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}
354 .. opcode:: LDEXP - Multiply Number by Integral Power of 2
360 dst.x = src0.x * 2^{src1.x}
361 dst.y = src0.y * 2^{src1.y}
362 dst.z = src0.z * 2^{src1.z}
363 dst.w = src0.w * 2^{src1.w}
366 .. opcode:: COS - Cosine
368 This instruction replicates its result.
375 .. opcode:: DDX, DDX_FINE - Derivative Relative To X
377 The fine variant is only used when ``PIPE_CAP_TGSI_FS_FINE_DERIVATIVE`` is
378 advertised. When it is, the fine version guarantees one derivative per row
379 while DDX is allowed to be the same for the entire 2x2 quad.
383 dst.x = partialx(src.x)
385 dst.y = partialx(src.y)
387 dst.z = partialx(src.z)
389 dst.w = partialx(src.w)
392 .. opcode:: DDY, DDY_FINE - Derivative Relative To Y
394 The fine variant is only used when ``PIPE_CAP_TGSI_FS_FINE_DERIVATIVE`` is
395 advertised. When it is, the fine version guarantees one derivative per column
396 while DDY is allowed to be the same for the entire 2x2 quad.
400 dst.x = partialy(src.x)
402 dst.y = partialy(src.y)
404 dst.z = partialy(src.z)
406 dst.w = partialy(src.w)
409 .. opcode:: PK2H - Pack Two 16-bit Floats
411 This instruction replicates its result.
415 dst = f32\_to\_f16(src.x) | f32\_to\_f16(src.y) << 16
418 .. opcode:: PK2US - Pack Two Unsigned 16-bit Scalars
423 .. opcode:: PK4B - Pack Four Signed 8-bit Scalars
428 .. opcode:: PK4UB - Pack Four Unsigned 8-bit Scalars
433 .. opcode:: SEQ - Set On Equal
437 dst.x = (src0.x == src1.x) ? 1.0F : 0.0F
439 dst.y = (src0.y == src1.y) ? 1.0F : 0.0F
441 dst.z = (src0.z == src1.z) ? 1.0F : 0.0F
443 dst.w = (src0.w == src1.w) ? 1.0F : 0.0F
446 .. opcode:: SGT - Set On Greater Than
450 dst.x = (src0.x > src1.x) ? 1.0F : 0.0F
452 dst.y = (src0.y > src1.y) ? 1.0F : 0.0F
454 dst.z = (src0.z > src1.z) ? 1.0F : 0.0F
456 dst.w = (src0.w > src1.w) ? 1.0F : 0.0F
459 .. opcode:: SIN - Sine
461 This instruction replicates its result.
468 .. opcode:: SLE - Set On Less Equal Than
472 dst.x = (src0.x <= src1.x) ? 1.0F : 0.0F
474 dst.y = (src0.y <= src1.y) ? 1.0F : 0.0F
476 dst.z = (src0.z <= src1.z) ? 1.0F : 0.0F
478 dst.w = (src0.w <= src1.w) ? 1.0F : 0.0F
481 .. opcode:: SNE - Set On Not Equal
485 dst.x = (src0.x != src1.x) ? 1.0F : 0.0F
487 dst.y = (src0.y != src1.y) ? 1.0F : 0.0F
489 dst.z = (src0.z != src1.z) ? 1.0F : 0.0F
491 dst.w = (src0.w != src1.w) ? 1.0F : 0.0F
494 .. opcode:: TEX - Texture Lookup
496 for array textures src0.y contains the slice for 1D,
497 and src0.z contain the slice for 2D.
499 for shadow textures with no arrays (and not cube map),
500 src0.z contains the reference value.
502 for shadow textures with arrays, src0.z contains
503 the reference value for 1D arrays, and src0.w contains
504 the reference value for 2D arrays and cube maps.
506 for cube map array shadow textures, the reference value
507 cannot be passed in src0.w, and TEX2 must be used instead.
513 shadow_ref = src0.z or src0.w (optional)
517 dst = texture\_sample(unit, coord, shadow_ref)
520 .. opcode:: TEX2 - Texture Lookup (for shadow cube map arrays only)
522 this is the same as TEX, but uses another reg to encode the
533 dst = texture\_sample(unit, coord, shadow_ref)
538 .. opcode:: TXD - Texture Lookup with Derivatives
550 dst = texture\_sample\_deriv(unit, coord, ddx, ddy)
553 .. opcode:: TXP - Projective Texture Lookup
557 coord.x = src0.x / src0.w
559 coord.y = src0.y / src0.w
561 coord.z = src0.z / src0.w
567 dst = texture\_sample(unit, coord)
570 .. opcode:: UP2H - Unpack Two 16-Bit Floats
574 dst.x = f16\_to\_f32(src0.x \& 0xffff)
576 dst.y = f16\_to\_f32(src0.x >> 16)
578 dst.z = f16\_to\_f32(src0.x \& 0xffff)
580 dst.w = f16\_to\_f32(src0.x >> 16)
584 Considered for removal.
586 .. opcode:: UP2US - Unpack Two Unsigned 16-Bit Scalars
592 Considered for removal.
594 .. opcode:: UP4B - Unpack Four Signed 8-Bit Values
600 Considered for removal.
602 .. opcode:: UP4UB - Unpack Four Unsigned 8-Bit Scalars
608 Considered for removal.
611 .. opcode:: ARR - Address Register Load With Round
615 dst.x = (int) round(src.x)
617 dst.y = (int) round(src.y)
619 dst.z = (int) round(src.z)
621 dst.w = (int) round(src.w)
624 .. opcode:: SSG - Set Sign
628 dst.x = (src.x > 0) ? 1 : (src.x < 0) ? -1 : 0
630 dst.y = (src.y > 0) ? 1 : (src.y < 0) ? -1 : 0
632 dst.z = (src.z > 0) ? 1 : (src.z < 0) ? -1 : 0
634 dst.w = (src.w > 0) ? 1 : (src.w < 0) ? -1 : 0
637 .. opcode:: CMP - Compare
641 dst.x = (src0.x < 0) ? src1.x : src2.x
643 dst.y = (src0.y < 0) ? src1.y : src2.y
645 dst.z = (src0.z < 0) ? src1.z : src2.z
647 dst.w = (src0.w < 0) ? src1.w : src2.w
650 .. opcode:: KILL_IF - Conditional Discard
652 Conditional discard. Allowed in fragment shaders only.
656 if (src.x < 0 || src.y < 0 || src.z < 0 || src.w < 0)
661 .. opcode:: KILL - Discard
663 Unconditional discard. Allowed in fragment shaders only.
666 .. opcode:: TXB - Texture Lookup With Bias
668 for cube map array textures and shadow cube maps, the bias value
669 cannot be passed in src0.w, and TXB2 must be used instead.
671 if the target is a shadow texture, the reference value is always
672 in src.z (this prevents shadow 3d and shadow 2d arrays from
673 using this instruction, but this is not needed).
689 dst = texture\_sample(unit, coord, bias)
692 .. opcode:: TXB2 - Texture Lookup With Bias (some cube maps only)
694 this is the same as TXB, but uses another reg to encode the
695 lod bias value for cube map arrays and shadow cube maps.
696 Presumably shadow 2d arrays and shadow 3d targets could use
697 this encoding too, but this is not legal.
699 shadow cube map arrays are neither possible nor required.
709 dst = texture\_sample(unit, coord, bias)
712 .. opcode:: DIV - Divide
716 dst.x = \frac{src0.x}{src1.x}
718 dst.y = \frac{src0.y}{src1.y}
720 dst.z = \frac{src0.z}{src1.z}
722 dst.w = \frac{src0.w}{src1.w}
725 .. opcode:: DP2 - 2-component Dot Product
727 This instruction replicates its result.
731 dst = src0.x \times src1.x + src0.y \times src1.y
734 .. opcode:: TEX_LZ - Texture Lookup With LOD = 0
736 This is the same as TXL with LOD = 0. Like every texture opcode, it obeys
737 pipe_sampler_view::u.tex.first_level and pipe_sampler_state::min_lod.
738 There is no way to override those two in shaders.
754 dst = texture\_sample(unit, coord, lod)
757 .. opcode:: TXL - Texture Lookup With explicit LOD
759 for cube map array textures, the explicit lod value
760 cannot be passed in src0.w, and TXL2 must be used instead.
762 if the target is a shadow texture, the reference value is always
763 in src.z (this prevents shadow 3d / 2d array / cube targets from
764 using this instruction, but this is not needed).
780 dst = texture\_sample(unit, coord, lod)
783 .. opcode:: TXL2 - Texture Lookup With explicit LOD (for cube map arrays only)
785 this is the same as TXL, but uses another reg to encode the
787 Presumably shadow 3d / 2d array / cube targets could use
788 this encoding too, but this is not legal.
790 shadow cube map arrays are neither possible nor required.
800 dst = texture\_sample(unit, coord, lod)
804 ^^^^^^^^^^^^^^^^^^^^^^^^
806 These opcodes are primarily provided for special-use computational shaders.
807 Support for these opcodes indicated by a special pipe capability bit (TBD).
809 XXX doesn't look like most of the opcodes really belong here.
811 .. opcode:: CEIL - Ceiling
815 dst.x = \lceil src.x\rceil
817 dst.y = \lceil src.y\rceil
819 dst.z = \lceil src.z\rceil
821 dst.w = \lceil src.w\rceil
824 .. opcode:: TRUNC - Truncate
837 .. opcode:: MOD - Modulus
841 dst.x = src0.x \bmod src1.x
843 dst.y = src0.y \bmod src1.y
845 dst.z = src0.z \bmod src1.z
847 dst.w = src0.w \bmod src1.w
850 .. opcode:: UARL - Integer Address Register Load
852 Moves the contents of the source register, assumed to be an integer, into the
853 destination register, which is assumed to be an address (ADDR) register.
856 .. opcode:: TXF - Texel Fetch
858 As per NV_gpu_shader4, extract a single texel from a specified texture
859 image or PIPE_BUFFER resource. The source sampler may not be a CUBE or
861 four-component signed integer vector used to identify the single texel
862 accessed. 3 components + level. If the texture is multisampled, then
863 the fourth component indicates the sample, not the mipmap level.
864 Just like texture instructions, an optional
865 offset vector is provided, which is subject to various driver restrictions
866 (regarding range, source of offsets). This instruction ignores the sampler
869 TXF(uint_vec coord, int_vec offset).
872 .. opcode:: TXQ - Texture Size Query
874 As per NV_gpu_program4, retrieve the dimensions of the texture depending on
875 the target. For 1D (width), 2D/RECT/CUBE (width, height), 3D (width, height,
876 depth), 1D array (width, layers), 2D array (width, height, layers).
877 Also return the number of accessible levels (last_level - first_level + 1)
880 For components which don't return a resource dimension, their value
887 dst.x = texture\_width(unit, lod)
889 dst.y = texture\_height(unit, lod)
891 dst.z = texture\_depth(unit, lod)
893 dst.w = texture\_levels(unit)
896 .. opcode:: TXQS - Texture Samples Query
898 This retrieves the number of samples in the texture, and stores it
899 into the x component as an unsigned integer. The other components are
900 undefined. If the texture is not multisampled, this function returns
901 (1, undef, undef, undef).
905 dst.x = texture\_samples(unit)
908 .. opcode:: TG4 - Texture Gather
910 As per ARB_texture_gather, gathers the four texels to be used in a bi-linear
911 filtering operation and packs them into a single register. Only works with
912 2D, 2D array, cubemaps, and cubemaps arrays. For 2D textures, only the
913 addressing modes of the sampler and the top level of any mip pyramid are
914 used. Set W to zero. It behaves like the TEX instruction, but a filtered
915 sample is not generated. The four samples that contribute to filtering are
916 placed into xyzw in clockwise order, starting with the (u,v) texture
917 coordinate delta at the following locations (-, +), (+, +), (+, -), (-, -),
918 where the magnitude of the deltas are half a texel.
920 PIPE_CAP_TEXTURE_SM5 enhances this instruction to support shadow per-sample
921 depth compares, single component selection, and a non-constant offset. It
922 doesn't allow support for the GL independent offset to get i0,j0. This would
923 require another CAP is hw can do it natively. For now we lower that before
932 dst = texture\_gather4 (unit, coord, component)
934 (with SM5 - cube array shadow)
942 dst = texture\_gather (uint, coord, compare)
944 .. opcode:: LODQ - level of detail query
946 Compute the LOD information that the texture pipe would use to access the
947 texture. The Y component contains the computed LOD lambda_prime. The X
948 component contains the LOD that will be accessed, based on min/max lod's
955 dst.xy = lodq(uint, coord);
957 .. opcode:: CLOCK - retrieve the current shader time
959 Invoking this instruction multiple times in the same shader should
960 cause monotonically increasing values to be returned. The values
961 are implicitly 64-bit, so if fewer than 64 bits of precision are
962 available, to provide expected wraparound semantics, the value
963 should be shifted up so that the most significant bit of the time
964 is the most significant bit of the 64-bit value.
972 ^^^^^^^^^^^^^^^^^^^^^^^^
973 These opcodes are used for integer operations.
974 Support for these opcodes indicated by PIPE_SHADER_CAP_INTEGERS (all of them?)
977 .. opcode:: I2F - Signed Integer To Float
979 Rounding is unspecified (round to nearest even suggested).
983 dst.x = (float) src.x
985 dst.y = (float) src.y
987 dst.z = (float) src.z
989 dst.w = (float) src.w
992 .. opcode:: U2F - Unsigned Integer To Float
994 Rounding is unspecified (round to nearest even suggested).
998 dst.x = (float) src.x
1000 dst.y = (float) src.y
1002 dst.z = (float) src.z
1004 dst.w = (float) src.w
1007 .. opcode:: F2I - Float to Signed Integer
1009 Rounding is towards zero (truncate).
1010 Values outside signed range (including NaNs) produce undefined results.
1023 .. opcode:: F2U - Float to Unsigned Integer
1025 Rounding is towards zero (truncate).
1026 Values outside unsigned range (including NaNs) produce undefined results.
1030 dst.x = (unsigned) src.x
1032 dst.y = (unsigned) src.y
1034 dst.z = (unsigned) src.z
1036 dst.w = (unsigned) src.w
1039 .. opcode:: UADD - Integer Add
1041 This instruction works the same for signed and unsigned integers.
1042 The low 32bit of the result is returned.
1046 dst.x = src0.x + src1.x
1048 dst.y = src0.y + src1.y
1050 dst.z = src0.z + src1.z
1052 dst.w = src0.w + src1.w
1055 .. opcode:: UMAD - Integer Multiply And Add
1057 This instruction works the same for signed and unsigned integers.
1058 The multiplication returns the low 32bit (as does the result itself).
1062 dst.x = src0.x \times src1.x + src2.x
1064 dst.y = src0.y \times src1.y + src2.y
1066 dst.z = src0.z \times src1.z + src2.z
1068 dst.w = src0.w \times src1.w + src2.w
1071 .. opcode:: UMUL - Integer Multiply
1073 This instruction works the same for signed and unsigned integers.
1074 The low 32bit of the result is returned.
1078 dst.x = src0.x \times src1.x
1080 dst.y = src0.y \times src1.y
1082 dst.z = src0.z \times src1.z
1084 dst.w = src0.w \times src1.w
1087 .. opcode:: IMUL_HI - Signed Integer Multiply High Bits
1089 The high 32bits of the multiplication of 2 signed integers are returned.
1093 dst.x = (src0.x \times src1.x) >> 32
1095 dst.y = (src0.y \times src1.y) >> 32
1097 dst.z = (src0.z \times src1.z) >> 32
1099 dst.w = (src0.w \times src1.w) >> 32
1102 .. opcode:: UMUL_HI - Unsigned Integer Multiply High Bits
1104 The high 32bits of the multiplication of 2 unsigned integers are returned.
1108 dst.x = (src0.x \times src1.x) >> 32
1110 dst.y = (src0.y \times src1.y) >> 32
1112 dst.z = (src0.z \times src1.z) >> 32
1114 dst.w = (src0.w \times src1.w) >> 32
1117 .. opcode:: IDIV - Signed Integer Division
1119 TBD: behavior for division by zero.
1123 dst.x = \frac{src0.x}{src1.x}
1125 dst.y = \frac{src0.y}{src1.y}
1127 dst.z = \frac{src0.z}{src1.z}
1129 dst.w = \frac{src0.w}{src1.w}
1132 .. opcode:: UDIV - Unsigned Integer Division
1134 For division by zero, 0xffffffff is returned.
1138 dst.x = \frac{src0.x}{src1.x}
1140 dst.y = \frac{src0.y}{src1.y}
1142 dst.z = \frac{src0.z}{src1.z}
1144 dst.w = \frac{src0.w}{src1.w}
1147 .. opcode:: UMOD - Unsigned Integer Remainder
1149 If second arg is zero, 0xffffffff is returned.
1153 dst.x = src0.x \bmod src1.x
1155 dst.y = src0.y \bmod src1.y
1157 dst.z = src0.z \bmod src1.z
1159 dst.w = src0.w \bmod src1.w
1162 .. opcode:: NOT - Bitwise Not
1175 .. opcode:: AND - Bitwise And
1179 dst.x = src0.x \& src1.x
1181 dst.y = src0.y \& src1.y
1183 dst.z = src0.z \& src1.z
1185 dst.w = src0.w \& src1.w
1188 .. opcode:: OR - Bitwise Or
1192 dst.x = src0.x | src1.x
1194 dst.y = src0.y | src1.y
1196 dst.z = src0.z | src1.z
1198 dst.w = src0.w | src1.w
1201 .. opcode:: XOR - Bitwise Xor
1205 dst.x = src0.x \oplus src1.x
1207 dst.y = src0.y \oplus src1.y
1209 dst.z = src0.z \oplus src1.z
1211 dst.w = src0.w \oplus src1.w
1214 .. opcode:: IMAX - Maximum of Signed Integers
1218 dst.x = max(src0.x, src1.x)
1220 dst.y = max(src0.y, src1.y)
1222 dst.z = max(src0.z, src1.z)
1224 dst.w = max(src0.w, src1.w)
1227 .. opcode:: UMAX - Maximum of Unsigned Integers
1231 dst.x = max(src0.x, src1.x)
1233 dst.y = max(src0.y, src1.y)
1235 dst.z = max(src0.z, src1.z)
1237 dst.w = max(src0.w, src1.w)
1240 .. opcode:: IMIN - Minimum of Signed Integers
1244 dst.x = min(src0.x, src1.x)
1246 dst.y = min(src0.y, src1.y)
1248 dst.z = min(src0.z, src1.z)
1250 dst.w = min(src0.w, src1.w)
1253 .. opcode:: UMIN - Minimum of Unsigned Integers
1257 dst.x = min(src0.x, src1.x)
1259 dst.y = min(src0.y, src1.y)
1261 dst.z = min(src0.z, src1.z)
1263 dst.w = min(src0.w, src1.w)
1266 .. opcode:: SHL - Shift Left
1268 The shift count is masked with 0x1f before the shift is applied.
1272 dst.x = src0.x << (0x1f \& src1.x)
1274 dst.y = src0.y << (0x1f \& src1.y)
1276 dst.z = src0.z << (0x1f \& src1.z)
1278 dst.w = src0.w << (0x1f \& src1.w)
1281 .. opcode:: ISHR - Arithmetic Shift Right (of Signed Integer)
