Kenneth Graunke <kenn...@whitecape.org> writes: > Shamelessly ripped off from Eric Anholt's tgsi_to_nir pass. > > Not compiled on SCons, like the rest of NIR. > > Signed-off-by: Kenneth Graunke <kenn...@whitecape.org>
Nice. I think at some point we'll want a lowering pass for texture projection. The recommendation to do fsub in algebraic instead of i965 sounded good. The use of nir variables here also looks way less unpleasant than my first interaction with them was, so I may take a look at it again for my pass. This patch is: Reviewed-by: Eric Anholt <e...@anholt.net> Possible improvement: ptn_imm_f()/d() could get a scalar value instead of a vector. I found that this was a win (now that nir_builder can handle scalars mixed with vectors reasonably). Similarly, ttn_channel() returning a scalar helped.
signature.asc
Description: PGP signature
_______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
