While we cannot recommend specific models, here are some general considerations:
- Solving of ODE equations is inherently single-threaded. The single-threaded performance of a CPU therefore has the biggest impact on simulation speed. This is roughly equivalent with "more GHz". Additional cores help with certain tasks, e.g. parameter sweeps or code generation.
- The amount of memory needed for a simulation depends on the number of scope signals, the step size and the simulation time range. For long-running simulations with many scopes more RAM is helpful. If the simulation requires more memory than physical RAM, the operating system starts swapping out memory to disk which has a big performance penalty, which can be partially compensated by using a fast SSD.
Having said that, speeding up a simulation is usually a question of improving the model rather than throwing raw CPU power at it.