I'm trying to estimate a large-scale model with about 70 parameters and 26 shocks. I use Dynare parallel estimation (the PsTools toolbox) with 8 chains (mh_nblocks=8) on a server with 8 cores. It works, but it seems like it is slower than doing consecutive chains on the same server. Anyone faced a similar problem? Any ideas to speed up the process? Ideas welcome!
