On Friday, November 13, 2015 10:14:56 AM Jason Ekstrand wrote: > Cc: Kenneth Graunke <kenn...@whitecape.org> > > > --- > src/mesa/drivers/dri/i965/brw_fs_generator.cpp | 11 +++++++++++ > 1 file changed, 11 insertions(+) > > diff --git a/src/mesa/drivers/dri/i965/brw_fs_generator.cpp > b/src/mesa/drivers/dri/i965/brw_fs_generator.cpp > index 4877504..61c63d4 100644 > --- a/src/mesa/drivers/dri/i965/brw_fs_generator.cpp > +++ b/src/mesa/drivers/dri/i965/brw_fs_generator.cpp > @@ -690,6 +690,17 @@ fs_generator::generate_tex(fs_inst *inst, struct brw_reg > dst, struct brw_reg src > break; > } > > + /* Stomp the resinfo output type to UINT32. On gens 4-5, the output type > + * is set as part of the message descriptor. On gen4, the PRM seems to > + * allow UINT32 and FLOAT32 (i965 PRM, Vol. 4 Section 4.8.1.1), but on > + * later gens UINT32 is required. Once you hit Sandy Bridge, the bit is > + * gone from the message descriptor entirely and you just get UINT32 all > + * the time regasrdless. Since we can really only do non-UINT32 on gen4, > + * just stomp it to UINT32 all the time. > + */ > + if (inst->opcode == SHADER_OPCODE_TXS) > + return_format = BRW_SAMPLER_RETURN_FORMAT_UINT32; > + > switch (inst->exec_size) { > case 8: > simd_mode = BRW_SAMPLER_SIMD_MODE_SIMD8; >
I like this one better. Reviewed-by: Kenneth Graunke <kenn...@whitecape.org>
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev