Okay, but I think these SM1 tests should be part of a different MR, since we can't specify those requirements in the shader_runner yet. Perhaps they should be in a different file too.
That's fine, as long as we do indeed get them in at some point soon (and as long as the implementation doesn't contradict them).