In a transient simulation the results of the next time step depend on the results of the current time step. Such a simulation is therefore not well suited for parallel computation.
As a result, the (single thread) speed of your CPU has the biggest impact on simulation time. PLECS profits from multi-core CPUs because it can offload things like scope rendering to another core (or run multiple simulations in parallel on different cores). The performance of your graphics card, however, has no impact on the simulation speed.