XGBOOST GridSearchCV with GPU_HIST uses both GPU and CPU simultaneously

Question

I am using this code to train an xgboost model on a GPU

The problem is that both GPU (NVIDIA 1050) and CPU cores are being used at the same time. NVIDIA system monitor shows a utilization of 85 to 90% and linux system monitor shows all cores working.

there are two problems here

1 .Why is xgb_cv using both when the tree_method defined is 'gpu_hist'

2. When the model is trained with 'hist' and not 'gpu_hist' it finishes in half the time using CPU cores only

Thanks

model_xgb = XGBClassifier(verbose= 1,objective = 'multi:softmax',num_classes=3,tree_method='gpu_hist',predictor = 'gpu_predictor')

xgb_cv    = GridSearchCV(model_xgb, 
                      {"colsample_bytree":[0.8,0.6]
                        ,"min_child_weight":[0,5 ]
                        ,'max_depth': [3,4,] 
                        ,'n_estimators': [500]
                        ,'learning_rate' :[0.01, 0.1]},cv = 2,verbose = 1)





## Fit with cross validation
start_time = time.time()
xgb_cv.fit(X_train,Y_train,verbose = 1)   
duration = (time.time() - start_time)/60                    
print("XGBOOST HyperParameter Tuning  %s minutes ---" % + duration)