Do some template magic so we don't have to cast to void**

[jakub-homola]

The following example line of code

hipMalloc(&d_x, count * sizeof(float));

fails to compile (using g++ 9.4.0) with error

saxpy.hip.cpp:46:15: error: invalid conversion from ‘float**’ to ‘void**’ [-fpermissive]
   46 |     hipMalloc(&d_x, count * sizeof(float));
      |               ^~~~
      |               |
      |               float**

I know the fix is to just cast the pointer,

hipMalloc((void**)&d_x, count * sizeof(float));

but who likes that? CUDA and HIP don't require the cast, so neither should HIP-CPU. In the classic HIP (include/hip_runtime_api.h right at the bottom in 4.3) there is the function

template <class T>
static inline hipError_t hipMalloc(T** devPtr, size_t size) {
    return hipMalloc((void**)devPtr, size);
}

so please do something similar in HIP-CPU too, for all similar functions. If there are any cons to this, please explain or point me to an explanation.

Thanks, Jakub Homola

Edit: I am using the current HIP-CPU master branch, changelog says version 0.1.4142 from December 2020