On multicores, build_third_party_packages could up the number of CPUs used by setting the MAKEFLAGS environment variables. this would make it go a lot faster on current machines.
On OSX you can get the number through the command "sysctl hw.cpu", on Linux you can read /proc/cpuinfo. Using N=(half the cores)+1 or something like that seems to be a decent choice, assuming -jN actually works for all the Makefiles that are used within build-third_party_packages.