We were doing this late after nir_lower_io, but we can just reuse the core code. By doing it at this stage, we won't even set up the VS attributes as inputs, reducing our VPM size.
We were doing this late after nir_lower_io, but we can just reuse the core code. By doing it at this stage, we won't even set up the VS attributes as inputs, reducing our VPM size.