In this investigation, the specific cutting energy (SCE) and average surface roughness (Ra) were decreased using the hard-rotary turning (HRT) factors, including the inclined angle (I), depth of cut (D), feed rate (f), and spindle speed (S). The Bayesian regularized feed-forward neural network was applied to develop the SCE and Ra models. The entropy method and vibration and communication particle swarm optimization (VCPSO) algorithm were employed to compute the weights and determine optimal factors. The optimizing outcomes presented that the optimal I, D, f, and S were 35 deg., 0.45 mm, 0.50 mm/rev., and 1200 rpm, respectively, while the SCE and Ra were decreased by 37.4% and 6.6%, respectively. The total turning cost was saved by 7.5% at the selected solution. The valuable outcomes could be applied to the practical HRT process to decrease performance measures, while the developed HRT operation could be utilized for machining difficult-to-cut materials.