In the past week, I was working on the followings.

First, I've helped Lerong with his journal revision and proof read
Changbo's journal.

For Lerong's DAC paper, I've read the updated version. It seems
that for timing, one method applies the same placement to all dies
according to the writing. If spatial variation map is not correlated
across dies, I suspect that the gain may be averaged out. I think what
Lerong implemented was to apply different placements across dies and
compare the best location against the worst one within one die for this
method. For another method, Lerong compares the best one within a few dies
against the worst one.

For power optimiation, the optimization within a die does not achieve
significant gain while the optimization accross dies does achieve gain.
One explanation is that the power is the sum of a large number of random
variables as stated in the paper. I suspect that the gain accross dies
then comes from global variation instead of spatial variation.

For VPR integration, I've integrated the statistical timing-driven
placement algorithm into our low power VPR package. I've tested one
combinational and one sequential design for the integrated package. 
Ongoing work includes integrating Yu's part, extending timing model 
considering spatial correlated variations, and Altera proposal.