I am curious. Why do you think it helps it doing things on the stack. For me, I understand this to provide more generic type level checks on the size/shape of values. Perhaps, I am oversimplifying this?
Basically, the const-generics feature gives the programmer the ability to create fixed-size custom types that doesn't allocate on the heap. For example, with current Rust you can't make your own generic fixed-size hashmap (as in FixedSizeHashMap<K, V, N: uint>, where N is the maximum number of keys you can use) that doesn't use the heap, since you can't use size-parameterized arrays in structs. Although this isn't a big deal for most general programming purposes, it still matters a lot in domains like embedded/HFT/gamedev/simulation where minimal latency is incredibly important. It's also important when you want to create any sort of linear algebra library (such as Eigen). For me const-generics is an important dealbreaker when choosing between Rust and C++ (where C++ already had the ability to do this for decades).