1283 The shift count is masked with 0x1f before the shift is applied.
1287 dst.x = src0.x >> (0x1f \& src1.x)
1289 dst.y = src0.y >> (0x1f \& src1.y)
1291 dst.z = src0.z >> (0x1f \& src1.z)
1293 dst.w = src0.w >> (0x1f \& src1.w)
1296 .. opcode:: USHR - Logical Shift Right
1298 The shift count is masked with 0x1f before the shift is applied.
1302 dst.x = src0.x >> (unsigned) (0x1f \& src1.x)
1304 dst.y = src0.y >> (unsigned) (0x1f \& src1.y)
1306 dst.z = src0.z >> (unsigned) (0x1f \& src1.z)
1308 dst.w = src0.w >> (unsigned) (0x1f \& src1.w)
1311 .. opcode:: UCMP - Integer Conditional Move
1315 dst.x = src0.x ? src1.x : src2.x
1317 dst.y = src0.y ? src1.y : src2.y
1319 dst.z = src0.z ? src1.z : src2.z
1321 dst.w = src0.w ? src1.w : src2.w
1325 .. opcode:: ISSG - Integer Set Sign
1329 dst.x = (src0.x < 0) ? -1 : (src0.x > 0) ? 1 : 0
1331 dst.y = (src0.y < 0) ? -1 : (src0.y > 0) ? 1 : 0
1333 dst.z = (src0.z < 0) ? -1 : (src0.z > 0) ? 1 : 0
1335 dst.w = (src0.w < 0) ? -1 : (src0.w > 0) ? 1 : 0
1339 .. opcode:: FSLT - Float Set On Less Than (ordered)
1341 Same comparison as SLT but returns integer instead of 1.0/0.0 float
1345 dst.x = (src0.x < src1.x) ? \sim 0 : 0
1347 dst.y = (src0.y < src1.y) ? \sim 0 : 0
1349 dst.z = (src0.z < src1.z) ? \sim 0 : 0
1351 dst.w = (src0.w < src1.w) ? \sim 0 : 0
1354 .. opcode:: ISLT - Signed Integer Set On Less Than
1358 dst.x = (src0.x < src1.x) ? \sim 0 : 0
1360 dst.y = (src0.y < src1.y) ? \sim 0 : 0
1362 dst.z = (src0.z < src1.z) ? \sim 0 : 0
1364 dst.w = (src0.w < src1.w) ? \sim 0 : 0
1367 .. opcode:: USLT - Unsigned Integer Set On Less Than
1371 dst.x = (src0.x < src1.x) ? \sim 0 : 0
1373 dst.y = (src0.y < src1.y) ? \sim 0 : 0
1375 dst.z = (src0.z < src1.z) ? \sim 0 : 0
1377 dst.w = (src0.w < src1.w) ? \sim 0 : 0
1380 .. opcode:: FSGE - Float Set On Greater Equal Than (ordered)
1382 Same comparison as SGE but returns integer instead of 1.0/0.0 float
1386 dst.x = (src0.x >= src1.x) ? \sim 0 : 0
1388 dst.y = (src0.y >= src1.y) ? \sim 0 : 0
1390 dst.z = (src0.z >= src1.z) ? \sim 0 : 0
1392 dst.w = (src0.w >= src1.w) ? \sim 0 : 0
1395 .. opcode:: ISGE - Signed Integer Set On Greater Equal Than
1399 dst.x = (src0.x >= src1.x) ? \sim 0 : 0
1401 dst.y = (src0.y >= src1.y) ? \sim 0 : 0
1403 dst.z = (src0.z >= src1.z) ? \sim 0 : 0
1405 dst.w = (src0.w >= src1.w) ? \sim 0 : 0
1408 .. opcode:: USGE - Unsigned Integer Set On Greater Equal Than
1412 dst.x = (src0.x >= src1.x) ? \sim 0 : 0
1414 dst.y = (src0.y >= src1.y) ? \sim 0 : 0
1416 dst.z = (src0.z >= src1.z) ? \sim 0 : 0
1418 dst.w = (src0.w >= src1.w) ? \sim 0 : 0
1421 .. opcode:: FSEQ - Float Set On Equal (ordered)
1423 Same comparison as SEQ but returns integer instead of 1.0/0.0 float
1427 dst.x = (src0.x == src1.x) ? \sim 0 : 0
1429 dst.y = (src0.y == src1.y) ? \sim 0 : 0
1431 dst.z = (src0.z == src1.z) ? \sim 0 : 0
1433 dst.w = (src0.w == src1.w) ? \sim 0 : 0
1436 .. opcode:: USEQ - Integer Set On Equal
1440 dst.x = (src0.x == src1.x) ? \sim 0 : 0
1442 dst.y = (src0.y == src1.y) ? \sim 0 : 0
1444 dst.z = (src0.z == src1.z) ? \sim 0 : 0
1446 dst.w = (src0.w == src1.w) ? \sim 0 : 0
1449 .. opcode:: FSNE - Float Set On Not Equal (unordered)
1451 Same comparison as SNE but returns integer instead of 1.0/0.0 float
1455 dst.x = (src0.x != src1.x) ? \sim 0 : 0
1457 dst.y = (src0.y != src1.y) ? \sim 0 : 0
1459 dst.z = (src0.z != src1.z) ? \sim 0 : 0
1461 dst.w = (src0.w != src1.w) ? \sim 0 : 0
1464 .. opcode:: USNE - Integer Set On Not Equal
1468 dst.x = (src0.x != src1.x) ? \sim 0 : 0
1470 dst.y = (src0.y != src1.y) ? \sim 0 : 0
1472 dst.z = (src0.z != src1.z) ? \sim 0 : 0
1474 dst.w = (src0.w != src1.w) ? \sim 0 : 0
1477 .. opcode:: INEG - Integer Negate
1492 .. opcode:: IABS - Integer Absolute Value
1506 These opcodes are used for bit-level manipulation of integers.
1508 .. opcode:: IBFE - Signed Bitfield Extract
1510 Like GLSL bitfieldExtract. Extracts a set of bits from the input, and
1511 sign-extends them if the high bit of the extracted window is set.
