Later on that could be accomplished just by routing the exhaust down into the canyons, but early on that may have to go through greenhouses, which either vent above a certain pressure or aren't quite hermetically sealed.
While mars may have been geologically dynamic at one point, it either is drastically much less so now or pretty much dead. Terraforming by means of burning through the surface would have to be an ongoing thing process because not everything we’d burn would be beneficial for our understanding of life (us and the ecosystems we are used to on earth), but it would also be very hard to keep what we generate on Mars in an atmosphere. Could be wrong, though.
So the thermal decomposition of the rocks will require much more energy.
As you say, the thermal decomposition could be done directly with solar light, bypassing the conversion efficiency of the photovoltaic cells.
That brings the energy requirements to similar values. Which would be the more efficient process would depend on details, e.g. the type of photovoltaic cells, the exact composition of the rocks whose decomposition is attempted, the usefulness of the byproducts, e.g. reduced carbon or reduced metals, and so on.
When taking into account the byproducts, it is likely that both processes will be needed anyway.
The council of extraterrestrial life has acknowledged that.