They also have a manual where they explain how they got to these estimates. Basically, they estimate a Solow model and use the residuals to find an estimate for TFP.
Unfortunately, I'm not fvery familiar with the Penn world tables, so I can't help you there.