1515 def ibfe(value, offset, bits):
1516 if offset < 0 or bits < 0 or offset + bits > 32:
1518 if bits == 0: return 0
1519 # Note: >> sign-extends
1520 return (value << (32 - offset - bits)) >> (32 - bits)
1522 .. opcode:: UBFE - Unsigned Bitfield Extract
1524 Like GLSL bitfieldExtract. Extracts a set of bits from the input, without
1529 def ubfe(value, offset, bits):
1530 if offset < 0 or bits < 0 or offset + bits > 32:
1532 if bits == 0: return 0
1533 # Note: >> does not sign-extend
1534 return (value << (32 - offset - bits)) >> (32 - bits)
1536 .. opcode:: BFI - Bitfield Insert
1538 Like GLSL bitfieldInsert. Replaces a bit region of 'base' with the low bits
1543 def bfi(base, insert, offset, bits):
1544 if offset < 0 or bits < 0 or offset + bits > 32:
1546 # << defined such that mask == ~0 when bits == 32, offset == 0
1547 mask = ((1 << bits) - 1) << offset
1548 return ((insert << offset) & mask) | (base & ~mask)
1550 .. opcode:: BREV - Bitfield Reverse
1552 See SM5 instruction BFREV. Reverses the bits of the argument.
1554 .. opcode:: POPC - Population Count
1556 See SM5 instruction COUNTBITS. Counts the number of set bits in the argument.
1558 .. opcode:: LSB - Index of lowest set bit
1560 See SM5 instruction FIRSTBIT_LO. Computes the 0-based index of the first set
1561 bit of the argument. Returns -1 if none are set.
1563 .. opcode:: IMSB - Index of highest non-sign bit
1565 See SM5 instruction FIRSTBIT_SHI. Computes the 0-based index of the highest
1566 non-sign bit of the argument (i.e. highest 0 bit for negative numbers,
1567 highest 1 bit for positive numbers). Returns -1 if all bits are the same
1568 (i.e. for inputs 0 and -1).
1570 .. opcode:: UMSB - Index of highest set bit
1572 See SM5 instruction FIRSTBIT_HI. Computes the 0-based index of the highest
1573 set bit of the argument. Returns -1 if none are set.
1576 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1578 These opcodes are only supported in geometry shaders; they have no meaning
1579 in any other type of shader.
1581 .. opcode:: EMIT - Emit
1583 Generate a new vertex for the current primitive into the specified vertex
1584 stream using the values in the output registers.
1587 .. opcode:: ENDPRIM - End Primitive
1589 Complete the current primitive in the specified vertex stream (consisting of
1590 the emitted vertices), and start a new one.
1596 These opcodes are part of :term:`GLSL`'s opcode set. Support for these
1597 opcodes is determined by a special capability bit, ``GLSL``.
1598 Some require glsl version 1.30 (UIF/SWITCH/CASE/DEFAULT/ENDSWITCH).
1600 .. opcode:: CAL - Subroutine Call
1606 .. opcode:: RET - Subroutine Call Return
1611 .. opcode:: CONT - Continue
1613 Unconditionally moves the point of execution to the instruction after the
1614 last bgnloop. The instruction must appear within a bgnloop/endloop.
1618 Support for CONT is determined by a special capability bit,
1619 ``TGSI_CONT_SUPPORTED``. See :ref:`Screen` for more information.
1622 .. opcode:: BGNLOOP - Begin a Loop
1624 Start a loop. Must have a matching endloop.
1627 .. opcode:: BGNSUB - Begin Subroutine
1629 Starts definition of a subroutine. Must have a matching endsub.
1632 .. opcode:: ENDLOOP - End a Loop
1634 End a loop started with bgnloop.
1637 .. opcode:: ENDSUB - End Subroutine
1639 Ends definition of a subroutine.
1642 .. opcode:: NOP - No Operation
1647 .. opcode:: BRK - Break
1649 Unconditionally moves the point of execution to the instruction after the
1650 next endloop or endswitch. The instruction must appear within a loop/endloop
1651 or switch/endswitch.
1654 .. opcode:: IF - Float If
1656 Start an IF ... ELSE .. ENDIF block. Condition evaluates to true if
1660 where src0.x is interpreted as a floating point register.
1663 .. opcode:: UIF - Bitwise If
1665 Start an UIF ... ELSE .. ENDIF block. Condition evaluates to true if
1669 where src0.x is interpreted as an integer register.
1672 .. opcode:: ELSE - Else
1674 Starts an else block, after an IF or UIF statement.
1677 .. opcode:: ENDIF - End If
1679 Ends an IF or UIF block.
1682 .. opcode:: SWITCH - Switch
1684 Starts a C-style switch expression. The switch consists of one or multiple
1685 CASE statements, and at most one DEFAULT statement. Execution of a statement
1686 ends when a BRK is hit, but just like in C falling through to other cases
1687 without a break is allowed. Similarly, DEFAULT label is allowed anywhere not
1688 just as last statement, and fallthrough is allowed into/from it.
1689 CASE src arguments are evaluated at bit level against the SWITCH src argument.
1695 (some instructions here)
1698 (some instructions here)
1701 (some instructions here)
1706 .. opcode:: CASE - Switch case
1708 This represents a switch case label. The src arg must be an integer immediate.
1711 .. opcode:: DEFAULT - Switch default
1713 This represents the default case in the switch, which is taken if no other
1717 .. opcode:: ENDSWITCH - End of switch
1719 Ends a switch expression.
1725 The interpolation instructions allow an input to be interpolated in a
1726 different way than its declaration. This corresponds to the GLSL 4.00
1727 interpolateAt* functions. The first argument of each of these must come from
1728 ``TGSI_FILE_INPUT``.
1730 .. opcode:: INTERP_CENTROID - Interpolate at the centroid
1732 Interpolates the varying specified by src0 at the centroid
1734 .. opcode:: INTERP_SAMPLE - Interpolate at the specified sample
1736 Interpolates the varying specified by src0 at the sample id specified by
1737 src1.x (interpreted as an integer)
1739 .. opcode:: INTERP_OFFSET - Interpolate at the specified offset
1741 Interpolates the varying specified by src0 at the offset src1.xy from the
1742 pixel center (interpreted as floats)
1750 The double-precision opcodes reinterpret four-component vectors into
1751 two-component vectors with doubled precision in each component.
