GDP should be able to vary at least in factor of 1000.
That is from 300 GDP/capita (poorest countries/after total collapse to pre-industrial level) to 300 000 GDP/capita (ultrawealthy ultra high tech country with serious space economy).
Everything, that affects GDP would affect GDP Growth.
GDP would be numerical value - it would be used to calculate GDP/Capita, which then could be normalized to 0 - 1.
Then this GDP/Capita and GDP growth rate would affect things, that GDP normally affects now.
Increasing GDP/Capita would lower GDP growth rate forcing policies, that favor more growth.
GDP growth would be from 0 to 1 representing -+ 10% yearly growth.
Logarithms and exponents would have to be used to translate between exponentially growing GDP and 0-1 simulation bounds.
Exponential effects can be approximated with Taylor approximation if exponents and logarithms are unsupported directly in equations.
Growth being between ±10% yearly and GDP/c being between 300/300 000 $ (0-1) would cover most of scenarios.
GDP/c simulation would be extremely slow to move:
Not more than 0.01 per year, even if GDP Growth is at extreme value.
GDP and GDP/c could be displayed next to population.