Kenneth Graunke <kenn...@whitecape.org> writes: > The way nir_lower_clip_vs() works with store_output intrinsics makes a > ton of assumptions about the driver_location field. > > In i965, I'd rather do this lowering early and work with variables. > ir3 and vc4 could probably do that as well, but I'm not sure exactly > what paths would need updating, so for now, handle both methods. > --- > src/compiler/nir/nir.h | 2 +- > src/compiler/nir/nir_lower_clip.c | 45 ++++++++++++++++----- > src/gallium/drivers/freedreno/ir3/ir3_nir.c | 2 +- > src/gallium/drivers/vc4/vc4_program.c | 3 +-
If you hit src/broadcom/compiler/vir.c as well, the series is: Reviewed-by: Eric Anholt <e...@anholt.net> Switching vc4/v3d stuff over to variables has been going well and showing wins -- I think we should probably finish that work and then simplify these paths by moving this lowering before lower_io. In the past, I thought we would want to lower_io at gallium state creation time, since it was some compile work we could do at compile/link instead of draw.
signature.asc
Description: PGP signature
_______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev