Flux-tube (local) gyrokinetic codes are widely used to simulate drift-wave turbulence in magnetic confinement devices. While a large number of studies show that flux-tube codes provide an excellent approximation for turbulent transport in medium-large devices, it still needs to be determined whether they are sufficient for modeling supra-thermal particle effects on core turbulence. This is called into question given the large temperature of energetic particles (EPs), which makes them hardly confined on a single flux-surface, but also due to the radially broad mode structure of EP-driven modes. The primary focus of this manuscript is to assess the range of validity of flux-tube codes in modeling fast ion effects by comparing radially global turbulence simulations with flux-tube results at different radial locations for realistic JET parameters using the gyrokinetic code GENE. To extend our study to a broad range of different plasma scenarios, this comparison is made for four different plasma regimes, which differ only by the profile of the ratio between the plasma kinetic and magnetic pressure. The latter is artificially rescaled to address the (i) electrostatic limit and regimes with (ii) marginally stable, (iii) weakly unstable and (iv) strongly unstable fast ion modes. These EP-driven modes are identified as Alfvenic ion temperature gradient modes (AITG)/kinetic beta-induced Alfven eigenmodes (KBAE) via linear ORB5 and LIGKA simulations. It is found that the local flux-tube simulations can recover well the global results only in the electrostatic and marginally stable cases. When the AITG/KBAE becomes linearly unstable, the local approximation fails to correctly model the radially broad fast ion mode structure and the consequent global zonal patterns. According to this study, global turbulence simulations are likely required in regimes with linearly unstable AITG/KBAEs. In conditions with different fast ion-driven modes, these results might change.