Data structures and a good statistics class. I feel that an algorithms class might be overkill. I worked in computational physics and I mostly used statistics, monte carlo especially, also many numerical methods of PDE's. We had three CS people working there and they jumped in if we had any problems with the algorithm type stuff. I highly doubt that a job would require someone to do the CS-type work along with the physics work. The physicists that I worked with focused on the models almost exclusively, they would never sacrifice the completeness of the model for a more efficient program. That was mostly my job. Getting dirty and working in the trenches, lol.
When you say for computational physics, what are you envisioning that you'll be doing?