What is the optimal way to split up the cores (n_jobs) between an sklearn model and RandomizedSearch?

Question

I am trying to figure out how to use the cores that I have available most effectively for a random forest that is being tuned using RandomizedSearchCV. Is it enough to specify n_jobs = -1 in the RandomizedSearch, should I include this in the model itself as well, or should I even divide up the cores between the two (e.g., if I have 32 cores available, n_jobs = 16 in the model and n_jobs = 16 in the RandomizedSearch)?

This is an excerpt of my code:

rf_model = RandomForestClassifier()

tuning_grid_rf = {'n_estimators': [500, 1000, 1500],
               'max_features': ['auto', 'sqrt'],
               'max_depth': [10, 20, 30],
               'min_samples_split': [1, 3, 5],
               'min_samples_leaf': [1, 3, 5],
               'max_samples': [0.5, 0.75, None]}

rf_tuning = RandomizedSearchCV(estimator = rf_model, 
                               param_distributions = tuning_grid_rf, 
                               n_iter = 50, 
                               cv = 5, 
                               verbose = 2,
                               n_jobs = -1,
                               random_state = 42)

history_rf = rf_tuning.fit(X_train, y_train)

This is my first question here, so please let me know if I have left out any relevant information.