There is pure simulation render on the left (notice a grid artifacts) and the result of applying advection of the cached density by velocity in the shader on the right. The method(s) is still under development, so it has many problems yet.
The fast dissipation in thin parts of the smoke is probably caused in too high volume step (in the future tests, I'll try to lowered it as possible)
Loading more stuff…
Hmm…it looks like things are taking a while to load. Try again?