PR's #954 and #1459 are super-sets of this request. However, this smaller request is to have available in a compute shader the functions anyInvocation() and allInvocation() in compute shader. For certain cases (for example mine) these make a big difference for performance, especially to prevent divergence within an SIMD group.