Use of hipMalloc in the "Finite difference method" labs

[AlexisEspinosaGayosso]

Why are you using:
hipMalloc((void**)&d_u, numbytes); hipMalloc((void**)&d_f, numbytes);

instead of the "typical":
hipMalloc(&d_u, numbytes); hipMalloc(&d_f, numbytes);

?

OK. Got it. hipMalloc receives, as first argument a void **ptr. Then, the (void **)&d_u is performing an explicit cast to the type definition of that first argument, rather than letting the compiler perform the casting. I guess the decision of using an explict cast is more a matter of style. Or is there any "AMD GPU guru" hint that make you decide to use the explicit cast?

[jychang48]

@AlexisEspinosaGayosso yes, this was simply a matter of style, the former is compatible with both C and C++ whereas the latter is only compatible with C++, but since the provided example is written in C++, the latter should be sufficient.