Everybody knows that Brook+ generated IL is hard to understand(harder than HLSL or GLSL generated). But I was surprised to see that gpu specific assemler for the same kernel written in HLSL and brook+ is the same. Will the same happen for any kernels or just for simpe kernels. Does newest brook+ use all the feautures included in R7xx assembler or there are a lot of optimization to do by hands. Can I be sure that brook+ code will be compiled properly for any device (no matter R6xx or R7xx) ? And why does brook+ need runtime to be installed (what optimization happens in runtime ?) ? Does brook+ runtime integrated with newest drivers ?
Brook+ and HLSL comparison. SDK 1.3 - questions.