Over the last two decades computational engineering has gained credibility for virtually testing novel technological solutions (prostheses or surgical procedures) with the advantage of providing potentially unlimited access to hemodynamics data [1]. Furthermore, computer simulations allow to run ideal tests in which some parameters of the cardiac configuration are varied continuously while fixing the others so to isolate their effect, which would be exceedingly difficult or impossible in regular trials involving humans. To this aim, our group has developed a GPU accelerated cardiac model to accurately solve cardiovascular flows [2], which can cope with the electrophysiology of the myocardium, including the fibers orientation, its active contraction and passive relaxation, the dynamics of the valves and the hemodynamics within the heart chambers and arteries. All these models are three-way coupled with each other, thus capturing the fully synergistic physics of the heart functioning. Specifically, the pulsatile and transitional character of the hemodynamics is obtained by solving directly the incompressible Navier-Stokes equations comple- mented with immersed boundary (IB) techniques to handle complex moving and deforming geometries. The structural mechanics is based on the Fedosov’s interaction potential approach to account for the or- thotropic and nonlinear properties of the biological tissues. The electrophysiology, responsible for the active potential propagation in the cardiac tissue triggering the muscular contraction, is incorporated by means of a bidomain model accounting for the heterogeneous physiologic properties of the myocardium. This computational model applied to solve the normal cardiac dynamics is currently being applied to in- vestigate pathologies and optimize cardiac devices, such as the cardiac resynchronization therapy (CRT), which is a surgical treatment to restore the normal timing pattern of the heartbeat in patients with heart failure.