4 TGSI, Tungsten Graphics Shader Infrastructure, is an intermediate language
5 for describing shaders. Since Gallium is inherently shaderful, shaders are
6 an important part of the API. TGSI is the only intermediate representation
12 All TGSI instructions, known as *opcodes*, operate on arbitrary-precision
13 floating-point four-component vectors. An opcode may have up to one
14 destination register, known as *dst*, and between zero and three source
15 registers, called *src0* through *src2*, or simply *src* if there is only
18 Some instructions, like :opcode:`I2F`, permit re-interpretation of vector
19 components as integers. Other instructions permit using registers as
20 two-component vectors with double precision; see :ref:`doubleopcodes`.
22 When an instruction has a scalar result, the result is usually copied into
23 each of the components of *dst*. When this happens, the result is said to be
24 *replicated* to *dst*. :opcode:`RCP` is one such instruction.
29 TGSI supports modifiers on inputs (as well as saturate and precise modifier
32 For arithmetic instruction having a precise modifier certain optimizations
33 which may alter the result are disallowed. Example: *add(mul(a,b),c)* can't be
34 optimized to TGSI_OPCODE_MAD, because some hardware only supports the fused
37 For inputs which have a floating point type, both absolute value and
38 negation modifiers are supported (with absolute value being applied
39 first). The only source of TGSI_OPCODE_MOV and the second and third
40 sources of TGSI_OPCODE_UCMP are considered to have float type for
43 For inputs which have signed or unsigned type only the negate modifier is
50 ^^^^^^^^^^^^^^^^^^^^^^^^^
52 These opcodes are guaranteed to be available regardless of the driver being
55 .. opcode:: ARL - Address Register Load
59 dst.x = (int) \lfloor src.x\rfloor
61 dst.y = (int) \lfloor src.y\rfloor
63 dst.z = (int) \lfloor src.z\rfloor
65 dst.w = (int) \lfloor src.w\rfloor
68 .. opcode:: MOV - Move
81 .. opcode:: LIT - Light Coefficients
86 dst.y &= max(src.x, 0) \\
87 dst.z &= (src.x > 0) ? max(src.y, 0)^{clamp(src.w, -128, 128))} : 0 \\
91 .. opcode:: RCP - Reciprocal
93 This instruction replicates its result.
100 .. opcode:: RSQ - Reciprocal Square Root
102 This instruction replicates its result. The results are undefined for src <= 0.
106 dst = \frac{1}{\sqrt{src.x}}
109 .. opcode:: SQRT - Square Root
111 This instruction replicates its result. The results are undefined for src < 0.
118 .. opcode:: EXP - Approximate Exponential Base 2
122 dst.x &= 2^{\lfloor src.x\rfloor} \\
123 dst.y &= src.x - \lfloor src.x\rfloor \\
124 dst.z &= 2^{src.x} \\
128 .. opcode:: LOG - Approximate Logarithm Base 2
132 dst.x &= \lfloor\log_2{|src.x|}\rfloor \\
133 dst.y &= \frac{|src.x|}{2^{\lfloor\log_2{|src.x|}\rfloor}} \\
134 dst.z &= \log_2{|src.x|} \\
138 .. opcode:: MUL - Multiply
142 dst.x = src0.x \times src1.x
144 dst.y = src0.y \times src1.y
146 dst.z = src0.z \times src1.z
148 dst.w = src0.w \times src1.w
151 .. opcode:: ADD - Add
155 dst.x = src0.x + src1.x
157 dst.y = src0.y + src1.y
159 dst.z = src0.z + src1.z
161 dst.w = src0.w + src1.w
164 .. opcode:: DP3 - 3-component Dot Product
166 This instruction replicates its result.
170 dst = src0.x \times src1.x + src0.y \times src1.y + src0.z \times src1.z
173 .. opcode:: DP4 - 4-component Dot Product
175 This instruction replicates its result.
179 dst = src0.x \times src1.x + src0.y \times src1.y + src0.z \times src1.z + src0.w \times src1.w
182 .. opcode:: DST - Distance Vector
187 dst.y &= src0.y \times src1.y\\
192 .. opcode:: MIN - Minimum
196 dst.x = min(src0.x, src1.x)
198 dst.y = min(src0.y, src1.y)
200 dst.z = min(src0.z, src1.z)
202 dst.w = min(src0.w, src1.w)
205 .. opcode:: MAX - Maximum
209 dst.x = max(src0.x, src1.x)
211 dst.y = max(src0.y, src1.y)
213 dst.z = max(src0.z, src1.z)
215 dst.w = max(src0.w, src1.w)
218 .. opcode:: SLT - Set On Less Than
222 dst.x = (src0.x < src1.x) ? 1.0F : 0.0F
224 dst.y = (src0.y < src1.y) ? 1.0F : 0.0F
226 dst.z = (src0.z < src1.z) ? 1.0F : 0.0F
228 dst.w = (src0.w < src1.w) ? 1.0F : 0.0F
231 .. opcode:: SGE - Set On Greater Equal Than
235 dst.x = (src0.x >= src1.x) ? 1.0F : 0.0F
237 dst.y = (src0.y >= src1.y) ? 1.0F : 0.0F
239 dst.z = (src0.z >= src1.z) ? 1.0F : 0.0F
241 dst.w = (src0.w >= src1.w) ? 1.0F : 0.0F
244 .. opcode:: MAD - Multiply And Add
246 Perform a * b + c. The implementation is free to decide whether there is an
247 intermediate rounding step or not.
251 dst.x = src0.x \times src1.x + src2.x
253 dst.y = src0.y \times src1.y + src2.y
255 dst.z = src0.z \times src1.z + src2.z
257 dst.w = src0.w \times src1.w + src2.w
260 .. opcode:: LRP - Linear Interpolate
264 dst.x = src0.x \times src1.x + (1 - src0.x) \times src2.x
266 dst.y = src0.y \times src1.y + (1 - src0.y) \times src2.y
268 dst.z = src0.z \times src1.z + (1 - src0.z) \times src2.z
270 dst.w = src0.w \times src1.w + (1 - src0.w) \times src2.w
273 .. opcode:: FMA - Fused Multiply-Add
275 Perform a * b + c with no intermediate rounding step.
279 dst.x = src0.x \times src1.x + src2.x
281 dst.y = src0.y \times src1.y + src2.y
283 dst.z = src0.z \times src1.z + src2.z
285 dst.w = src0.w \times src1.w + src2.w
288 .. opcode:: FRC - Fraction
292 dst.x = src.x - \lfloor src.x\rfloor
294 dst.y = src.y - \lfloor src.y\rfloor
296 dst.z = src.z - \lfloor src.z\rfloor
298 dst.w = src.w - \lfloor src.w\rfloor
301 .. opcode:: FLR - Floor
305 dst.x = \lfloor src.x\rfloor
307 dst.y = \lfloor src.y\rfloor
309 dst.z = \lfloor src.z\rfloor
311 dst.w = \lfloor src.w\rfloor
314 .. opcode:: ROUND - Round
327 .. opcode:: EX2 - Exponential Base 2
329 This instruction replicates its result.
336 .. opcode:: LG2 - Logarithm Base 2
338 This instruction replicates its result.
345 .. opcode:: POW - Power
347 This instruction replicates its result.
351 dst = src0.x^{src1.x}
353 .. opcode:: XPD - Cross Product
357 dst.x = src0.y \times src1.z - src1.y \times src0.z
359 dst.y = src0.z \times src1.x - src1.z \times src0.x
361 dst.z = src0.x \times src1.y - src1.x \times src0.y
366 .. opcode:: DPH - Homogeneous Dot Product
368 This instruction replicates its result.
372 dst = src0.x \times src1.x + src0.y \times src1.y + src0.z \times src1.z + src1.w
375 .. opcode:: COS - Cosine
377 This instruction replicates its result.
384 .. opcode:: DDX, DDX_FINE - Derivative Relative To X
386 The fine variant is only used when ``PIPE_CAP_TGSI_FS_FINE_DERIVATIVE`` is
387 advertised. When it is, the fine version guarantees one derivative per row
388 while DDX is allowed to be the same for the entire 2x2 quad.
392 dst.x = partialx(src.x)
394 dst.y = partialx(src.y)
396 dst.z = partialx(src.z)
398 dst.w = partialx(src.w)
401 .. opcode:: DDY, DDY_FINE - Derivative Relative To Y
403 The fine variant is only used when ``PIPE_CAP_TGSI_FS_FINE_DERIVATIVE`` is
404 advertised. When it is, the fine version guarantees one derivative per column
405 while DDY is allowed to be the same for the entire 2x2 quad.
409 dst.x = partialy(src.x)
411 dst.y = partialy(src.y)
413 dst.z = partialy(src.z)
415 dst.w = partialy(src.w)
418 .. opcode:: PK2H - Pack Two 16-bit Floats
420 This instruction replicates its result.
424 dst = f32\_to\_f16(src.x) | f32\_to\_f16(src.y) << 16
427 .. opcode:: PK2US - Pack Two Unsigned 16-bit Scalars
432 .. opcode:: PK4B - Pack Four Signed 8-bit Scalars
437 .. opcode:: PK4UB - Pack Four Unsigned 8-bit Scalars
442 .. opcode:: SEQ - Set On Equal
446 dst.x = (src0.x == src1.x) ? 1.0F : 0.0F
448 dst.y = (src0.y == src1.y) ? 1.0F : 0.0F
450 dst.z = (src0.z == src1.z) ? 1.0F : 0.0F
452 dst.w = (src0.w == src1.w) ? 1.0F : 0.0F
455 .. opcode:: SGT - Set On Greater Than
459 dst.x = (src0.x > src1.x) ? 1.0F : 0.0F
461 dst.y = (src0.y > src1.y) ? 1.0F : 0.0F
463 dst.z = (src0.z > src1.z) ? 1.0F : 0.0F
465 dst.w = (src0.w > src1.w) ? 1.0F : 0.0F
468 .. opcode:: SIN - Sine
470 This instruction replicates its result.
477 .. opcode:: SLE - Set On Less Equal Than
481 dst.x = (src0.x <= src1.x) ? 1.0F : 0.0F
483 dst.y = (src0.y <= src1.y) ? 1.0F : 0.0F
485 dst.z = (src0.z <= src1.z) ? 1.0F : 0.0F
487 dst.w = (src0.w <= src1.w) ? 1.0F : 0.0F
490 .. opcode:: SNE - Set On Not Equal
494 dst.x = (src0.x != src1.x) ? 1.0F : 0.0F
496 dst.y = (src0.y != src1.y) ? 1.0F : 0.0F
498 dst.z = (src0.z != src1.z) ? 1.0F : 0.0F
500 dst.w = (src0.w != src1.w) ? 1.0F : 0.0F
503 .. opcode:: TEX - Texture Lookup
505 for array textures src0.y contains the slice for 1D,
506 and src0.z contain the slice for 2D.
508 for shadow textures with no arrays (and not cube map),
509 src0.z contains the reference value.
511 for shadow textures with arrays, src0.z contains
512 the reference value for 1D arrays, and src0.w contains
513 the reference value for 2D arrays and cube maps.
515 for cube map array shadow textures, the reference value
516 cannot be passed in src0.w, and TEX2 must be used instead.
522 shadow_ref = src0.z or src0.w (optional)
526 dst = texture\_sample(unit, coord, shadow_ref)
529 .. opcode:: TEX2 - Texture Lookup (for shadow cube map arrays only)
531 this is the same as TEX, but uses another reg to encode the
542 dst = texture\_sample(unit, coord, shadow_ref)
547 .. opcode:: TXD - Texture Lookup with Derivatives
559 dst = texture\_sample\_deriv(unit, coord, ddx, ddy)
562 .. opcode:: TXP - Projective Texture Lookup
566 coord.x = src0.x / src0.w
568 coord.y = src0.y / src0.w
570 coord.z = src0.z / src0.w
576 dst = texture\_sample(unit, coord)
579 .. opcode:: UP2H - Unpack Two 16-Bit Floats
583 dst.x = f16\_to\_f32(src0.x \& 0xffff)
585 dst.y = f16\_to\_f32(src0.x >> 16)
587 dst.z = f16\_to\_f32(src0.x \& 0xffff)
589 dst.w = f16\_to\_f32(src0.x >> 16)
593 Considered for removal.
595 .. opcode:: UP2US - Unpack Two Unsigned 16-Bit Scalars
601 Considered for removal.
603 .. opcode:: UP4B - Unpack Four Signed 8-Bit Values
609 Considered for removal.
611 .. opcode:: UP4UB - Unpack Four Unsigned 8-Bit Scalars
617 Considered for removal.
620 .. opcode:: ARR - Address Register Load With Round
624 dst.x = (int) round(src.x)
626 dst.y = (int) round(src.y)
628 dst.z = (int) round(src.z)
630 dst.w = (int) round(src.w)
633 .. opcode:: SSG - Set Sign
637 dst.x = (src.x > 0) ? 1 : (src.x < 0) ? -1 : 0
639 dst.y = (src.y > 0) ? 1 : (src.y < 0) ? -1 : 0
641 dst.z = (src.z > 0) ? 1 : (src.z < 0) ? -1 : 0
643 dst.w = (src.w > 0) ? 1 : (src.w < 0) ? -1 : 0
646 .. opcode:: CMP - Compare
650 dst.x = (src0.x < 0) ? src1.x : src2.x
652 dst.y = (src0.y < 0) ? src1.y : src2.y
654 dst.z = (src0.z < 0) ? src1.z : src2.z
656 dst.w = (src0.w < 0) ? src1.w : src2.w
659 .. opcode:: KILL_IF - Conditional Discard
661 Conditional discard. Allowed in fragment shaders only.
665 if (src.x < 0 || src.y < 0 || src.z < 0 || src.w < 0)
670 .. opcode:: KILL - Discard
672 Unconditional discard. Allowed in fragment shaders only.
675 .. opcode:: SCS - Sine Cosine
688 .. opcode:: TXB - Texture Lookup With Bias
690 for cube map array textures and shadow cube maps, the bias value
691 cannot be passed in src0.w, and TXB2 must be used instead.