1753 .. opcode:: DABS - Absolute
1761 .. opcode:: DADD - Add
1765 dst.xy = src0.xy + src1.xy
1767 dst.zw = src0.zw + src1.zw
1769 .. opcode:: DSEQ - Set on Equal
1773 dst.x = src0.xy == src1.xy ? \sim 0 : 0
1775 dst.z = src0.zw == src1.zw ? \sim 0 : 0
1777 .. opcode:: DSNE - Set on Not Equal
1781 dst.x = src0.xy != src1.xy ? \sim 0 : 0
1783 dst.z = src0.zw != src1.zw ? \sim 0 : 0
1785 .. opcode:: DSLT - Set on Less than
1789 dst.x = src0.xy < src1.xy ? \sim 0 : 0
1791 dst.z = src0.zw < src1.zw ? \sim 0 : 0
1793 .. opcode:: DSGE - Set on Greater equal
1797 dst.x = src0.xy >= src1.xy ? \sim 0 : 0
1799 dst.z = src0.zw >= src1.zw ? \sim 0 : 0
1801 .. opcode:: DFRAC - Fraction
1805 dst.xy = src.xy - \lfloor src.xy\rfloor
1807 dst.zw = src.zw - \lfloor src.zw\rfloor
1809 .. opcode:: DTRUNC - Truncate
1813 dst.xy = trunc(src.xy)
1815 dst.zw = trunc(src.zw)
1817 .. opcode:: DCEIL - Ceiling
1821 dst.xy = \lceil src.xy\rceil
1823 dst.zw = \lceil src.zw\rceil
1825 .. opcode:: DFLR - Floor
1829 dst.xy = \lfloor src.xy\rfloor
1831 dst.zw = \lfloor src.zw\rfloor
1833 .. opcode:: DROUND - Fraction
1837 dst.xy = round(src.xy)
1839 dst.zw = round(src.zw)
1841 .. opcode:: DSSG - Set Sign
1845 dst.xy = (src.xy > 0) ? 1.0 : (src.xy < 0) ? -1.0 : 0.0
1847 dst.zw = (src.zw > 0) ? 1.0 : (src.zw < 0) ? -1.0 : 0.0
1849 .. opcode:: DFRACEXP - Convert Number to Fractional and Integral Components
1851 Like the ``frexp()`` routine in many math libraries, this opcode stores the
1852 exponent of its source to ``dst0``, and the significand to ``dst1``, such that
1853 :math:`dst1 \times 2^{dst0} = src` . The results are replicated across
1858 dst0.xy = dst.zw = frac(src.xy)
1863 .. opcode:: DLDEXP - Multiply Number by Integral Power of 2
1865 This opcode is the inverse of :opcode:`DFRACEXP`. The second
1866 source is an integer.
1870 dst.xy = src0.xy \times 2^{src1.x}
1872 dst.zw = src0.zw \times 2^{src1.z}
1874 .. opcode:: DMIN - Minimum
1878 dst.xy = min(src0.xy, src1.xy)
1880 dst.zw = min(src0.zw, src1.zw)
1882 .. opcode:: DMAX - Maximum
1886 dst.xy = max(src0.xy, src1.xy)
1888 dst.zw = max(src0.zw, src1.zw)
1890 .. opcode:: DMUL - Multiply
1894 dst.xy = src0.xy \times src1.xy
1896 dst.zw = src0.zw \times src1.zw
1899 .. opcode:: DMAD - Multiply And Add
1903 dst.xy = src0.xy \times src1.xy + src2.xy
1905 dst.zw = src0.zw \times src1.zw + src2.zw
1908 .. opcode:: DFMA - Fused Multiply-Add
1910 Perform a * b + c with no intermediate rounding step.
1914 dst.xy = src0.xy \times src1.xy + src2.xy
1916 dst.zw = src0.zw \times src1.zw + src2.zw
1919 .. opcode:: DDIV - Divide
1923 dst.xy = \frac{src0.xy}{src1.xy}
1925 dst.zw = \frac{src0.zw}{src1.zw}
1928 .. opcode:: DRCP - Reciprocal
1932 dst.xy = \frac{1}{src.xy}
1934 dst.zw = \frac{1}{src.zw}
1936 .. opcode:: DSQRT - Square Root
1940 dst.xy = \sqrt{src.xy}
1942 dst.zw = \sqrt{src.zw}
1944 .. opcode:: DRSQ - Reciprocal Square Root
1948 dst.xy = \frac{1}{\sqrt{src.xy}}
1950 dst.zw = \frac{1}{\sqrt{src.zw}}
1952 .. opcode:: F2D - Float to Double
1956 dst.xy = double(src0.x)
1958 dst.zw = double(src0.y)
1960 .. opcode:: D2F - Double to Float
1964 dst.x = float(src0.xy)
1966 dst.y = float(src0.zw)
1968 .. opcode:: I2D - Int to Double
1972 dst.xy = double(src0.x)
1974 dst.zw = double(src0.y)
1976 .. opcode:: D2I - Double to Int
1980 dst.x = int(src0.xy)
1982 dst.y = int(src0.zw)
1984 .. opcode:: U2D - Unsigned Int to Double
1988 dst.xy = double(src0.x)
1990 dst.zw = double(src0.y)
1992 .. opcode:: D2U - Double to Unsigned Int
1996 dst.x = unsigned(src0.xy)
1998 dst.y = unsigned(src0.zw)
2003 The 64-bit integer opcodes reinterpret four-component vectors into
2004 two-component vectors with 64-bits in each component.
2006 .. opcode:: I64ABS - 64-bit Integer Absolute Value
2014 .. opcode:: I64NEG - 64-bit Integer Negate
2024 .. opcode:: I64SSG - 64-bit Integer Set Sign
2028 dst.xy = (src0.xy < 0) ? -1 : (src0.xy > 0) ? 1 : 0
2030 dst.zw = (src0.zw < 0) ? -1 : (src0.zw > 0) ? 1 : 0
2032 .. opcode:: U64ADD - 64-bit Integer Add
2036 dst.xy = src0.xy + src1.xy
2038 dst.zw = src0.zw + src1.zw
2040 .. opcode:: U64MUL - 64-bit Integer Multiply
2044 dst.xy = src0.xy * src1.xy
2046 dst.zw = src0.zw * src1.zw
2048 .. opcode:: U64SEQ - 64-bit Integer Set on Equal
2052 dst.x = src0.xy == src1.xy ? \sim 0 : 0
2054 dst.z = src0.zw == src1.zw ? \sim 0 : 0
2056 .. opcode:: U64SNE - 64-bit Integer Set on Not Equal
2060 dst.x = src0.xy != src1.xy ? \sim 0 : 0
2062 dst.z = src0.zw != src1.zw ? \sim 0 : 0
2064 .. opcode:: U64SLT - 64-bit Unsigned Integer Set on Less Than
2068 dst.x = src0.xy < src1.xy ? \sim 0 : 0
2070 dst.z = src0.zw < src1.zw ? \sim 0 : 0
2072 .. opcode:: U64SGE - 64-bit Unsigned Integer Set on Greater Equal
2076 dst.x = src0.xy >= src1.xy ? \sim 0 : 0
2078 dst.z = src0.zw >= src1.zw ? \sim 0 : 0
2080 .. opcode:: I64SLT - 64-bit Signed Integer Set on Less Than
2084 dst.x = src0.xy < src1.xy ? \sim 0 : 0
2086 dst.z = src0.zw < src1.zw ? \sim 0 : 0
2088 .. opcode:: I64SGE - 64-bit Signed Integer Set on Greater Equal
2092 dst.x = src0.xy >= src1.xy ? \sim 0 : 0
2094 dst.z = src0.zw >= src1.zw ? \sim 0 : 0
2096 .. opcode:: I64MIN - Minimum of 64-bit Signed Integers
2100 dst.xy = min(src0.xy, src1.xy)
2102 dst.zw = min(src0.zw, src1.zw)
2104 .. opcode:: U64MIN - Minimum of 64-bit Unsigned Integers
2108 dst.xy = min(src0.xy, src1.xy)
2110 dst.zw = min(src0.zw, src1.zw)
2112 .. opcode:: I64MAX - Maximum of 64-bit Signed Integers
2116 dst.xy = max(src0.xy, src1.xy)
2118 dst.zw = max(src0.zw, src1.zw)
2120 .. opcode:: U64MAX - Maximum of 64-bit Unsigned Integers
2124 dst.xy = max(src0.xy, src1.xy)
2126 dst.zw = max(src0.zw, src1.zw)
