QCD on the connection machine

In this talk I give a brief introduction to the standard model of particle interactions and illustrate why analytical methods fail to solve QCD. I then give some details of our implementation of the high performance QCD code on the CM2 and highlight the important lessons learned. The sustained speed of the code at the time of this conference is 5.2 Gigaflops (scaled to a full 64K machine). Since this is a conference dedicated to computing in the 21st century, I will tailor my expectations (somewhat idiosyncratic) of the physics objectives to reflect what we will be able to do in 10 years time, extrapolating from where we stand today. This work is being done under a joint LANL-TMC collaboration consisting of C. Baillie, R. Brickner, D. Daniel, G. Kilcup, L. Johnson, A. Patel. S. Sharpe and myself. 5 refs