Unigine uses Voxel Cone based GI and not Raytracing, which we already have called VXGI. Here is a paper on Bounding Voume Hierarchies which is what the RT Cores in the 20xx cards mainly accellerate: https://graphics.stanford.edu/~boulo…ient_notes.pdf
Now you can use the compute shaders for that, but it just isn‘t fast enough on top of AMD and Maxwell cards having a lot less raw horsepower than something like a 2080 Ti even without the dedicated processing units.