693 if the target is a shadow texture, the reference value is always
694 in src.z (this prevents shadow 3d and shadow 2d arrays from
695 using this instruction, but this is not needed).
711 dst = texture\_sample(unit, coord, bias)
714 .. opcode:: TXB2 - Texture Lookup With Bias (some cube maps only)
716 this is the same as TXB, but uses another reg to encode the
717 lod bias value for cube map arrays and shadow cube maps.
718 Presumably shadow 2d arrays and shadow 3d targets could use
719 this encoding too, but this is not legal.
721 shadow cube map arrays are neither possible nor required.
731 dst = texture\_sample(unit, coord, bias)
734 .. opcode:: DIV - Divide
738 dst.x = \frac{src0.x}{src1.x}
740 dst.y = \frac{src0.y}{src1.y}
742 dst.z = \frac{src0.z}{src1.z}
744 dst.w = \frac{src0.w}{src1.w}
747 .. opcode:: DP2 - 2-component Dot Product
749 This instruction replicates its result.
753 dst = src0.x \times src1.x + src0.y \times src1.y
756 .. opcode:: TEX_LZ - Texture Lookup With LOD = 0
758 This is the same as TXL with LOD = 0. Like every texture opcode, it obeys
759 pipe_sampler_view::u.tex.first_level and pipe_sampler_state::min_lod.
760 There is no way to override those two in shaders.
776 dst = texture\_sample(unit, coord, lod)
779 .. opcode:: TXL - Texture Lookup With explicit LOD
781 for cube map array textures, the explicit lod value
782 cannot be passed in src0.w, and TXL2 must be used instead.
784 if the target is a shadow texture, the reference value is always
785 in src.z (this prevents shadow 3d / 2d array / cube targets from
786 using this instruction, but this is not needed).
802 dst = texture\_sample(unit, coord, lod)
805 .. opcode:: TXL2 - Texture Lookup With explicit LOD (for cube map arrays only)
807 this is the same as TXL, but uses another reg to encode the
809 Presumably shadow 3d / 2d array / cube targets could use
810 this encoding too, but this is not legal.
812 shadow cube map arrays are neither possible nor required.
822 dst = texture\_sample(unit, coord, lod)
826 ^^^^^^^^^^^^^^^^^^^^^^^^
828 These opcodes are primarily provided for special-use computational shaders.
829 Support for these opcodes indicated by a special pipe capability bit (TBD).
831 XXX doesn't look like most of the opcodes really belong here.
833 .. opcode:: CEIL - Ceiling
837 dst.x = \lceil src.x\rceil
839 dst.y = \lceil src.y\rceil
841 dst.z = \lceil src.z\rceil
843 dst.w = \lceil src.w\rceil
846 .. opcode:: TRUNC - Truncate
859 .. opcode:: MOD - Modulus
863 dst.x = src0.x \bmod src1.x
865 dst.y = src0.y \bmod src1.y
867 dst.z = src0.z \bmod src1.z
869 dst.w = src0.w \bmod src1.w
872 .. opcode:: UARL - Integer Address Register Load
874 Moves the contents of the source register, assumed to be an integer, into the
875 destination register, which is assumed to be an address (ADDR) register.
878 .. opcode:: TXF - Texel Fetch
880 As per NV_gpu_shader4, extract a single texel from a specified texture
881 image or PIPE_BUFFER resource. The source sampler may not be a CUBE or
883 four-component signed integer vector used to identify the single texel
884 accessed. 3 components + level. If the texture is multisampled, then
885 the fourth component indicates the sample, not the mipmap level.
886 Just like texture instructions, an optional
887 offset vector is provided, which is subject to various driver restrictions
888 (regarding range, source of offsets). This instruction ignores the sampler
891 TXF(uint_vec coord, int_vec offset).
894 .. opcode:: TXQ - Texture Size Query
896 As per NV_gpu_program4, retrieve the dimensions of the texture depending on
897 the target. For 1D (width), 2D/RECT/CUBE (width, height), 3D (width, height,
898 depth), 1D array (width, layers), 2D array (width, height, layers).
899 Also return the number of accessible levels (last_level - first_level + 1)
902 For components which don't return a resource dimension, their value
909 dst.x = texture\_width(unit, lod)
911 dst.y = texture\_height(unit, lod)
913 dst.z = texture\_depth(unit, lod)
915 dst.w = texture\_levels(unit)
918 .. opcode:: TXQS - Texture Samples Query
920 This retrieves the number of samples in the texture, and stores it
921 into the x component as an unsigned integer. The other components are
922 undefined. If the texture is not multisampled, this function returns
923 (1, undef, undef, undef).
927 dst.x = texture\_samples(unit)
930 .. opcode:: TG4 - Texture Gather
932 As per ARB_texture_gather, gathers the four texels to be used in a bi-linear
933 filtering operation and packs them into a single register. Only works with
934 2D, 2D array, cubemaps, and cubemaps arrays. For 2D textures, only the
935 addressing modes of the sampler and the top level of any mip pyramid are
936 used. Set W to zero. It behaves like the TEX instruction, but a filtered
937 sample is not generated. The four samples that contribute to filtering are
938 placed into xyzw in clockwise order, starting with the (u,v) texture
939 coordinate delta at the following locations (-, +), (+, +), (+, -), (-, -),
940 where the magnitude of the deltas are half a texel.
942 PIPE_CAP_TEXTURE_SM5 enhances this instruction to support shadow per-sample
943 depth compares, single component selection, and a non-constant offset. It
944 doesn't allow support for the GL independent offset to get i0,j0. This would
945 require another CAP is hw can do it natively. For now we lower that before
954 dst = texture\_gather4 (unit, coord, component)
956 (with SM5 - cube array shadow)
964 dst = texture\_gather (uint, coord, compare)
966 .. opcode:: LODQ - level of detail query
968 Compute the LOD information that the texture pipe would use to access the
969 texture. The Y component contains the computed LOD lambda_prime. The X
970 component contains the LOD that will be accessed, based on min/max lod's
977 dst.xy = lodq(uint, coord);
979 .. opcode:: CLOCK - retrieve the current shader time
981 Invoking this instruction multiple times in the same shader should
982 cause monotonically increasing values to be returned. The values
983 are implicitly 64-bit, so if fewer than 64 bits of precision are
984 available, to provide expected wraparound semantics, the value
985 should be shifted up so that the most significant bit of the time
986 is the most significant bit of the 64-bit value.
994 ^^^^^^^^^^^^^^^^^^^^^^^^
995 These opcodes are used for integer operations.
996 Support for these opcodes indicated by PIPE_SHADER_CAP_INTEGERS (all of them?)
999 .. opcode:: I2F - Signed Integer To Float
1001 Rounding is unspecified (round to nearest even suggested).
1005 dst.x = (float) src.x
1007 dst.y = (float) src.y
1009 dst.z = (float) src.z
1011 dst.w = (float) src.w
1014 .. opcode:: U2F - Unsigned Integer To Float
1016 Rounding is unspecified (round to nearest even suggested).
1020 dst.x = (float) src.x
1022 dst.y = (float) src.y
1024 dst.z = (float) src.z
1026 dst.w = (float) src.w
1029 .. opcode:: F2I - Float to Signed Integer
1031 Rounding is towards zero (truncate).
1032 Values outside signed range (including NaNs) produce undefined results.
1045 .. opcode:: F2U - Float to Unsigned Integer
1047 Rounding is towards zero (truncate).
1048 Values outside unsigned range (including NaNs) produce undefined results.
1052 dst.x = (unsigned) src.x
1054 dst.y = (unsigned) src.y
1056 dst.z = (unsigned) src.z
1058 dst.w = (unsigned) src.w
1061 .. opcode:: UADD - Integer Add
1063 This instruction works the same for signed and unsigned integers.
1064 The low 32bit of the result is returned.
1068 dst.x = src0.x + src1.x
1070 dst.y = src0.y + src1.y
1072 dst.z = src0.z + src1.z
1074 dst.w = src0.w + src1.w
1077 .. opcode:: UMAD - Integer Multiply And Add
1079 This instruction works the same for signed and unsigned integers.
1080 The multiplication returns the low 32bit (as does the result itself).
1084 dst.x = src0.x \times src1.x + src2.x
1086 dst.y = src0.y \times src1.y + src2.y
1088 dst.z = src0.z \times src1.z + src2.z
1090 dst.w = src0.w \times src1.w + src2.w
1093 .. opcode:: UMUL - Integer Multiply
1095 This instruction works the same for signed and unsigned integers.
1096 The low 32bit of the result is returned.
1100 dst.x = src0.x \times src1.x
1102 dst.y = src0.y \times src1.y
1104 dst.z = src0.z \times src1.z
1106 dst.w = src0.w \times src1.w
1109 .. opcode:: IMUL_HI - Signed Integer Multiply High Bits
1111 The high 32bits of the multiplication of 2 signed integers are returned.
1115 dst.x = (src0.x \times src1.x) >> 32
1117 dst.y = (src0.y \times src1.y) >> 32
1119 dst.z = (src0.z \times src1.z) >> 32
1121 dst.w = (src0.w \times src1.w) >> 32
1124 .. opcode:: UMUL_HI - Unsigned Integer Multiply High Bits
1126 The high 32bits of the multiplication of 2 unsigned integers are returned.
1130 dst.x = (src0.x \times src1.x) >> 32
1132 dst.y = (src0.y \times src1.y) >> 32
1134 dst.z = (src0.z \times src1.z) >> 32
1136 dst.w = (src0.w \times src1.w) >> 32
1139 .. opcode:: IDIV - Signed Integer Division
1141 TBD: behavior for division by zero.
1145 dst.x = \frac{src0.x}{src1.x}
1147 dst.y = \frac{src0.y}{src1.y}
1149 dst.z = \frac{src0.z}{src1.z}
1151 dst.w = \frac{src0.w}{src1.w}
1154 .. opcode:: UDIV - Unsigned Integer Division
1156 For division by zero, 0xffffffff is returned.
1160 dst.x = \frac{src0.x}{src1.x}
1162 dst.y = \frac{src0.y}{src1.y}
1164 dst.z = \frac{src0.z}{src1.z}
1166 dst.w = \frac{src0.w}{src1.w}
1169 .. opcode:: UMOD - Unsigned Integer Remainder
1171 If second arg is zero, 0xffffffff is returned.
1175 dst.x = src0.x \bmod src1.x
1177 dst.y = src0.y \bmod src1.y
1179 dst.z = src0.z \bmod src1.z
1181 dst.w = src0.w \bmod src1.w
1184 .. opcode:: NOT - Bitwise Not
1197 .. opcode:: AND - Bitwise And
1201 dst.x = src0.x \& src1.x
1203 dst.y = src0.y \& src1.y
1205 dst.z = src0.z \& src1.z
1207 dst.w = src0.w \& src1.w
1210 .. opcode:: OR - Bitwise Or
1214 dst.x = src0.x | src1.x
1216 dst.y = src0.y | src1.y
1218 dst.z = src0.z | src1.z
1220 dst.w = src0.w | src1.w
1223 .. opcode:: XOR - Bitwise Xor
1227 dst.x = src0.x \oplus src1.x
1229 dst.y = src0.y \oplus src1.y
1231 dst.z = src0.z \oplus src1.z
1233 dst.w = src0.w \oplus src1.w
1236 .. opcode:: IMAX - Maximum of Signed Integers
1240 dst.x = max(src0.x, src1.x)
1242 dst.y = max(src0.y, src1.y)
1244 dst.z = max(src0.z, src1.z)
1246 dst.w = max(src0.w, src1.w)
1249 .. opcode:: UMAX - Maximum of Unsigned Integers
1253 dst.x = max(src0.x, src1.x)
1255 dst.y = max(src0.y, src1.y)
1257 dst.z = max(src0.z, src1.z)
1259 dst.w = max(src0.w, src1.w)
1262 .. opcode:: IMIN - Minimum of Signed Integers
1266 dst.x = min(src0.x, src1.x)
1268 dst.y = min(src0.y, src1.y)
1270 dst.z = min(src0.z, src1.z)
1272 dst.w = min(src0.w, src1.w)
1275 .. opcode:: UMIN - Minimum of Unsigned Integers
1279 dst.x = min(src0.x, src1.x)
1281 dst.y = min(src0.y, src1.y)
1283 dst.z = min(src0.z, src1.z)
1285 dst.w = min(src0.w, src1.w)
1288 .. opcode:: SHL - Shift Left
1290 The shift count is masked with 0x1f before the shift is applied.
1294 dst.x = src0.x << (0x1f \& src1.x)
1296 dst.y = src0.y << (0x1f \& src1.y)
1298 dst.z = src0.z << (0x1f \& src1.z)
1300 dst.w = src0.w << (0x1f \& src1.w)
1303 .. opcode:: ISHR - Arithmetic Shift Right (of Signed Integer)
1305 The shift count is masked with 0x1f before the shift is applied.
1309 dst.x = src0.x >> (0x1f \& src1.x)
1311 dst.y = src0.y >> (0x1f \& src1.y)
1313 dst.z = src0.z >> (0x1f \& src1.z)
1315 dst.w = src0.w >> (0x1f \& src1.w)
1318 .. opcode:: USHR - Logical Shift Right
1320 The shift count is masked with 0x1f before the shift is applied.
1324 dst.x = src0.x >> (unsigned) (0x1f \& src1.x)
1326 dst.y = src0.y >> (unsigned) (0x1f \& src1.y)
1328 dst.z = src0.z >> (unsigned) (0x1f \& src1.z)
1330 dst.w = src0.w >> (unsigned) (0x1f \& src1.w)
1333 .. opcode:: UCMP - Integer Conditional Move
1337 dst.x = src0.x ? src1.x : src2.x
1339 dst.y = src0.y ? src1.y : src2.y
1341 dst.z = src0.z ? src1.z : src2.z
1343 dst.w = src0.w ? src1.w : src2.w
1347 .. opcode:: ISSG - Integer Set Sign
1351 dst.x = (src0.x < 0) ? -1 : (src0.x > 0) ? 1 : 0
1353 dst.y = (src0.y < 0) ? -1 : (src0.y > 0) ? 1 : 0
1355 dst.z = (src0.z < 0) ? -1 : (src0.z > 0) ? 1 : 0
1357 dst.w = (src0.w < 0) ? -1 : (src0.w > 0) ? 1 : 0
1361 .. opcode:: FSLT - Float Set On Less Than (ordered)
1363 Same comparison as SLT but returns integer instead of 1.0/0.0 float
1367 dst.x = (src0.x < src1.x) ? \sim 0 : 0
1369 dst.y = (src0.y < src1.y) ? \sim 0 : 0
1371 dst.z = (src0.z < src1.z) ? \sim 0 : 0
1373 dst.w = (src0.w < src1.w) ? \sim 0 : 0
1376 .. opcode:: ISLT - Signed Integer Set On Less Than
1380 dst.x = (src0.x < src1.x) ? \sim 0 : 0
1382 dst.y = (src0.y < src1.y) ? \sim 0 : 0
1384 dst.z = (src0.z < src1.z) ? \sim 0 : 0
1386 dst.w = (src0.w < src1.w) ? \sim 0 : 0
1389 .. opcode:: USLT - Unsigned Integer Set On Less Than
1393 dst.x = (src0.x < src1.x) ? \sim 0 : 0
1395 dst.y = (src0.y < src1.y) ? \sim 0 : 0
1397 dst.z = (src0.z < src1.z) ? \sim 0 : 0
1399 dst.w = (src0.w < src1.w) ? \sim 0 : 0
1402 .. opcode:: FSGE - Float Set On Greater Equal Than (ordered)
1404 Same comparison as SGE but returns integer instead of 1.0/0.0 float
1408 dst.x = (src0.x >= src1.x) ? \sim 0 : 0
1410 dst.y = (src0.y >= src1.y) ? \sim 0 : 0
1412 dst.z = (src0.z >= src1.z) ? \sim 0 : 0
1414 dst.w = (src0.w >= src1.w) ? \sim 0 : 0
1417 .. opcode:: ISGE - Signed Integer Set On Greater Equal Than
1421 dst.x = (src0.x >= src1.x) ? \sim 0 : 0
1423 dst.y = (src0.y >= src1.y) ? \sim 0 : 0
1425 dst.z = (src0.z >= src1.z) ? \sim 0 : 0
1427 dst.w = (src0.w >= src1.w) ? \sim 0 : 0
1430 .. opcode:: USGE - Unsigned Integer Set On Greater Equal Than
1434 dst.x = (src0.x >= src1.x) ? \sim 0 : 0
1436 dst.y = (src0.y >= src1.y) ? \sim 0 : 0
1438 dst.z = (src0.z >= src1.z) ? \sim 0 : 0
1440 dst.w = (src0.w >= src1.w) ? \sim 0 : 0
1443 .. opcode:: FSEQ - Float Set On Equal (ordered)
1445 Same comparison as SEQ but returns integer instead of 1.0/0.0 float
1449 dst.x = (src0.x == src1.x) ? \sim 0 : 0
1451 dst.y = (src0.y == src1.y) ? \sim 0 : 0
1453 dst.z = (src0.z == src1.z) ? \sim 0 : 0
1455 dst.w = (src0.w == src1.w) ? \sim 0 : 0
1458 .. opcode:: USEQ - Integer Set On Equal
1462 dst.x = (src0.x == src1.x) ? \sim 0 : 0
1464 dst.y = (src0.y == src1.y) ? \sim 0 : 0
1466 dst.z = (src0.z == src1.z) ? \sim 0 : 0
1468 dst.w = (src0.w == src1.w) ? \sim 0 : 0
1471 .. opcode:: FSNE - Float Set On Not Equal (unordered)
1473 Same comparison as SNE but returns integer instead of 1.0/0.0 float
1477 dst.x = (src0.x != src1.x) ? \sim 0 : 0
1479 dst.y = (src0.y != src1.y) ? \sim 0 : 0
1481 dst.z = (src0.z != src1.z) ? \sim 0 : 0
1483 dst.w = (src0.w != src1.w) ? \sim 0 : 0
1486 .. opcode:: USNE - Integer Set On Not Equal
1490 dst.x = (src0.x != src1.x) ? \sim 0 : 0
1492 dst.y = (src0.y != src1.y) ? \sim 0 : 0
1494 dst.z = (src0.z != src1.z) ? \sim 0 : 0
1496 dst.w = (src0.w != src1.w) ? \sim 0 : 0
1499 .. opcode:: INEG - Integer Negate
1514 .. opcode:: IABS - Integer Absolute Value
1528 These opcodes are used for bit-level manipulation of integers.
1530 .. opcode:: IBFE - Signed Bitfield Extract
1532 Like GLSL bitfieldExtract. Extracts a set of bits from the input, and
1533 sign-extends them if the high bit of the extracted window is set.
1537 def ibfe(value, offset, bits):
1538 if offset < 0 or bits < 0 or offset + bits > 32:
1540 if bits == 0: return 0
1541 # Note: >> sign-extends
1542 return (value << (32 - offset - bits)) >> (32 - bits)
1544 .. opcode:: UBFE - Unsigned Bitfield Extract
1546 Like GLSL bitfieldExtract. Extracts a set of bits from the input, without
1551 def ubfe(value, offset, bits):
1552 if offset < 0 or bits < 0 or offset + bits > 32:
1554 if bits == 0: return 0
1555 # Note: >> does not sign-extend
1556 return (value << (32 - offset - bits)) >> (32 - bits)
1558 .. opcode:: BFI - Bitfield Insert
1560 Like GLSL bitfieldInsert. Replaces a bit region of 'base' with the low bits
1565 def bfi(base, insert, offset, bits):
1566 if offset < 0 or bits < 0 or offset + bits > 32:
1568 # << defined such that mask == ~0 when bits == 32, offset == 0
1569 mask = ((1 << bits) - 1) << offset
1570 return ((insert << offset) & mask) | (base & ~mask)
1572 .. opcode:: BREV - Bitfield Reverse
1574 See SM5 instruction BFREV. Reverses the bits of the argument.
1576 .. opcode:: POPC - Population Count
1578 See SM5 instruction COUNTBITS. Counts the number of set bits in the argument.
1580 .. opcode:: LSB - Index of lowest set bit
1582 See SM5 instruction FIRSTBIT_LO. Computes the 0-based index of the first set
1583 bit of the argument. Returns -1 if none are set.
1585 .. opcode:: IMSB - Index of highest non-sign bit
1587 See SM5 instruction FIRSTBIT_SHI. Computes the 0-based index of the highest
1588 non-sign bit of the argument (i.e. highest 0 bit for negative numbers,
1589 highest 1 bit for positive numbers). Returns -1 if all bits are the same
1590 (i.e. for inputs 0 and -1).
1592 .. opcode:: UMSB - Index of highest set bit
1594 See SM5 instruction FIRSTBIT_HI. Computes the 0-based index of the highest
1595 set bit of the argument. Returns -1 if none are set.
1598 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1600 These opcodes are only supported in geometry shaders; they have no meaning
1601 in any other type of shader.
1603 .. opcode:: EMIT - Emit
1605 Generate a new vertex for the current primitive into the specified vertex
1606 stream using the values in the output registers.
1609 .. opcode:: ENDPRIM - End Primitive
1611 Complete the current primitive in the specified vertex stream (consisting of
1612 the emitted vertices), and start a new one.
1618 These opcodes are part of :term:`GLSL`'s opcode set. Support for these
1619 opcodes is determined by a special capability bit, ``GLSL``.
1620 Some require glsl version 1.30 (UIF/BREAKC/SWITCH/CASE/DEFAULT/ENDSWITCH).
1622 .. opcode:: CAL - Subroutine Call
1628 .. opcode:: RET - Subroutine Call Return
1633 .. opcode:: CONT - Continue
1635 Unconditionally moves the point of execution to the instruction after the
1636 last bgnloop. The instruction must appear within a bgnloop/endloop.
1640 Support for CONT is determined by a special capability bit,
1641 ``TGSI_CONT_SUPPORTED``. See :ref:`Screen` for more information.
1644 .. opcode:: BGNLOOP - Begin a Loop
1646 Start a loop. Must have a matching endloop.
1649 .. opcode:: BGNSUB - Begin Subroutine
1651 Starts definition of a subroutine. Must have a matching endsub.
1654 .. opcode:: ENDLOOP - End a Loop
1656 End a loop started with bgnloop.
1659 .. opcode:: ENDSUB - End Subroutine
1661 Ends definition of a subroutine.
1664 .. opcode:: NOP - No Operation
1669 .. opcode:: BRK - Break
1671 Unconditionally moves the point of execution to the instruction after the
1672 next endloop or endswitch. The instruction must appear within a loop/endloop
1673 or switch/endswitch.
1676 .. opcode:: BREAKC - Break Conditional
1678 Conditionally moves the point of execution to the instruction after the
1679 next endloop or endswitch. The instruction must appear within a loop/endloop
1680 or switch/endswitch.
1681 Condition evaluates to true if src0.x != 0 where src0.x is interpreted
1682 as an integer register.
1686 Considered for removal as it's quite inconsistent wrt other opcodes
1687 (could emulate with UIF/BRK/ENDIF).
1690 .. opcode:: IF - Float If
1692 Start an IF ... ELSE .. ENDIF block. Condition evaluates to true if
1696 where src0.x is interpreted as a floating point register.
1699 .. opcode:: UIF - Bitwise If
1701 Start an UIF ... ELSE .. ENDIF block. Condition evaluates to true if
1705 where src0.x is interpreted as an integer register.
1708 .. opcode:: ELSE - Else
1710 Starts an else block, after an IF or UIF statement.
1713 .. opcode:: ENDIF - End If
1715 Ends an IF or UIF block.
1718 .. opcode:: SWITCH - Switch
1720 Starts a C-style switch expression. The switch consists of one or multiple
1721 CASE statements, and at most one DEFAULT statement. Execution of a statement
1722 ends when a BRK is hit, but just like in C falling through to other cases
1723 without a break is allowed. Similarly, DEFAULT label is allowed anywhere not
1724 just as last statement, and fallthrough is allowed into/from it.
1725 CASE src arguments are evaluated at bit level against the SWITCH src argument.
1731 (some instructions here)
1734 (some instructions here)
1737 (some instructions here)
1742 .. opcode:: CASE - Switch case
1744 This represents a switch case label. The src arg must be an integer immediate.
1747 .. opcode:: DEFAULT - Switch default
1749 This represents the default case in the switch, which is taken if no other
1753 .. opcode:: ENDSWITCH - End of switch
1755 Ends a switch expression.
1761 The interpolation instructions allow an input to be interpolated in a
1762 different way than its declaration. This corresponds to the GLSL 4.00
1763 interpolateAt* functions. The first argument of each of these must come from
1764 ``TGSI_FILE_INPUT``.
1766 .. opcode:: INTERP_CENTROID - Interpolate at the centroid
1768 Interpolates the varying specified by src0 at the centroid
1770 .. opcode:: INTERP_SAMPLE - Interpolate at the specified sample
1772 Interpolates the varying specified by src0 at the sample id specified by
1773 src1.x (interpreted as an integer)
1775 .. opcode:: INTERP_OFFSET - Interpolate at the specified offset
1777 Interpolates the varying specified by src0 at the offset src1.xy from the
1778 pixel center (interpreted as floats)
1786 The double-precision opcodes reinterpret four-component vectors into
1787 two-component vectors with doubled precision in each component.
1789 .. opcode:: DABS - Absolute
1797 .. opcode:: DADD - Add
1801 dst.xy = src0.xy + src1.xy
1803 dst.zw = src0.zw + src1.zw
1805 .. opcode:: DSEQ - Set on Equal
1809 dst.x = src0.xy == src1.xy ? \sim 0 : 0
1811 dst.z = src0.zw == src1.zw ? \sim 0 : 0
1813 .. opcode:: DSNE - Set on Equal
1817 dst.x = src0.xy != src1.xy ? \sim 0 : 0
1819 dst.z = src0.zw != src1.zw ? \sim 0 : 0
1821 .. opcode:: DSLT - Set on Less than
1825 dst.x = src0.xy < src1.xy ? \sim 0 : 0
1827 dst.z = src0.zw < src1.zw ? \sim 0 : 0
1829 .. opcode:: DSGE - Set on Greater equal
1833 dst.x = src0.xy >= src1.xy ? \sim 0 : 0
1835 dst.z = src0.zw >= src1.zw ? \sim 0 : 0
1837 .. opcode:: DFRAC - Fraction
1841 dst.xy = src.xy - \lfloor src.xy\rfloor
1843 dst.zw = src.zw - \lfloor src.zw\rfloor
1845 .. opcode:: DTRUNC - Truncate
1849 dst.xy = trunc(src.xy)
1851 dst.zw = trunc(src.zw)
1853 .. opcode:: DCEIL - Ceiling
1857 dst.xy = \lceil src.xy\rceil
1859 dst.zw = \lceil src.zw\rceil
1861 .. opcode:: DFLR - Floor
1865 dst.xy = \lfloor src.xy\rfloor
1867 dst.zw = \lfloor src.zw\rfloor
1869 .. opcode:: DROUND - Fraction
1873 dst.xy = round(src.xy)
1875 dst.zw = round(src.zw)
1877 .. opcode:: DSSG - Set Sign
1881 dst.xy = (src.xy > 0) ? 1.0 : (src.xy < 0) ? -1.0 : 0.0
1883 dst.zw = (src.zw > 0) ? 1.0 : (src.zw < 0) ? -1.0 : 0.0
1885 .. opcode:: DFRACEXP - Convert Number to Fractional and Integral Components
1887 Like the ``frexp()`` routine in many math libraries, this opcode stores the
1888 exponent of its source to ``dst0``, and the significand to ``dst1``, such that
1889 :math:`dst1 \times 2^{dst0} = src` .
1893 dst0.xy = exp(src.xy)
1895 dst1.xy = frac(src.xy)
1897 dst0.zw = exp(src.zw)
1899 dst1.zw = frac(src.zw)
1901 .. opcode:: DLDEXP - Multiply Number by Integral Power of 2
1903 This opcode is the inverse of :opcode:`DFRACEXP`. The second
1904 source is an integer.
1908 dst.xy = src0.xy \times 2^{src1.x}
1910 dst.zw = src0.zw \times 2^{src1.y}
1912 .. opcode:: DMIN - Minimum
1916 dst.xy = min(src0.xy, src1.xy)
1918 dst.zw = min(src0.zw, src1.zw)
1920 .. opcode:: DMAX - Maximum
1924 dst.xy = max(src0.xy, src1.xy)
1926 dst.zw = max(src0.zw, src1.zw)
1928 .. opcode:: DMUL - Multiply
1932 dst.xy = src0.xy \times src1.xy
1934 dst.zw = src0.zw \times src1.zw
1937 .. opcode:: DMAD - Multiply And Add
1941 dst.xy = src0.xy \times src1.xy + src2.xy
1943 dst.zw = src0.zw \times src1.zw + src2.zw
1946 .. opcode:: DFMA - Fused Multiply-Add
1948 Perform a * b + c with no intermediate rounding step.
1952 dst.xy = src0.xy \times src1.xy + src2.xy
1954 dst.zw = src0.zw \times src1.zw + src2.zw
1957 .. opcode:: DDIV - Divide
1961 dst.xy = \frac{src0.xy}{src1.xy}
1963 dst.zw = \frac{src0.zw}{src1.zw}
1966 .. opcode:: DRCP - Reciprocal
1970 dst.xy = \frac{1}{src.xy}
1972 dst.zw = \frac{1}{src.zw}
1974 .. opcode:: DSQRT - Square Root
1978 dst.xy = \sqrt{src.xy}
1980 dst.zw = \sqrt{src.zw}
1982 .. opcode:: DRSQ - Reciprocal Square Root
1986 dst.xy = \frac{1}{\sqrt{src.xy}}
1988 dst.zw = \frac{1}{\sqrt{src.zw}}
1990 .. opcode:: F2D - Float to Double
1994 dst.xy = double(src0.x)
1996 dst.zw = double(src0.y)
1998 .. opcode:: D2F - Double to Float
2002 dst.x = float(src0.xy)
2004 dst.y = float(src0.zw)
2006 .. opcode:: I2D - Int to Double
2010 dst.xy = double(src0.x)
2012 dst.zw = double(src0.y)
2014 .. opcode:: D2I - Double to Int
2018 dst.x = int(src0.xy)
2020 dst.y = int(src0.zw)
2022 .. opcode:: U2D - Unsigned Int to Double
2026 dst.xy = double(src0.x)
2028 dst.zw = double(src0.y)
2030 .. opcode:: D2U - Double to Unsigned Int
2034 dst.x = unsigned(src0.xy)
2036 dst.y = unsigned(src0.zw)
2041 The 64-bit integer opcodes reinterpret four-component vectors into
2042 two-component vectors with 64-bits in each component.
2044 .. opcode:: I64ABS - 64-bit Integer Absolute Value
2052 .. opcode:: I64NEG - 64-bit Integer Negate
2062 .. opcode:: I64SSG - 64-bit Integer Set Sign
2066 dst.xy = (src0.xy < 0) ? -1 : (src0.xy > 0) ? 1 : 0
2068 dst.zw = (src0.zw < 0) ? -1 : (src0.zw > 0) ? 1 : 0
2070 .. opcode:: U64ADD - 64-bit Integer Add
2074 dst.xy = src0.xy + src1.xy
2076 dst.zw = src0.zw + src1.zw
2078 .. opcode:: U64MUL - 64-bit Integer Multiply
2082 dst.xy = src0.xy * src1.xy
2084 dst.zw = src0.zw * src1.zw
2086 .. opcode:: U64SEQ - 64-bit Integer Set on Equal
2090 dst.x = src0.xy == src1.xy ? \sim 0 : 0
2092 dst.z = src0.zw == src1.zw ? \sim 0 : 0
2094 .. opcode:: U64SNE - 64-bit Integer Set on Not Equal
2098 dst.x = src0.xy != src1.xy ? \sim 0 : 0
2100 dst.z = src0.zw != src1.zw ? \sim 0 : 0
2102 .. opcode:: U64SLT - 64-bit Unsigned Integer Set on Less Than
2106 dst.x = src0.xy < src1.xy ? \sim 0 : 0
2108 dst.z = src0.zw < src1.zw ? \sim 0 : 0
2110 .. opcode:: U64SGE - 64-bit Unsigned Integer Set on Greater Equal
2114 dst.x = src0.xy >= src1.xy ? \sim 0 : 0
2116 dst.z = src0.zw >= src1.zw ? \sim 0 : 0
2118 .. opcode:: I64SLT - 64-bit Signed Integer Set on Less Than
2122 dst.x = src0.xy < src1.xy ? \sim 0 : 0
2124 dst.z = src0.zw < src1.zw ? \sim 0 : 0
2126 .. opcode:: I64SGE - 64-bit Signed Integer Set on Greater Equal
2130 dst.x = src0.xy >= src1.xy ? \sim 0 : 0
2132 dst.z = src0.zw >= src1.zw ? \sim 0 : 0
2134 .. opcode:: I64MIN - Minimum of 64-bit Signed Integers
2138 dst.xy = min(src0.xy, src1.xy)
2140 dst.zw = min(src0.zw, src1.zw)
2142 .. opcode:: U64MIN - Minimum of 64-bit Unsigned Integers
2146 dst.xy = min(src0.xy, src1.xy)
2148 dst.zw = min(src0.zw, src1.zw)
2150 .. opcode:: I64MAX - Maximum of 64-bit Signed Integers
2154 dst.xy = max(src0.xy, src1.xy)
2156 dst.zw = max(src0.zw, src1.zw)
2158 .. opcode:: U64MAX - Maximum of 64-bit Unsigned Integers
2162 dst.xy = max(src0.xy, src1.xy)
2164 dst.zw = max(src0.zw, src1.zw)
2166 .. opcode:: U64SHL - Shift Left 64-bit Unsigned Integer
2168 The shift count is masked with 0x3f before the shift is applied.
2172 dst.xy = src0.xy << (0x3f \& src1.x)
2174 dst.zw = src0.zw << (0x3f \& src1.y)
2176 .. opcode:: I64SHR - Arithmetic Shift Right (of 64-bit Signed Integer)
2178 The shift count is masked with 0x3f before the shift is applied.
2182 dst.xy = src0.xy >> (0x3f \& src1.x)
2184 dst.zw = src0.zw >> (0x3f \& src1.y)
2186 .. opcode:: U64SHR - Logical Shift Right (of 64-bit Unsigned Integer)
2188 The shift count is masked with 0x3f before the shift is applied.
2192 dst.xy = src0.xy >> (unsigned) (0x3f \& src1.x)
2194 dst.zw = src0.zw >> (unsigned) (0x3f \& src1.y)
2196 .. opcode:: I64DIV - 64-bit Signed Integer Division
2200 dst.xy = \frac{src0.xy}{src1.xy}
2202 dst.zw = \frac{src0.zw}{src1.zw}
2204 .. opcode:: U64DIV - 64-bit Unsigned Integer Division
2208 dst.xy = \frac{src0.xy}{src1.xy}
2210 dst.zw = \frac{src0.zw}{src1.zw}
2212 .. opcode:: U64MOD - 64-bit Unsigned Integer Remainder
2216 dst.xy = src0.xy \bmod src1.xy
2218 dst.zw = src0.zw \bmod src1.zw
2220 .. opcode:: I64MOD - 64-bit Signed Integer Remainder
2224 dst.xy = src0.xy \bmod src1.xy
2226 dst.zw = src0.zw \bmod src1.zw
2228 .. opcode:: F2U64 - Float to 64-bit Unsigned Int
2232 dst.xy = (uint64_t) src0.x
2234 dst.zw = (uint64_t) src0.y
2236 .. opcode:: F2I64 - Float to 64-bit Int
2240 dst.xy = (int64_t) src0.x
2242 dst.zw = (int64_t) src0.y
2244 .. opcode:: U2I64 - Unsigned Integer to 64-bit Integer
2246 This is a zero extension.
2250 dst.xy = (uint64_t) src0.x
2252 dst.zw = (uint64_t) src0.y
2254 .. opcode:: I2I64 - Signed Integer to 64-bit Integer
2256 This is a sign extension.
2260 dst.xy = (int64_t) src0.x
2262 dst.zw = (int64_t) src0.y
2264 .. opcode:: D2U64 - Double to 64-bit Unsigned Int
2268 dst.xy = (uint64_t) src0.xy
2270 dst.zw = (uint64_t) src0.zw
2272 .. opcode:: D2I64 - Double to 64-bit Int
2276 dst.xy = (int64_t) src0.xy
2278 dst.zw = (int64_t) src0.zw
2280 .. opcode:: U642F - 64-bit unsigned integer to float
2284 dst.x = (float) src0.xy
2286 dst.y = (float) src0.zw
2288 .. opcode:: I642F - 64-bit Int to Float
2292 dst.x = (float) src0.xy
2294 dst.y = (float) src0.zw
2296 .. opcode:: U642D - 64-bit unsigned integer to double
2300 dst.xy = (double) src0.xy
2302 dst.zw = (double) src0.zw
2304 .. opcode:: I642D - 64-bit Int to double
2308 dst.xy = (double) src0.xy
2310 dst.zw = (double) src0.zw
2312 .. _samplingopcodes:
2314 Resource Sampling Opcodes
2315 ^^^^^^^^^^^^^^^^^^^^^^^^^
2317 Those opcodes follow very closely semantics of the respective Direct3D
2318 instructions. If in doubt double check Direct3D documentation.
2319 Note that the swizzle on SVIEW (src1) determines texel swizzling
2324 Using provided address, sample data from the specified texture using the
2325 filtering mode identified by the given sampler. The source data may come from
2326 any resource type other than buffers.
2328 Syntax: ``SAMPLE dst, address, sampler_view, sampler``
2330 Example: ``SAMPLE TEMP[0], TEMP[1], SVIEW[0], SAMP[0]``
2332 .. opcode:: SAMPLE_I
2334 Simplified alternative to the SAMPLE instruction. Using the provided
2335 integer address, SAMPLE_I fetches data from the specified sampler view
2336 without any filtering. The source data may come from any resource type
2339 Syntax: ``SAMPLE_I dst, address, sampler_view``
2341 Example: ``SAMPLE_I TEMP[0], TEMP[1], SVIEW[0]``
2343 The 'address' is specified as unsigned integers. If the 'address' is out of
2344 range [0...(# texels - 1)] the result of the fetch is always 0 in all
2345 components. As such the instruction doesn't honor address wrap modes, in
2346 cases where that behavior is desirable 'SAMPLE' instruction should be used.
2347 address.w always provides an unsigned integer mipmap level. If the value is
2348 out of the range then the instruction always returns 0 in all components.
2349 address.yz are ignored for buffers and 1d textures. address.z is ignored
2350 for 1d texture arrays and 2d textures.
2352 For 1D texture arrays address.y provides the array index (also as unsigned
2353 integer). If the value is out of the range of available array indices
2354 [0... (array size - 1)] then the opcode always returns 0 in all components.
2355 For 2D texture arrays address.z provides the array index, otherwise it
2356 exhibits the same behavior as in the case for 1D texture arrays. The exact
2357 semantics of the source address are presented in the table below:
2359 +---------------------------+----+-----+-----+---------+
2360 | resource type | X | Y | Z | W |
2361 +===========================+====+=====+=====+=========+
2362 | ``PIPE_BUFFER`` | x | | | ignored |
2363 +---------------------------+----+-----+-----+---------+
2364 | ``PIPE_TEXTURE_1D`` | x | | | mpl |
2365 +---------------------------+----+-----+-----+---------+
2366 | ``PIPE_TEXTURE_2D`` | x | y | | mpl |
2367 +---------------------------+----+-----+-----+---------+
2368 | ``PIPE_TEXTURE_3D`` | x | y | z | mpl |
2369 +---------------------------+----+-----+-----+---------+
2370 | ``PIPE_TEXTURE_RECT`` | x | y | | mpl |
2371 +---------------------------+----+-----+-----+---------+
2372 | ``PIPE_TEXTURE_CUBE`` | not allowed as source |
2373 +---------------------------+----+-----+-----+---------+
2374 | ``PIPE_TEXTURE_1D_ARRAY`` | x | idx | | mpl |
2375 +---------------------------+----+-----+-----+---------+
2376 | ``PIPE_TEXTURE_2D_ARRAY`` | x | y | idx | mpl |
2377 +---------------------------+----+-----+-----+---------+
2379 Where 'mpl' is a mipmap level and 'idx' is the array index.
2381 .. opcode:: SAMPLE_I_MS
2383 Just like SAMPLE_I but allows fetch data from multi-sampled surfaces.
2385 Syntax: ``SAMPLE_I_MS dst, address, sampler_view, sample``
2387 .. opcode:: SAMPLE_B
2389 Just like the SAMPLE instruction with the exception that an additional bias
2390 is applied to the level of detail computed as part of the instruction
2393 Syntax: ``SAMPLE_B dst, address, sampler_view, sampler, lod_bias``
2395 Example: ``SAMPLE_B TEMP[0], TEMP[1], SVIEW[0], SAMP[0], TEMP[2].x``
2397 .. opcode:: SAMPLE_C
2399 Similar to the SAMPLE instruction but it performs a comparison filter. The
2400 operands to SAMPLE_C are identical to SAMPLE, except that there is an
2401 additional float32 operand, reference value, which must be a register with
2402 single-component, or a scalar literal. SAMPLE_C makes the hardware use the
2403 current samplers compare_func (in pipe_sampler_state) to compare reference
2404 value against the red component value for the surce resource at each texel
2405 that the currently configured texture filter covers based on the provided
2408 Syntax: ``SAMPLE_C dst, address, sampler_view.r, sampler, ref_value``
2410 Example: ``SAMPLE_C TEMP[0], TEMP[1], SVIEW[0].r, SAMP[0], TEMP[2].x``
2412 .. opcode:: SAMPLE_C_LZ
2414 Same as SAMPLE_C, but LOD is 0 and derivatives are ignored. The LZ stands
2417 Syntax: ``SAMPLE_C_LZ dst, address, sampler_view.r, sampler, ref_value``
2419 Example: ``SAMPLE_C_LZ TEMP[0], TEMP[1], SVIEW[0].r, SAMP[0], TEMP[2].x``
2422 .. opcode:: SAMPLE_D
2424 SAMPLE_D is identical to the SAMPLE opcode except that the derivatives for
2425 the source address in the x direction and the y direction are provided by
2428 Syntax: ``SAMPLE_D dst, address, sampler_view, sampler, der_x, der_y``
2430 Example: ``SAMPLE_D TEMP[0], TEMP[1], SVIEW[0], SAMP[0], TEMP[2], TEMP[3]``
2432 .. opcode:: SAMPLE_L
2434 SAMPLE_L is identical to the SAMPLE opcode except that the LOD is provided
2435 directly as a scalar value, representing no anisotropy.
2437 Syntax: ``SAMPLE_L dst, address, sampler_view, sampler, explicit_lod``
2439 Example: ``SAMPLE_L TEMP[0], TEMP[1], SVIEW[0], SAMP[0], TEMP[2].x``
2443 Gathers the four texels to be used in a bi-linear filtering operation and
2444 packs them into a single register. Only works with 2D, 2D array, cubemaps,
2445 and cubemaps arrays. For 2D textures, only the addressing modes of the
2446 sampler and the top level of any mip pyramid are used. Set W to zero. It
2447 behaves like the SAMPLE instruction, but a filtered sample is not
2448 generated. The four samples that contribute to filtering are placed into
2449 xyzw in counter-clockwise order, starting with the (u,v) texture coordinate
2450 delta at the following locations (-, +), (+, +), (+, -), (-, -), where the
2451 magnitude of the deltas are half a texel.
2454 .. opcode:: SVIEWINFO
2456 Query the dimensions of a given sampler view. dst receives width, height,
2457 depth or array size and number of mipmap levels as int4. The dst can have a
2458 writemask which will specify what info is the caller interested in.
2460 Syntax: ``SVIEWINFO dst, src_mip_level, sampler_view``
2462 Example: ``SVIEWINFO TEMP[0], TEMP[1].x, SVIEW[0]``
2464 src_mip_level is an unsigned integer scalar. If it's out of range then
2465 returns 0 for width, height and depth/array size but the total number of
2466 mipmap is still returned correctly for the given sampler view. The returned
2467 width, height and depth values are for the mipmap level selected by the
2468 src_mip_level and are in the number of texels. For 1d texture array width
2469 is in dst.x, array size is in dst.y and dst.z is 0. The number of mipmaps is
2470 still in dst.w. In contrast to d3d10 resinfo, there's no way in the tgsi
2471 instruction encoding to specify the return type (float/rcpfloat/uint), hence
2472 always using uint. Also, unlike the SAMPLE instructions, the swizzle on src1
2473 resinfo allowing swizzling dst values is ignored (due to the interaction
2474 with rcpfloat modifier which requires some swizzle handling in the state
2477 .. opcode:: SAMPLE_POS
2479 Query the position of a sample in the given resource or render target
2480 when per-sample fragment shading is in effect.
2482 Syntax: ``SAMPLE_POS dst, source, sample_index``
2484 dst receives float4 (x, y, undef, undef) indicated where the sample is
2485 located. Sample locations are in the range [0, 1] where 0.5 is the center
2488 source is either a sampler view (to indicate a shader resource) or temp
2489 register (to indicate the render target). The source register may have
2490 an optional swizzle to apply to the returned result
2492 sample_index is an integer scalar indicating which sample position is to
2495 If per-sample shading is not in effect or the source resource or render
2496 target is not multisampled, the result is (0.5, 0.5, undef, undef).
2498 NOTE: no driver has implemented this opcode yet (and no state tracker
2499 emits it). This information is subject to change.
2501 .. opcode:: SAMPLE_INFO
2503 Query the number of samples in a multisampled resource or render target.
2505 Syntax: ``SAMPLE_INFO dst, source``
2507 dst receives int4 (n, 0, 0, 0) where n is the number of samples in a
2508 resource or the render target.
2510 source is either a sampler view (to indicate a shader resource) or temp
2511 register (to indicate the render target). The source register may have
2512 an optional swizzle to apply to the returned result
2514 If per-sample shading is not in effect or the source resource or render
2515 target is not multisampled, the result is (1, 0, 0, 0).
2517 NOTE: no driver has implemented this opcode yet (and no state tracker
2518 emits it). This information is subject to change.
2520 .. _resourceopcodes:
2522 Resource Access Opcodes
2523 ^^^^^^^^^^^^^^^^^^^^^^^
2525 For these opcodes, the resource can be a BUFFER, IMAGE, or MEMORY.
2527 .. opcode:: LOAD - Fetch data from a shader buffer or image
2529 Syntax: ``LOAD dst, resource, address``
2531 Example: ``LOAD TEMP[0], BUFFER[0], TEMP[1]``
2533 Using the provided integer address, LOAD fetches data
2534 from the specified buffer or texture without any
2537 The 'address' is specified as a vector of unsigned
2538 integers. If the 'address' is out of range the result
2541 Only the first mipmap level of a resource can be read
2542 from using this instruction.
2544 For 1D or 2D texture arrays, the array index is
2545 provided as an unsigned integer in address.y or
2546 address.z, respectively. address.yz are ignored for
2547 buffers and 1D textures. address.z is ignored for 1D
2548 texture arrays and 2D textures. address.w is always
2551 A swizzle suffix may be added to the resource argument
2552 this will cause the resource data to be swizzled accordingly.
2554 .. opcode:: STORE - Write data to a shader resource
2556 Syntax: ``STORE resource, address, src``
2558 Example: ``STORE BUFFER[0], TEMP[0], TEMP[1]``
2560 Using the provided integer address, STORE writes data
2561 to the specified buffer or texture.
2563 The 'address' is specified as a vector of unsigned
2564 integers. If the 'address' is out of range the result
2567 Only the first mipmap level of a resource can be
2568 written to using this instruction.
2570 For 1D or 2D texture arrays, the array index is
2571 provided as an unsigned integer in address.y or
2572 address.z, respectively. address.yz are ignored for
2573 buffers and 1D textures. address.z is ignored for 1D
2574 texture arrays and 2D textures. address.w is always
2577 .. opcode:: RESQ - Query information about a resource
2579 Syntax: ``RESQ dst, resource``
2581 Example: ``RESQ TEMP[0], BUFFER[0]``
2583 Returns information about the buffer or image resource. For buffer
2584 resources, the size (in bytes) is returned in the x component. For
2585 image resources, .xyz will contain the width/height/layers of the
2586 image, while .w will contain the number of samples for multi-sampled
2589 .. opcode:: FBFETCH - Load data from framebuffer
2591 Syntax: ``FBFETCH dst, output``
2593 Example: ``FBFETCH TEMP[0], OUT[0]``
2595 This is only valid on ``COLOR`` semantic outputs. Returns the color
2596 of the current position in the framebuffer from before this fragment
2597 shader invocation. May return the same value from multiple calls for
2598 a particular output within a single invocation. Note that result may
2599 be undefined if a fragment is drawn multiple times without a blend
2603 .. _threadsyncopcodes:
2605 Inter-thread synchronization opcodes
2606 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2608 These opcodes are intended for communication between threads running
2609 within the same compute grid. For now they're only valid in compute
2612 .. opcode:: BARRIER - Thread group barrier
2616 This opcode suspends the execution of the current thread until all
2617 the remaining threads in the working group reach the same point of
2618 the program. Results are unspecified if any of the remaining
2619 threads terminates or never reaches an executed BARRIER instruction.
2621 .. opcode:: MEMBAR - Memory barrier
2625 This opcode waits for the completion of all memory accesses based on
2626 the type passed in. The type is an immediate bitfield with the following
2629 Bit 0: Shader storage buffers
2630 Bit 1: Atomic buffers
2632 Bit 3: Shared memory
2635 These may be passed in in any combination. An implementation is free to not
2636 distinguish between these as it sees fit. However these map to all the
2637 possibilities made available by GLSL.
2644 These opcodes provide atomic variants of some common arithmetic and
2645 logical operations. In this context atomicity means that another
2646 concurrent memory access operation that affects the same memory
2647 location is guaranteed to be performed strictly before or after the
2648 entire execution of the atomic operation. The resource may be a BUFFER,
2649 IMAGE, or MEMORY. In the case of an image, the offset works the same as for
2650 ``LOAD`` and ``STORE``, specified above. These atomic operations may
2651 only be used with 32-bit integer image formats.
2653 .. opcode:: ATOMUADD - Atomic integer addition
2655 Syntax: ``ATOMUADD dst, resource, offset, src``
2657 Example: ``ATOMUADD TEMP[0], BUFFER[0], TEMP[1], TEMP[2]``
2659 The following operation is performed atomically:
2663 dst_x = resource[offset]
2665 resource[offset] = dst_x + src_x
2668 .. opcode:: ATOMXCHG - Atomic exchange
2670 Syntax: ``ATOMXCHG dst, resource, offset, src``
2672 Example: ``ATOMXCHG TEMP[0], BUFFER[0], TEMP[1], TEMP[2]``
2674 The following operation is performed atomically:
2678 dst_x = resource[offset]
2680 resource[offset] = src_x
2683 .. opcode:: ATOMCAS - Atomic compare-and-exchange
2685 Syntax: ``ATOMCAS dst, resource, offset, cmp, src``
2687 Example: ``ATOMCAS TEMP[0], BUFFER[0], TEMP[1], TEMP[2], TEMP[3]``
2689 The following operation is performed atomically:
2693 dst_x = resource[offset]
2695 resource[offset] = (dst_x == cmp_x ? src_x : dst_x)
2698 .. opcode:: ATOMAND - Atomic bitwise And
2700 Syntax: ``ATOMAND dst, resource, offset, src``
2702 Example: ``ATOMAND TEMP[0], BUFFER[0], TEMP[1], TEMP[2]``
2704 The following operation is performed atomically:
2708 dst_x = resource[offset]
2710 resource[offset] = dst_x \& src_x
2713 .. opcode:: ATOMOR - Atomic bitwise Or
2715 Syntax: ``ATOMOR dst, resource, offset, src``
2717 Example: ``ATOMOR TEMP[0], BUFFER[0], TEMP[1], TEMP[2]``
2719 The following operation is performed atomically:
2723 dst_x = resource[offset]
2725 resource[offset] = dst_x | src_x
2728 .. opcode:: ATOMXOR - Atomic bitwise Xor
2730 Syntax: ``ATOMXOR dst, resource, offset, src``
2732 Example: ``ATOMXOR TEMP[0], BUFFER[0], TEMP[1], TEMP[2]``
2734 The following operation is performed atomically:
2738 dst_x = resource[offset]
2740 resource[offset] = dst_x \oplus src_x
2743 .. opcode:: ATOMUMIN - Atomic unsigned minimum
2745 Syntax: ``ATOMUMIN dst, resource, offset, src``
2747 Example: ``ATOMUMIN TEMP[0], BUFFER[0], TEMP[1], TEMP[2]``
2749 The following operation is performed atomically:
2753 dst_x = resource[offset]
2755 resource[offset] = (dst_x < src_x ? dst_x : src_x)
2758 .. opcode:: ATOMUMAX - Atomic unsigned maximum
2760 Syntax: ``ATOMUMAX dst, resource, offset, src``
2762 Example: ``ATOMUMAX TEMP[0], BUFFER[0], TEMP[1], TEMP[2]``
2764 The following operation is performed atomically:
2768 dst_x = resource[offset]
2770 resource[offset] = (dst_x > src_x ? dst_x : src_x)
2773 .. opcode:: ATOMIMIN - Atomic signed minimum
2775 Syntax: ``ATOMIMIN dst, resource, offset, src``
2777 Example: ``ATOMIMIN TEMP[0], BUFFER[0], TEMP[1], TEMP[2]``
2779 The following operation is performed atomically:
2783 dst_x = resource[offset]
2785 resource[offset] = (dst_x < src_x ? dst_x : src_x)
2788 .. opcode:: ATOMIMAX - Atomic signed maximum
2790 Syntax: ``ATOMIMAX dst, resource, offset, src``
2792 Example: ``ATOMIMAX TEMP[0], BUFFER[0], TEMP[1], TEMP[2]``
2794 The following operation is performed atomically:
2798 dst_x = resource[offset]
2800 resource[offset] = (dst_x > src_x ? dst_x : src_x)
2803 .. _interlaneopcodes:
2808 These opcodes reduce the given value across the shader invocations
2809 running in the current SIMD group. Every thread in the subgroup will receive
2810 the same result. The BALLOT operations accept a single-channel argument that
2811 is treated as a boolean and produce a 64-bit value.
2813 .. opcode:: VOTE_ANY - Value is set in any of the active invocations
2815 Syntax: ``VOTE_ANY dst, value``
2817 Example: ``VOTE_ANY TEMP[0].x, TEMP[1].x``
2820 .. opcode:: VOTE_ALL - Value is set in all of the active invocations
2822 Syntax: ``VOTE_ALL dst, value``
2824 Example: ``VOTE_ALL TEMP[0].x, TEMP[1].x``
2827 .. opcode:: VOTE_EQ - Value is the same in all of the active invocations
2829 Syntax: ``VOTE_EQ dst, value``
2831 Example: ``VOTE_EQ TEMP[0].x, TEMP[1].x``
2834 .. opcode:: BALLOT - Lanemask of whether the value is set in each active
2837 Syntax: ``BALLOT dst, value``
2839 Example: ``BALLOT TEMP[0].xy, TEMP[1].x``
2841 When the argument is a constant true, this produces a bitmask of active
2842 invocations. In fragment shaders, this can include helper invocations
2843 (invocations whose outputs and writes to memory are discarded, but which
2844 are used to compute derivatives).
2847 .. opcode:: READ_FIRST - Broadcast the value from the first active
2848 invocation to all active lanes
2850 Syntax: ``READ_FIRST dst, value``
2852 Example: ``READ_FIRST TEMP[0], TEMP[1]``
2855 .. opcode:: READ_INVOC - Retrieve the value from the given invocation
2856 (need not be uniform)
2858 Syntax: ``READ_INVOC dst, value, invocation``
2860 Example: ``READ_INVOC TEMP[0].xy, TEMP[1].xy, TEMP[2].x``
2862 invocation.x controls the invocation number to read from for all channels.
2863 The invocation number must be the same across all active invocations in a
2864 sub-group; otherwise, the results are undefined.
2867 Explanation of symbols used
2868 ------------------------------
2875 :math:`|x|` Absolute value of `x`.
2877 :math:`\lceil x \rceil` Ceiling of `x`.
2879 clamp(x,y,z) Clamp x between y and z.
2880 (x < y) ? y : (x > z) ? z : x
2882 :math:`\lfloor x\rfloor` Floor of `x`.
2884 :math:`\log_2{x}` Logarithm of `x`, base 2.
2886 max(x,y) Maximum of x and y.
2889 min(x,y) Minimum of x and y.
2892 partialx(x) Derivative of x relative to fragment's X.
2894 partialy(x) Derivative of x relative to fragment's Y.
2896 pop() Pop from stack.
2898 :math:`x^y` `x` to the power `y`.
2900 push(x) Push x on stack.
2904 trunc(x) Truncate x, i.e. drop the fraction bits.
2911 discard Discard fragment.
2915 target Label of target instruction.
2926 Declares a register that is will be referenced as an operand in Instruction
2929 File field contains register file that is being declared and is one
2932 UsageMask field specifies which of the register components can be accessed
2933 and is one of TGSI_WRITEMASK.
2935 The Local flag specifies that a given value isn't intended for
2936 subroutine parameter passing and, as a result, the implementation
2937 isn't required to give any guarantees of it being preserved across
2938 subroutine boundaries. As it's merely a compiler hint, the
2939 implementation is free to ignore it.
2941 If Dimension flag is set to 1, a Declaration Dimension token follows.
2943 If Semantic flag is set to 1, a Declaration Semantic token follows.
2945 If Interpolate flag is set to 1, a Declaration Interpolate token follows.
2947 If file is TGSI_FILE_RESOURCE, a Declaration Resource token follows.
2949 If Array flag is set to 1, a Declaration Array token follows.
2952 ^^^^^^^^^^^^^^^^^^^^^^^^
2954 Declarations can optional have an ArrayID attribute which can be referred by
2955 indirect addressing operands. An ArrayID of zero is reserved and treated as
2956 if no ArrayID is specified.
2958 If an indirect addressing operand refers to a specific declaration by using
2959 an ArrayID only the registers in this declaration are guaranteed to be
2960 accessed, accessing any register outside this declaration results in undefined
2961 behavior. Note that for compatibility the effective index is zero-based and
2962 not relative to the specified declaration
2964 If no ArrayID is specified with an indirect addressing operand the whole
2965 register file might be accessed by this operand. This is strongly discouraged
2966 and will prevent packing of scalar/vec2 arrays and effective alias analysis.
2967 This is only legal for TEMP and CONST register files.
2969 Declaration Semantic
2970 ^^^^^^^^^^^^^^^^^^^^^^^^
2972 Vertex and fragment shader input and output registers may be labeled
2973 with semantic information consisting of a name and index.
2975 Follows Declaration token if Semantic bit is set.
2977 Since its purpose is to link a shader with other stages of the pipeline,
2978 it is valid to follow only those Declaration tokens that declare a register
2979 either in INPUT or OUTPUT file.
2981 SemanticName field contains the semantic name of the register being declared.
2982 There is no default value.
2984 SemanticIndex is an optional subscript that can be used to distinguish
2985 different register declarations with the same semantic name. The default value
2988 The meanings of the individual semantic names are explained in the following
2991 TGSI_SEMANTIC_POSITION
2992 """"""""""""""""""""""
2994 For vertex shaders, TGSI_SEMANTIC_POSITION indicates the vertex shader
2995 output register which contains the homogeneous vertex position in the clip
2996 space coordinate system. After clipping, the X, Y and Z components of the
2997 vertex will be divided by the W value to get normalized device coordinates.
2999 For fragment shaders, TGSI_SEMANTIC_POSITION is used to indicate that
3000 fragment shader input (or system value, depending on which one is
3001 supported by the driver) contains the fragment's window position. The X
3002 component starts at zero and always increases from left to right.
3003 The Y component starts at zero and always increases but Y=0 may either
3004 indicate the top of the window or the bottom depending on the fragment
3005 coordinate origin convention (see TGSI_PROPERTY_FS_COORD_ORIGIN).
3006 The Z coordinate ranges from 0 to 1 to represent depth from the front
3007 to the back of the Z buffer. The W component contains the interpolated
3008 reciprocal of the vertex position W component (corresponding to gl_Fragcoord,
3009 but unlike d3d10 which interpolates the same 1/w but then gives back
3010 the reciprocal of the interpolated value).
3012 Fragment shaders may also declare an output register with
3013 TGSI_SEMANTIC_POSITION. Only the Z component is writable. This allows
3014 the fragment shader to change the fragment's Z position.
3021 For vertex shader outputs or fragment shader inputs/outputs, this
3022 label indicates that the register contains an R,G,B,A color.
3024 Several shader inputs/outputs may contain colors so the semantic index
3025 is used to distinguish them. For example, color[0] may be the diffuse
3026 color while color[1] may be the specular color.
3028 This label is needed so that the flat/smooth shading can be applied
3029 to the right interpolants during rasterization.
3033 TGSI_SEMANTIC_BCOLOR
3034 """"""""""""""""""""
3036 Back-facing colors are only used for back-facing polygons, and are only valid
3037 in vertex shader outputs. After rasterization, all polygons are front-facing
3038 and COLOR and BCOLOR end up occupying the same slots in the fragment shader,
3039 so all BCOLORs effectively become regular COLORs in the fragment shader.
3045 Vertex shader inputs and outputs and fragment shader inputs may be
3046 labeled with TGSI_SEMANTIC_FOG to indicate that the register contains
3047 a fog coordinate. Typically, the fragment shader will use the fog coordinate
3048 to compute a fog blend factor which is used to blend the normal fragment color
3049 with a constant fog color. But fog coord really is just an ordinary vec4
3050 register like regular semantics.
3056 Vertex shader input and output registers may be labeled with
3057 TGIS_SEMANTIC_PSIZE to indicate that the register contains a point size
3058 in the form (S, 0, 0, 1). The point size controls the width or diameter
3059 of points for rasterization. This label cannot be used in fragment
3062 When using this semantic, be sure to set the appropriate state in the
3063 :ref:`rasterizer` first.
3066 TGSI_SEMANTIC_TEXCOORD
3067 """"""""""""""""""""""
3069 Only available if PIPE_CAP_TGSI_TEXCOORD is exposed !
3071 Vertex shader outputs and fragment shader inputs may be labeled with
3072 this semantic to make them replaceable by sprite coordinates via the
3073 sprite_coord_enable state in the :ref:`rasterizer`.
3074 The semantic index permitted with this semantic is limited to <= 7.
3076 If the driver does not support TEXCOORD, sprite coordinate replacement
3077 applies to inputs with the GENERIC semantic instead.
3079 The intended use case for this semantic is gl_TexCoord.
3082 TGSI_SEMANTIC_PCOORD
3083 """"""""""""""""""""
3085 Only available if PIPE_CAP_TGSI_TEXCOORD is exposed !
3087 Fragment shader inputs may be labeled with TGSI_SEMANTIC_PCOORD to indicate
3088 that the register contains sprite coordinates in the form (x, y, 0, 1), if
3089 the current primitive is a point and point sprites are enabled. Otherwise,
3090 the contents of the register are undefined.
3092 The intended use case for this semantic is gl_PointCoord.
3095 TGSI_SEMANTIC_GENERIC
3096 """""""""""""""""""""
3098 All vertex/fragment shader inputs/outputs not labeled with any other
3099 semantic label can be considered to be generic attributes. Typical
3100 uses of generic inputs/outputs are texcoords and user-defined values.
3103 TGSI_SEMANTIC_NORMAL
3104 """"""""""""""""""""
3106 Indicates that a vertex shader input is a normal vector. This is
3107 typically only used for legacy graphics APIs.
3113 This label applies to fragment shader inputs (or system values,
3114 depending on which one is supported by the driver) and indicates that
3115 the register contains front/back-face information.
3117 If it is an input, it will be a floating-point vector in the form (F, 0, 0, 1),
3118 where F will be positive when the fragment belongs to a front-facing polygon,
3119 and negative when the fragment belongs to a back-facing polygon.
3121 If it is a system value, it will be an integer vector in the form (F, 0, 0, 1),
3122 where F is 0xffffffff when the fragment belongs to a front-facing polygon and
3123 0 when the fragment belongs to a back-facing polygon.
3126 TGSI_SEMANTIC_EDGEFLAG
3127 """"""""""""""""""""""
3129 For vertex shaders, this sematic label indicates that an input or
3130 output is a boolean edge flag. The register layout is [F, x, x, x]
3131 where F is 0.0 or 1.0 and x = don't care. Normally, the vertex shader
3132 simply copies the edge flag input to the edgeflag output.
3134 Edge flags are used to control which lines or points are actually
3135 drawn when the polygon mode converts triangles/quads/polygons into
3139 TGSI_SEMANTIC_STENCIL
3140 """""""""""""""""""""
3142 For fragment shaders, this semantic label indicates that an output
3143 is a writable stencil reference value. Only the Y component is writable.
3144 This allows the fragment shader to change the fragments stencilref value.
3147 TGSI_SEMANTIC_VIEWPORT_INDEX
3148 """"""""""""""""""""""""""""
3150 For geometry shaders, this semantic label indicates that an output
3151 contains the index of the viewport (and scissor) to use.
3152 This is an integer value, and only the X component is used.
3154 If PIPE_CAP_TGSI_VS_LAYER_VIEWPORT or PIPE_CAP_TGSI_TES_LAYER_VIEWPORT is
3155 supported, then this semantic label can also be used in vertex or
3156 tessellation evaluation shaders, respectively. Only the value written in the
3157 last vertex processing stage is used.
3163 For geometry shaders, this semantic label indicates that an output
3164 contains the layer value to use for the color and depth/stencil surfaces.
3165 This is an integer value, and only the X component is used.
3166 (Also known as rendertarget array index.)
3168 If PIPE_CAP_TGSI_VS_LAYER_VIEWPORT or PIPE_CAP_TGSI_TES_LAYER_VIEWPORT is
3169 supported, then this semantic label can also be used in vertex or
3170 tessellation evaluation shaders, respectively. Only the value written in the
3171 last vertex processing stage is used.
3174 TGSI_SEMANTIC_CULLDIST
3175 """"""""""""""""""""""
3177 Used as distance to plane for performing application-defined culling
3178 of individual primitives against a plane. When components of vertex
3179 elements are given this label, these values are assumed to be a
3180 float32 signed distance to a plane. Primitives will be completely
3181 discarded if the plane distance for all of the vertices in the
3182 primitive are < 0. If a vertex has a cull distance of NaN, that
3183 vertex counts as "out" (as if its < 0);
3184 The limits on both clip and cull distances are bound
3185 by the PIPE_MAX_CLIP_OR_CULL_DISTANCE_COUNT define which defines
3186 the maximum number of components that can be used to hold the
3187 distances and by the PIPE_MAX_CLIP_OR_CULL_DISTANCE_ELEMENT_COUNT
3188 which specifies the maximum number of registers which can be
3189 annotated with those semantics.
3192 TGSI_SEMANTIC_CLIPDIST
3193 """"""""""""""""""""""
3195 Note this covers clipping and culling distances.
3197 When components of vertex elements are identified this way, these
3198 values are each assumed to be a float32 signed distance to a plane.
3201 Primitive setup only invokes rasterization on pixels for which
3202 the interpolated plane distances are >= 0.
3205 Primitives will be completely discarded if the plane distance
3206 for all of the vertices in the primitive are < 0.
3207 If a vertex has a cull distance of NaN, that vertex counts as "out"
3210 Multiple clip/cull planes can be implemented simultaneously, by
3211 annotating multiple components of one or more vertex elements with
3212 the above specified semantic.
3213 The limits on both clip and cull distances are bound
3214 by the PIPE_MAX_CLIP_OR_CULL_DISTANCE_COUNT define which defines
3215 the maximum number of components that can be used to hold the
3216 distances and by the PIPE_MAX_CLIP_OR_CULL_DISTANCE_ELEMENT_COUNT
3217 which specifies the maximum number of registers which can be
3218 annotated with those semantics.
3219 The properties NUM_CLIPDIST_ENABLED and NUM_CULLDIST_ENABLED
3220 are used to divide up the 2 x vec4 space between clipping and culling.
3222 TGSI_SEMANTIC_SAMPLEID
3223 """"""""""""""""""""""
3225 For fragment shaders, this semantic label indicates that a system value
3226 contains the current sample id (i.e. gl_SampleID) as an unsigned int.
3227 Only the X component is used. If per-sample shading is not enabled,
3228 the result is (0, undef, undef, undef).
3230 Note that if the fragment shader uses this system value, the fragment
3231 shader is automatically executed at per sample frequency.
3233 TGSI_SEMANTIC_SAMPLEPOS
3234 """""""""""""""""""""""
3236 For fragment shaders, this semantic label indicates that a system
3237 value contains the current sample's position as float4(x, y, undef, undef)
3238 in the render target (i.e. gl_SamplePosition) when per-fragment shading
3239 is in effect. Position values are in the range [0, 1] where 0.5 is
3240 the center of the fragment.
3242 Note that if the fragment shader uses this system value, the fragment
3243 shader is automatically executed at per sample frequency.
3245 TGSI_SEMANTIC_SAMPLEMASK
3246 """"""""""""""""""""""""
3248 For fragment shaders, this semantic label can be applied to either a
3249 shader system value input or output.
3251 For a system value, the sample mask indicates the set of samples covered by
3252 the current primitive. If MSAA is not enabled, the value is (1, 0, 0, 0).
3254 For an output, the sample mask is used to disable further sample processing.
3256 For both, the register type is uint[4] but only the X component is used
3257 (i.e. gl_SampleMask[0]). Each bit corresponds to one sample position (up
3258 to 32x MSAA is supported).
3260 TGSI_SEMANTIC_INVOCATIONID
3261 """"""""""""""""""""""""""
3263 For geometry shaders, this semantic label indicates that a system value
3264 contains the current invocation id (i.e. gl_InvocationID).
3265 This is an integer value, and only the X component is used.
3267 TGSI_SEMANTIC_INSTANCEID
3268 """"""""""""""""""""""""
3270 For vertex shaders, this semantic label indicates that a system value contains
3271 the current instance id (i.e. gl_InstanceID). It does not include the base
3272 instance. This is an integer value, and only the X component is used.
3274 TGSI_SEMANTIC_VERTEXID
3275 """"""""""""""""""""""
3277 For vertex shaders, this semantic label indicates that a system value contains
3278 the current vertex id (i.e. gl_VertexID). It does (unlike in d3d10) include the
3279 base vertex. This is an integer value, and only the X component is used.
3281 TGSI_SEMANTIC_VERTEXID_NOBASE
3282 """""""""""""""""""""""""""""""
3284 For vertex shaders, this semantic label indicates that a system value contains
3285 the current vertex id without including the base vertex (this corresponds to
3286 d3d10 vertex id, so TGSI_SEMANTIC_VERTEXID_NOBASE + TGSI_SEMANTIC_BASEVERTEX
3287 == TGSI_SEMANTIC_VERTEXID). This is an integer value, and only the X component
3290 TGSI_SEMANTIC_BASEVERTEX
3291 """"""""""""""""""""""""
3293 For vertex shaders, this semantic label indicates that a system value contains
3294 the base vertex (i.e. gl_BaseVertex). Note that for non-indexed draw calls,
3295 this contains the first (or start) value instead.
3296 This is an integer value, and only the X component is used.
3298 TGSI_SEMANTIC_PRIMID
3299 """"""""""""""""""""
3301 For geometry and fragment shaders, this semantic label indicates the value
3302 contains the primitive id (i.e. gl_PrimitiveID). This is an integer value,
3303 and only the X component is used.
3304 FIXME: This right now can be either a ordinary input or a system value...
3310 For tessellation evaluation/control shaders, this semantic label indicates a
3311 generic per-patch attribute. Such semantics will not implicitly be per-vertex
3314 TGSI_SEMANTIC_TESSCOORD
3315 """""""""""""""""""""""
3317 For tessellation evaluation shaders, this semantic label indicates the
3318 coordinates of the vertex being processed. This is available in XYZ; W is
3321 TGSI_SEMANTIC_TESSOUTER
3322 """""""""""""""""""""""
3324 For tessellation evaluation/control shaders, this semantic label indicates the
3325 outer tessellation levels of the patch. Isoline tessellation will only have XY
3326 defined, triangle will have XYZ and quads will have XYZW defined. This
3327 corresponds to gl_TessLevelOuter.
3329 TGSI_SEMANTIC_TESSINNER
3330 """""""""""""""""""""""
3332 For tessellation evaluation/control shaders, this semantic label indicates the
3333 inner tessellation levels of the patch. The X value is only defined for
3334 triangle tessellation, while quads will have XY defined. This is entirely
3335 undefined for isoline tessellation.
3337 TGSI_SEMANTIC_VERTICESIN
3338 """"""""""""""""""""""""
3340 For tessellation evaluation/control shaders, this semantic label indicates the
3341 number of vertices provided in the input patch. Only the X value is defined.
3343 TGSI_SEMANTIC_HELPER_INVOCATION
3344 """""""""""""""""""""""""""""""
3346 For fragment shaders, this semantic indicates whether the current
3347 invocation is covered or not. Helper invocations are created in order
3348 to properly compute derivatives, however it may be desirable to skip
3349 some of the logic in those cases. See ``gl_HelperInvocation`` documentation.
3351 TGSI_SEMANTIC_BASEINSTANCE
3352 """"""""""""""""""""""""""
3354 For vertex shaders, the base instance argument supplied for this
3355 draw. This is an integer value, and only the X component is used.
3357 TGSI_SEMANTIC_DRAWID
3358 """"""""""""""""""""
3360 For vertex shaders, the zero-based index of the current draw in a
3361 ``glMultiDraw*`` invocation. This is an integer value, and only the X
3365 TGSI_SEMANTIC_WORK_DIM
3366 """"""""""""""""""""""
3368 For compute shaders started via opencl this retrieves the work_dim
3369 parameter to the clEnqueueNDRangeKernel call with which the shader
3373 TGSI_SEMANTIC_GRID_SIZE
3374 """""""""""""""""""""""
3376 For compute shaders, this semantic indicates the maximum (x, y, z) dimensions
3377 of a grid of thread blocks.
3380 TGSI_SEMANTIC_BLOCK_ID
3381 """"""""""""""""""""""
3383 For compute shaders, this semantic indicates the (x, y, z) coordinates of the
3384 current block inside of the grid.
3387 TGSI_SEMANTIC_BLOCK_SIZE
3388 """"""""""""""""""""""""
3390 For compute shaders, this semantic indicates the maximum (x, y, z) dimensions
3391 of a block in threads.
3394 TGSI_SEMANTIC_THREAD_ID
3395 """""""""""""""""""""""
3397 For compute shaders, this semantic indicates the (x, y, z) coordinates of the
3398 current thread inside of the block.
3401 TGSI_SEMANTIC_SUBGROUP_SIZE
3402 """""""""""""""""""""""""""
3404 This semantic indicates the subgroup size for the current invocation. This is
3405 an integer of at most 64, as it indicates the width of lanemasks. It does not
3406 depend on the number of invocations that are active.
