Are you talking seriously???
I'm a professional structural engineer and let me say I know what is FEM program, I use professional programs (like Strand7) almost every day and occasionally I wrote them too for my own professional work (I do not sell software).
Believe me or not, I can assure you that:
- "simple" linear elastic analysis of beams or shells connected in a model let's say of some hundreds nodes, is computed by a modern CPU in less than a tenth of a second. Not entering too much in detail, but, the fem method applied in vehicle dynamics is done with a real-time linear time-history analysis that could be implemented also in a excel spreadsheet. The problems arises with non-linear analysis, where many iterations may be necessary to solve the equilibrium equations of just a sampled instant of time, but this analysis are necessary to model ductile behaviour of structures after the yield stress and is definitely not the case of vehicle dynamics.
- it's definitely not true that a more sofisticated finite element model is always better of a simpler one
- the precision of the results of a finite element analysis are more close-fitting to reality depending on the accuracy of the costitutive models of the materials (for elastic materials: Young modulus and Poisson ratio) but most of all the boundary conditions (the inertial forces transmitted by the vehicle and the asphalt roughness & bumps)
- it's a matter of fact in the engineering practice that simple no-frills FEM models with few well-known parameters give in general more reliable results than over-detailed, complicated, huge models (in structural engineering we call them "monsters") where the analyst has to input a lot of parameters not completely mastered.
That said, your guesswork that a sim that do not stress much the CPU has a poor tyre model is completely ridiculous.