There was a presentation at snowbird about parallelized support vector machines. In many cases, people parallelize by ignoring serial operations, but that is not what happened here—they parallelize with optimizations. Consequently, this seems to be the fastest SVM in existence.
There is a related paper here.
For “fast”, take a look at this:
http://www.cs.ust.hk/~jamesk/papers/jmlr05.pdf
“…five million training patterns, in only 1.4 seconds on a 3.2GHz Pentium-4 PC.”