2128 .. opcode:: U64SHL - Shift Left 64-bit Unsigned Integer
2130 The shift count is masked with 0x3f before the shift is applied.
2134 dst.xy = src0.xy << (0x3f \& src1.x)
2136 dst.zw = src0.zw << (0x3f \& src1.y)
2138 .. opcode:: I64SHR - Arithmetic Shift Right (of 64-bit Signed Integer)
2140 The shift count is masked with 0x3f before the shift is applied.
2144 dst.xy = src0.xy >> (0x3f \& src1.x)
2146 dst.zw = src0.zw >> (0x3f \& src1.y)
2148 .. opcode:: U64SHR - Logical Shift Right (of 64-bit Unsigned Integer)
2150 The shift count is masked with 0x3f before the shift is applied.
2154 dst.xy = src0.xy >> (unsigned) (0x3f \& src1.x)
2156 dst.zw = src0.zw >> (unsigned) (0x3f \& src1.y)
2158 .. opcode:: I64DIV - 64-bit Signed Integer Division
2162 dst.xy = \frac{src0.xy}{src1.xy}
2164 dst.zw = \frac{src0.zw}{src1.zw}
2166 .. opcode:: U64DIV - 64-bit Unsigned Integer Division
2170 dst.xy = \frac{src0.xy}{src1.xy}
2172 dst.zw = \frac{src0.zw}{src1.zw}
2174 .. opcode:: U64MOD - 64-bit Unsigned Integer Remainder
2178 dst.xy = src0.xy \bmod src1.xy
2180 dst.zw = src0.zw \bmod src1.zw
2182 .. opcode:: I64MOD - 64-bit Signed Integer Remainder
2186 dst.xy = src0.xy \bmod src1.xy
2188 dst.zw = src0.zw \bmod src1.zw
2190 .. opcode:: F2U64 - Float to 64-bit Unsigned Int
2194 dst.xy = (uint64_t) src0.x
2196 dst.zw = (uint64_t) src0.y
2198 .. opcode:: F2I64 - Float to 64-bit Int
2202 dst.xy = (int64_t) src0.x
2204 dst.zw = (int64_t) src0.y
2206 .. opcode:: U2I64 - Unsigned Integer to 64-bit Integer
2208 This is a zero extension.
2212 dst.xy = (int64_t) src0.x
2214 dst.zw = (int64_t) src0.y
2216 .. opcode:: I2I64 - Signed Integer to 64-bit Integer
2218 This is a sign extension.
2222 dst.xy = (int64_t) src0.x
2224 dst.zw = (int64_t) src0.y
2226 .. opcode:: D2U64 - Double to 64-bit Unsigned Int
2230 dst.xy = (uint64_t) src0.xy
2232 dst.zw = (uint64_t) src0.zw
2234 .. opcode:: D2I64 - Double to 64-bit Int
2238 dst.xy = (int64_t) src0.xy
2240 dst.zw = (int64_t) src0.zw
2242 .. opcode:: U642F - 64-bit unsigned integer to float
2246 dst.x = (float) src0.xy
2248 dst.y = (float) src0.zw
2250 .. opcode:: I642F - 64-bit Int to Float
2254 dst.x = (float) src0.xy
2256 dst.y = (float) src0.zw
2258 .. opcode:: U642D - 64-bit unsigned integer to double
2262 dst.xy = (double) src0.xy
2264 dst.zw = (double) src0.zw
2266 .. opcode:: I642D - 64-bit Int to double
2270 dst.xy = (double) src0.xy
2272 dst.zw = (double) src0.zw
2274 .. _samplingopcodes:
2276 Resource Sampling Opcodes
2277 ^^^^^^^^^^^^^^^^^^^^^^^^^
2279 Those opcodes follow very closely semantics of the respective Direct3D
2280 instructions. If in doubt double check Direct3D documentation.
2281 Note that the swizzle on SVIEW (src1) determines texel swizzling
2286 Using provided address, sample data from the specified texture using the
2287 filtering mode identified by the given sampler. The source data may come from
2288 any resource type other than buffers.
2290 Syntax: ``SAMPLE dst, address, sampler_view, sampler``
2292 Example: ``SAMPLE TEMP[0], TEMP[1], SVIEW[0], SAMP[0]``
2294 .. opcode:: SAMPLE_I
2296 Simplified alternative to the SAMPLE instruction. Using the provided
2297 integer address, SAMPLE_I fetches data from the specified sampler view
2298 without any filtering. The source data may come from any resource type
2301 Syntax: ``SAMPLE_I dst, address, sampler_view``
2303 Example: ``SAMPLE_I TEMP[0], TEMP[1], SVIEW[0]``
2305 The 'address' is specified as unsigned integers. If the 'address' is out of
2306 range [0...(# texels - 1)] the result of the fetch is always 0 in all
2307 components. As such the instruction doesn't honor address wrap modes, in
2308 cases where that behavior is desirable 'SAMPLE' instruction should be used.
2309 address.w always provides an unsigned integer mipmap level. If the value is
2310 out of the range then the instruction always returns 0 in all components.
2311 address.yz are ignored for buffers and 1d textures. address.z is ignored
2312 for 1d texture arrays and 2d textures.
2314 For 1D texture arrays address.y provides the array index (also as unsigned
2315 integer). If the value is out of the range of available array indices
2316 [0... (array size - 1)] then the opcode always returns 0 in all components.
