We've discovered that our SVD software writes doubles rather than floats.
The algorithm in the original lab has not changed, but the byte offsets
into the "out" file are different (to compute "bytesPerVector," multiply
by sizeof(double) rather than sizeof(float)), and the vectors (for
example, qv[], queryVec[], etc) should be doubles rather than floats.
The lab 6 writeup has been changed to reflect this modification. It
can be found at
http://www.cs.utk.edu/~cs494/labs/lab6.1.ps
*** NOTE!!! *** that the filename for the lab has changed
(to try to subvert the Netscape caching mechanism - sigh...).
Todd