[Mesa-dev] [PATCH] gallium: add INTERP_* opcodes to support interpolateAt*

[Roland Scheidegger]

The only thing I'm wondering is about hw support for offset version
taking a generic float. In particular, it looks like intel graphics
actually uses 4bit immediates for the offset version (just like dictated
by the d3d11 specification).
At a quick glance I couldn't quite see how it's done with southern
islands, but this one has a 4bit int to float instruction "to be used in
shader interpolation" so my guess is it will indeed take any float
natively. Didn't look at nvidia, but a float makes sense then if not all
hw agrees.

Reviewed-by: Roland Scheidegger <sroland at vmware.com>

Am 06.07.2014 07:02, schrieb Ilia Mirkin:
> Signed-off-by: Ilia Mirkin <imirkin at alum.mit.edu>
> ---
> 
> There's an accompanying patch based on ChrisF's unpublished series to add
> interpolateAt* to mesa core that makes use of them, and a further couple of
> patches which inmplement this on nvc0.
> 
> With all that, nvc0 passes all the piglits in ChrisF's repo except for the
> interpolateAtCentroid-flat one, which fails on the blob driver in exactly the
> same way (and for a reason totally unrelated to these new opcodes).
> 
>  src/gallium/auxiliary/tgsi/tgsi_info.c     |  4 ++++
>  src/gallium/docs/source/tgsi.rst           | 23 +++++++++++++++++++++++
>  src/gallium/include/pipe/p_shader_tokens.h |  6 +++++-
>  3 files changed, 32 insertions(+), 1 deletion(-)
> 
> diff --git a/src/gallium/auxiliary/tgsi/tgsi_info.c b/src/gallium/auxiliary/tgsi/tgsi_info.c
> index e2c4e65..e24348f 100644
> --- a/src/gallium/auxiliary/tgsi/tgsi_info.c
> +++ b/src/gallium/auxiliary/tgsi/tgsi_info.c
> @@ -231,6 +231,10 @@ static const struct tgsi_opcode_info opcode_info[TGSI_OPCODE_LAST] =
>     { 1, 1, 0, 0, 0, 0, COMP, "LSB", TGSI_OPCODE_LSB },
>     { 1, 1, 0, 0, 0, 0, COMP, "IMSB", TGSI_OPCODE_IMSB },
>     { 1, 1, 0, 0, 0, 0, COMP, "UMSB", TGSI_OPCODE_UMSB },
> +
> +   { 1, 1, 0, 0, 0, 0, OTHR, "INTERP_CENTROID", TGSI_OPCODE_INTERP_CENTROID },
> +   { 1, 2, 0, 0, 0, 0, OTHR, "INTERP_SAMPLE", TGSI_OPCODE_INTERP_SAMPLE },
> +   { 1, 2, 0, 0, 0, 0, OTHR, "INTERP_OFFSET", TGSI_OPCODE_INTERP_OFFSET },
>  };
>  
>  const struct tgsi_opcode_info *
> diff --git a/src/gallium/docs/source/tgsi.rst b/src/gallium/docs/source/tgsi.rst
> index 4213b4a..3f4f0e6 100644
> --- a/src/gallium/docs/source/tgsi.rst
> +++ b/src/gallium/docs/source/tgsi.rst
> @@ -1808,6 +1808,29 @@ This instruction replicates its result.
>    dst = \frac{src.x}{src.x \times src.x + src.y \times src.y + src.z \times src.z + src.w \times src.w}
>  
>  
> +Interpolation ISA
> +^^^^^^^^^^^^^^^^^
> +
> +The interpolation instructions allow an input to be interpolated in a
> +different way than its declaration. This corresponds to the GLSL 4.00
> +interpolateAt* functions. The first argument of each of these must come from
> +``TGSI_FILE_INPUT``.
> +
> +.. opcode:: INTERP_CENTROID - Interpolate at the centroid
> +
> +   Interpolates the varying specified by src0 at the centroid
> +
> +.. opcode:: INTERP_SAMPLE - Interpolate at the specified sample
> +
> +   Interpolates the varying specified by src0 at the sample id specified by
> +   src1.x (interpreted as an integer)
> +
> +.. opcode:: INTERP_OFFSET - Interpolate at the specified offset
> +
> +   Interpolates the varying specified by src0 at the offset src1.xy from the
> +   pixel center (interpreted as floats)
> +
> +
>  .. _doubleopcodes:
>  
>  Double ISA
> diff --git a/src/gallium/include/pipe/p_shader_tokens.h b/src/gallium/include/pipe/p_shader_tokens.h
> index e68258d..0d3ad6a 100644
> --- a/src/gallium/include/pipe/p_shader_tokens.h
> +++ b/src/gallium/include/pipe/p_shader_tokens.h
> @@ -477,7 +477,11 @@ struct tgsi_property_data {
>  #define TGSI_OPCODE_IMSB                190
>  #define TGSI_OPCODE_UMSB                191
>  
> -#define TGSI_OPCODE_LAST                192
> +#define TGSI_OPCODE_INTERP_CENTROID     192
> +#define TGSI_OPCODE_INTERP_SAMPLE       193
> +#define TGSI_OPCODE_INTERP_OFFSET       194
> +
> +#define TGSI_OPCODE_LAST                195
>  
>  #define TGSI_SAT_NONE            0  /* do not saturate */
>  #define TGSI_SAT_ZERO_ONE        1  /* clamp to [0,1] */
>