Samuel sent out a patch to fix this already which I've reviewed. Feel free to merge it. I gave him comments on the second so maybe best to leave that one be.
That said, I do sort-of like your macro... On Wed, Mar 14, 2018 at 8:08 AM, Rob Clark <robdcl...@gmail.com> wrote: > These opcodes don't care about src/dst type, only src/dst size. > > Signed-off-by: Rob Clark <robdcl...@gmail.com> > --- > src/compiler/spirv/vtn_alu.c | 14 ++++++++++---- > 1 file changed, 10 insertions(+), 4 deletions(-) > > diff --git a/src/compiler/spirv/vtn_alu.c b/src/compiler/spirv/vtn_alu.c > index f0b69b38f83..1ca1f951200 100644 > --- a/src/compiler/spirv/vtn_alu.c > +++ b/src/compiler/spirv/vtn_alu.c > @@ -349,11 +349,17 @@ vtn_nir_alu_op_for_spirv_opcode(struct vtn_builder > *b, > > /* Conversions: */ > case SpvOpQuantizeToF16: return nir_op_fquantize2f16; > + > +#define nir_type_cast(type, basetype) (nir_alu_type_get_type_size(type) > | nir_type_ ## basetype) > +#define typed_conversion_op(src, srctype, dst, dsttype) \ > + nir_type_conversion_op(nir_type_cast(src, srctype), > nir_type_cast(dst, dsttype), nir_rounding_mode_undef) > + > + case SpvOpConvertFToU: return typed_conversion_op(src, float, dst, > uint); > + case SpvOpConvertFToS: return typed_conversion_op(src, float, dst, > int); > + case SpvOpConvertSToF: return typed_conversion_op(src, int, dst, > float); > + case SpvOpConvertUToF: return typed_conversion_op(src, uint, dst, > float); > + > case SpvOpUConvert: > - case SpvOpConvertFToU: > - case SpvOpConvertFToS: > - case SpvOpConvertSToF: > - case SpvOpConvertUToF: > case SpvOpSConvert: > case SpvOpFConvert: > return nir_type_conversion_op(src, dst, nir_rounding_mode_undef); > -- > 2.14.3 > >
_______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev
