for scientific simulations.