docs/tgsi: wrap overly long lines

While we're at it, use some alignment so the equations still reads
reasonably.

Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21893>
This commit is contained in:
Erik Faye-Lund
2023-03-13 16:28:39 +01:00
committed by Erik Faye-Lund
parent d22ee93f97
commit 02908b26bb

View File

@@ -170,7 +170,11 @@ used.
.. math::
dst = src0.x \times src1.x + src0.y \times src1.y + src0.z \times src1.z
\begin{aligned}
dst = & src0.x \times src1.x +\\
& src0.y \times src1.y +\\
& src0.z \times src1.z
\end{aligned}
.. opcode:: DP4 - 4-component Dot Product
@@ -179,7 +183,12 @@ used.
.. math::
dst = src0.x \times src1.x + src0.y \times src1.y + src0.z \times src1.z + src0.w \times src1.w
\begin{aligned}
dst = & src0.x \times src1.x +\\
& src0.y \times src1.y +\\
& src0.z \times src1.z +\\
& src0.w \times src1.w
\end{aligned}
.. opcode:: DST - Distance Vector
@@ -418,8 +427,10 @@ used.
.. math::
dst = f32\_to\_f16(src.x) | f32\_to\_f16(src.y) << 16
\begin{aligned}
dst = & f32\_to\_f16(src.x) | \\
( & f32\_to\_f16(src.y) << 16)
\end{aligned}
.. opcode:: PK2US - Pack Two Unsigned 16-bit Scalars
@@ -427,7 +438,10 @@ used.
.. math::
dst = f32\_to\_unorm16(src.x) | f32\_to\_unorm16(src.y) << 16
\begin{aligned}
dst = & f32\_to\_unorm16(src.x) | \\
( & f32\_to\_unorm16(src.y) << 16)
\end{aligned}
.. opcode:: PK4B - Pack Four Signed 8-bit Scalars
@@ -436,10 +450,12 @@ used.
.. math::
dst = f32\_to\_snorm8(src.x) |
(f32\_to\_snorm8(src.y) << 8) |
(f32\_to\_snorm8(src.z) << 16) |
(f32\_to\_snorm8(src.w) << 24)
\begin{aligned}
dst = & f32\_to\_snorm8(src.x) | \\
( & f32\_to\_snorm8(src.y) << 8) | \\
( & f32\_to\_snorm8(src.z) << 16) | \\
( & f32\_to\_snorm8(src.w) << 24)
\end{aligned}
.. opcode:: PK4UB - Pack Four Unsigned 8-bit Scalars
@@ -448,10 +464,12 @@ used.
.. math::
dst = f32\_to\_unorm8(src.x) |
(f32\_to\_unorm8(src.y) << 8) |
(f32\_to\_unorm8(src.z) << 16) |
(f32\_to\_unorm8(src.w) << 24)
\begin{aligned}
dst = & f32\_to\_unorm8(src.x) | \\
( & f32\_to\_unorm8(src.y) << 8) | \\
( & f32\_to\_unorm8(src.z) << 16) | \\
( & f32\_to\_unorm8(src.w) << 24)
\end{aligned}
.. opcode:: SEQ - Set On Equal
@@ -772,8 +790,10 @@ used.
.. math::
dst = src0.x \times src1.x + src0.y \times src1.y
\begin{aligned}
dst = & src0.x \times src1.x + \\
& src0.y \times src1.y
\end{aligned}
.. opcode:: TEX_LZ - Texture Lookup With LOD = 0