3409 TGSI_SEMANTIC_SUBGROUP_INVOCATION
3410 """""""""""""""""""""""""""""""""
3412 The index of the current invocation within its subgroup.
3415 TGSI_SEMANTIC_SUBGROUP_EQ_MASK
3416 """"""""""""""""""""""""""""""
3418 A bit mask of ``bit index == TGSI_SEMANTIC_SUBGROUP_INVOCATION``, i.e.
3419 ``1 << subgroup_invocation`` in arbitrary precision arithmetic.
3422 TGSI_SEMANTIC_SUBGROUP_GE_MASK
3423 """"""""""""""""""""""""""""""
3425 A bit mask of ``bit index >= TGSI_SEMANTIC_SUBGROUP_INVOCATION``, i.e.
3426 ``((1 << (subgroup_size - subgroup_invocation)) - 1) << subgroup_invocation``
3427 in arbitrary precision arithmetic.
3430 TGSI_SEMANTIC_SUBGROUP_GT_MASK
3431 """"""""""""""""""""""""""""""
3433 A bit mask of ``bit index > TGSI_SEMANTIC_SUBGROUP_INVOCATION``, i.e.
3434 ``((1 << (subgroup_size - subgroup_invocation - 1)) - 1) << (subgroup_invocation + 1)``
3435 in arbitrary precision arithmetic.
3438 TGSI_SEMANTIC_SUBGROUP_LE_MASK
3439 """"""""""""""""""""""""""""""
3441 A bit mask of ``bit index <= TGSI_SEMANTIC_SUBGROUP_INVOCATION``, i.e.
3442 ``(1 << (subgroup_invocation + 1)) - 1`` in arbitrary precision arithmetic.
3445 TGSI_SEMANTIC_SUBGROUP_LT_MASK
3446 """"""""""""""""""""""""""""""
3448 A bit mask of ``bit index > TGSI_SEMANTIC_SUBGROUP_INVOCATION``, i.e.
3449 ``(1 << subgroup_invocation) - 1`` in arbitrary precision arithmetic.
3452 Declaration Interpolate
3453 ^^^^^^^^^^^^^^^^^^^^^^^
3455 This token is only valid for fragment shader INPUT declarations.
3457 The Interpolate field specifes the way input is being interpolated by
3458 the rasteriser and is one of TGSI_INTERPOLATE_*.
3460 The Location field specifies the location inside the pixel that the
3461 interpolation should be done at, one of ``TGSI_INTERPOLATE_LOC_*``. Note that
3462 when per-sample shading is enabled, the implementation may choose to
3463 interpolate at the sample irrespective of the Location field.
3465 The CylindricalWrap bitfield specifies which register components
3466 should be subject to cylindrical wrapping when interpolating by the
3467 rasteriser. If TGSI_CYLINDRICAL_WRAP_X is set to 1, the X component
3468 should be interpolated according to cylindrical wrapping rules.
3471 Declaration Sampler View
3472 ^^^^^^^^^^^^^^^^^^^^^^^^
3474 Follows Declaration token if file is TGSI_FILE_SAMPLER_VIEW.
3476 DCL SVIEW[#], resource, type(s)
3478 Declares a shader input sampler view and assigns it to a SVIEW[#]
3481 resource can be one of BUFFER, 1D, 2D, 3D, 1DArray and 2DArray.
3483 type must be 1 or 4 entries (if specifying on a per-component
3484 level) out of UNORM, SNORM, SINT, UINT and FLOAT.
3486 For TEX\* style texture sample opcodes (as opposed to SAMPLE\* opcodes
3487 which take an explicit SVIEW[#] source register), there may be optionally
3488 SVIEW[#] declarations. In this case, the SVIEW index is implied by the
3489 SAMP index, and there must be a corresponding SVIEW[#] declaration for
3490 each SAMP[#] declaration. Drivers are free to ignore this if they wish.
3491 But note in particular that some drivers need to know the sampler type
3492 (float/int/unsigned) in order to generate the correct code, so cases
3493 where integer textures are sampled, SVIEW[#] declarations should be
3496 NOTE: It is NOT legal to mix SAMPLE\* style opcodes and TEX\* opcodes
3499 Declaration Resource
3500 ^^^^^^^^^^^^^^^^^^^^
3502 Follows Declaration token if file is TGSI_FILE_RESOURCE.
3504 DCL RES[#], resource [, WR] [, RAW]
3506 Declares a shader input resource and assigns it to a RES[#]
3509 resource can be one of BUFFER, 1D, 2D, 3D, CUBE, 1DArray and
3512 If the RAW keyword is not specified, the texture data will be
3513 subject to conversion, swizzling and scaling as required to yield
3514 the specified data type from the physical data format of the bound
3517 If the RAW keyword is specified, no channel conversion will be
3518 performed: the values read for each of the channels (X,Y,Z,W) will
3519 correspond to consecutive words in the same order and format
3520 they're found in memory. No element-to-address conversion will be
3521 performed either: the value of the provided X coordinate will be
3522 interpreted in byte units instead of texel units. The result of
3523 accessing a misaligned address is undefined.
3525 Usage of the STORE opcode is only allowed if the WR (writable) flag
3530 ^^^^^^^^^^^^^^^^^^^^^^^^
3532 Properties are general directives that apply to the whole TGSI program.
3537 Specifies the fragment shader TGSI_SEMANTIC_POSITION coordinate origin.
3538 The default value is UPPER_LEFT.
3540 If UPPER_LEFT, the position will be (0,0) at the upper left corner and
3541 increase downward and rightward.
3542 If LOWER_LEFT, the position will be (0,0) at the lower left corner and
3543 increase upward and rightward.
3545 OpenGL defaults to LOWER_LEFT, and is configurable with the
3546 GL_ARB_fragment_coord_conventions extension.
3548 DirectX 9/10 use UPPER_LEFT.
3550 FS_COORD_PIXEL_CENTER
3551 """""""""""""""""""""
3553 Specifies the fragment shader TGSI_SEMANTIC_POSITION pixel center convention.
3554 The default value is HALF_INTEGER.
3556 If HALF_INTEGER, the fractionary part of the position will be 0.5
3557 If INTEGER, the fractionary part of the position will be 0.0
3559 Note that this does not affect the set of fragments generated by
3560 rasterization, which is instead controlled by half_pixel_center in the
3563 OpenGL defaults to HALF_INTEGER, and is configurable with the
3564 GL_ARB_fragment_coord_conventions extension.
3566 DirectX 9 uses INTEGER.
3567 DirectX 10 uses HALF_INTEGER.
3569 FS_COLOR0_WRITES_ALL_CBUFS
3570 """"""""""""""""""""""""""
3571 Specifies that writes to the fragment shader color 0 are replicated to all
3572 bound cbufs. This facilitates OpenGL's fragColor output vs fragData[0] where
3573 fragData is directed to a single color buffer, but fragColor is broadcast.
3576 """"""""""""""""""""""""""
3577 If this property is set on the program bound to the shader stage before the
3578 fragment shader, user clip planes should have no effect (be disabled) even if
3579 that shader does not write to any clip distance outputs and the rasterizer's
3580 clip_plane_enable is non-zero.
3581 This property is only supported by drivers that also support shader clip
3583 This is useful for APIs that don't have UCPs and where clip distances written
3584 by a shader cannot be disabled.
3589 Specifies the number of times a geometry shader should be executed for each
3590 input primitive. Each invocation will have a different
3591 TGSI_SEMANTIC_INVOCATIONID system value set. If not specified, assumed to
3594 VS_WINDOW_SPACE_POSITION
3595 """"""""""""""""""""""""""
3596 If this property is set on the vertex shader, the TGSI_SEMANTIC_POSITION output
3597 is assumed to contain window space coordinates.
3598 Division of X,Y,Z by W and the viewport transformation are disabled, and 1/W is
3599 directly taken from the 4-th component of the shader output.
3600 Naturally, clipping is not performed on window coordinates either.
3601 The effect of this property is undefined if a geometry or tessellation shader
3607 The number of vertices written by the tessellation control shader. This
3608 effectively defines the patch input size of the tessellation evaluation shader
3614 This sets the tessellation primitive mode, one of ``PIPE_PRIM_TRIANGLES``,
3615 ``PIPE_PRIM_QUADS``, or ``PIPE_PRIM_LINES``. (Unlike in GL, there is no
3616 separate isolines settings, the regular lines is assumed to mean isolines.)
3621 This sets the spacing mode of the tessellation generator, one of
3622 ``PIPE_TESS_SPACING_*``.
3627 This sets the vertex order to be clockwise if the value is 1, or
3628 counter-clockwise if set to 0.
3633 If set to a non-zero value, this turns on point mode for the tessellator,
3634 which means that points will be generated instead of primitives.
3636 NUM_CLIPDIST_ENABLED
3637 """"""""""""""""""""
3639 How many clip distance scalar outputs are enabled.
3641 NUM_CULLDIST_ENABLED
3642 """"""""""""""""""""
3644 How many cull distance scalar outputs are enabled.
3646 FS_EARLY_DEPTH_STENCIL
3647 """"""""""""""""""""""
3649 Whether depth test, stencil test, and occlusion query should run before
3650 the fragment shader (regardless of fragment shader side effects). Corresponds
3651 to GLSL early_fragment_tests.
3656 Which shader stage will MOST LIKELY follow after this shader when the shader
3657 is bound. This is only a hint to the driver and doesn't have to be precise.
3658 Only set for VS and TES.
3660 CS_FIXED_BLOCK_WIDTH / HEIGHT / DEPTH
3661 """""""""""""""""""""""""""""""""""""
3663 Threads per block in each dimension, if known at compile time. If the block size
3664 is known all three should be at least 1. If it is unknown they should all be set
3670 The MUL TGSI operation (FP32 multiplication) will return 0 if either
3671 of the operands are equal to 0. That means that 0 * Inf = 0. This
3672 should be set the same way for an entire pipeline. Note that this
3673 applies not only to the literal MUL TGSI opcode, but all FP32
3674 multiplications implied by other operations, such as MAD, FMA, DP2,
3675 DP3, DP4, DPH, DST, LOG, LRP, XPD, and possibly others. If there is a
3676 mismatch between shaders, then it is unspecified whether this behavior
3679 FS_POST_DEPTH_COVERAGE
3680 """"""""""""""""""""""
3682 When enabled, the input for TGSI_SEMANTIC_SAMPLEMASK will exclude samples
3683 that have failed the depth/stencil tests. This is only valid when
3684 FS_EARLY_DEPTH_STENCIL is also specified.
3687 Texture Sampling and Texture Formats
3688 ------------------------------------
3690 This table shows how texture image components are returned as (x,y,z,w) tuples
3691 by TGSI texture instructions, such as :opcode:`TEX`, :opcode:`TXD`, and
3692 :opcode:`TXP`. For reference, OpenGL and Direct3D conventions are shown as
3695 +--------------------+--------------+--------------------+--------------+
3696 | Texture Components | Gallium | OpenGL | Direct3D 9 |
3697 +====================+==============+====================+==============+
3698 | R | (r, 0, 0, 1) | (r, 0, 0, 1) | (r, 1, 1, 1) |
3699 +--------------------+--------------+--------------------+--------------+
3700 | RG | (r, g, 0, 1) | (r, g, 0, 1) | (r, g, 1, 1) |
3701 +--------------------+--------------+--------------------+--------------+
3702 | RGB | (r, g, b, 1) | (r, g, b, 1) | (r, g, b, 1) |
3703 +--------------------+--------------+--------------------+--------------+
3704 | RGBA | (r, g, b, a) | (r, g, b, a) | (r, g, b, a) |
3705 +--------------------+--------------+--------------------+--------------+
3706 | A | (0, 0, 0, a) | (0, 0, 0, a) | (0, 0, 0, a) |
3707 +--------------------+--------------+--------------------+--------------+
3708 | L | (l, l, l, 1) | (l, l, l, 1) | (l, l, l, 1) |
3709 +--------------------+--------------+--------------------+--------------+
3710 | LA | (l, l, l, a) | (l, l, l, a) | (l, l, l, a) |
3711 +--------------------+--------------+--------------------+--------------+
3712 | I | (i, i, i, i) | (i, i, i, i) | N/A |
3713 +--------------------+--------------+--------------------+--------------+
3714 | UV | XXX TBD | (0, 0, 0, 1) | (u, v, 1, 1) |
3715 | | | [#envmap-bumpmap]_ | |
3716 +--------------------+--------------+--------------------+--------------+
3717 | Z | XXX TBD | (z, z, z, 1) | (0, z, 0, 1) |
3718 | | | [#depth-tex-mode]_ | |
3719 +--------------------+--------------+--------------------+--------------+
3720 | S | (s, s, s, s) | unknown | unknown |
3721 +--------------------+--------------+--------------------+--------------+
3723 .. [#envmap-bumpmap] http://www.opengl.org/registry/specs/ATI/envmap_bumpmap.txt
3724 .. [#depth-tex-mode] the default is (z, z, z, 1) but may also be (0, 0, 0, z)
3725 or (z, z, z, z) depending on the value of GL_DEPTH_TEXTURE_MODE.