2317 For 2D texture arrays address.z provides the array index, otherwise it
2318 exhibits the same behavior as in the case for 1D texture arrays. The exact
2319 semantics of the source address are presented in the table below:
2321 +---------------------------+----+-----+-----+---------+
2322 | resource type | X | Y | Z | W |
2323 +===========================+====+=====+=====+=========+
2324 | ``PIPE_BUFFER`` | x | | | ignored |
2325 +---------------------------+----+-----+-----+---------+
2326 | ``PIPE_TEXTURE_1D`` | x | | | mpl |
2327 +---------------------------+----+-----+-----+---------+
2328 | ``PIPE_TEXTURE_2D`` | x | y | | mpl |
2329 +---------------------------+----+-----+-----+---------+
2330 | ``PIPE_TEXTURE_3D`` | x | y | z | mpl |
2331 +---------------------------+----+-----+-----+---------+
2332 | ``PIPE_TEXTURE_RECT`` | x | y | | mpl |
2333 +---------------------------+----+-----+-----+---------+
2334 | ``PIPE_TEXTURE_CUBE`` | not allowed as source |
2335 +---------------------------+----+-----+-----+---------+
2336 | ``PIPE_TEXTURE_1D_ARRAY`` | x | idx | | mpl |
2337 +---------------------------+----+-----+-----+---------+
2338 | ``PIPE_TEXTURE_2D_ARRAY`` | x | y | idx | mpl |
2339 +---------------------------+----+-----+-----+---------+
2341 Where 'mpl' is a mipmap level and 'idx' is the array index.
2343 .. opcode:: SAMPLE_I_MS
2345 Just like SAMPLE_I but allows fetch data from multi-sampled surfaces.
2347 Syntax: ``SAMPLE_I_MS dst, address, sampler_view, sample``
2349 .. opcode:: SAMPLE_B
2351 Just like the SAMPLE instruction with the exception that an additional bias
2352 is applied to the level of detail computed as part of the instruction
2355 Syntax: ``SAMPLE_B dst, address, sampler_view, sampler, lod_bias``
2357 Example: ``SAMPLE_B TEMP[0], TEMP[1], SVIEW[0], SAMP[0], TEMP[2].x``
2359 .. opcode:: SAMPLE_C
2361 Similar to the SAMPLE instruction but it performs a comparison filter. The
2362 operands to SAMPLE_C are identical to SAMPLE, except that there is an
2363 additional float32 operand, reference value, which must be a register with
2364 single-component, or a scalar literal. SAMPLE_C makes the hardware use the
2365 current samplers compare_func (in pipe_sampler_state) to compare reference
2366 value against the red component value for the surce resource at each texel
2367 that the currently configured texture filter covers based on the provided
2370 Syntax: ``SAMPLE_C dst, address, sampler_view.r, sampler, ref_value``
2372 Example: ``SAMPLE_C TEMP[0], TEMP[1], SVIEW[0].r, SAMP[0], TEMP[2].x``
2374 .. opcode:: SAMPLE_C_LZ
2376 Same as SAMPLE_C, but LOD is 0 and derivatives are ignored. The LZ stands
2379 Syntax: ``SAMPLE_C_LZ dst, address, sampler_view.r, sampler, ref_value``
2381 Example: ``SAMPLE_C_LZ TEMP[0], TEMP[1], SVIEW[0].r, SAMP[0], TEMP[2].x``
2384 .. opcode:: SAMPLE_D
2386 SAMPLE_D is identical to the SAMPLE opcode except that the derivatives for
2387 the source address in the x direction and the y direction are provided by
2390 Syntax: ``SAMPLE_D dst, address, sampler_view, sampler, der_x, der_y``
2392 Example: ``SAMPLE_D TEMP[0], TEMP[1], SVIEW[0], SAMP[0], TEMP[2], TEMP[3]``
2394 .. opcode:: SAMPLE_L
2396 SAMPLE_L is identical to the SAMPLE opcode except that the LOD is provided
2397 directly as a scalar value, representing no anisotropy.
2399 Syntax: ``SAMPLE_L dst, address, sampler_view, sampler, explicit_lod``
2401 Example: ``SAMPLE_L TEMP[0], TEMP[1], SVIEW[0], SAMP[0], TEMP[2].x``
2405 Gathers the four texels to be used in a bi-linear filtering operation and
2406 packs them into a single register. Only works with 2D, 2D array, cubemaps,
2407 and cubemaps arrays. For 2D textures, only the addressing modes of the
2408 sampler and the top level of any mip pyramid are used. Set W to zero. It
2409 behaves like the SAMPLE instruction, but a filtered sample is not
2410 generated. The four samples that contribute to filtering are placed into
2411 xyzw in counter-clockwise order, starting with the (u,v) texture coordinate
2412 delta at the following locations (-, +), (+, +), (+, -), (-, -), where the
2413 magnitude of the deltas are half a texel.
2416 .. opcode:: SVIEWINFO
2418 Query the dimensions of a given sampler view. dst receives width, height,
2419 depth or array size and number of mipmap levels as int4. The dst can have a
2420 writemask which will specify what info is the caller interested in.
2422 Syntax: ``SVIEWINFO dst, src_mip_level, sampler_view``
2424 Example: ``SVIEWINFO TEMP[0], TEMP[1].x, SVIEW[0]``
2426 src_mip_level is an unsigned integer scalar. If it's out of range then
2427 returns 0 for width, height and depth/array size but the total number of
2428 mipmap is still returned correctly for the given sampler view. The returned
2429 width, height and depth values are for the mipmap level selected by the
2430 src_mip_level and are in the number of texels. For 1d texture array width
2431 is in dst.x, array size is in dst.y and dst.z is 0. The number of mipmaps is
2432 still in dst.w. In contrast to d3d10 resinfo, there's no way in the tgsi
2433 instruction encoding to specify the return type (float/rcpfloat/uint), hence
2434 always using uint. Also, unlike the SAMPLE instructions, the swizzle on src1
2435 resinfo allowing swizzling dst values is ignored (due to the interaction
2436 with rcpfloat modifier which requires some swizzle handling in the state
2439 .. opcode:: SAMPLE_POS
2441 Query the position of a sample in the given resource or render target
2442 when per-sample fragment shading is in effect.
2444 Syntax: ``SAMPLE_POS dst, source, sample_index``
2446 dst receives float4 (x, y, undef, undef) indicated where the sample is
2447 located. Sample locations are in the range [0, 1] where 0.5 is the center
2450 source is either a sampler view (to indicate a shader resource) or temp
2451 register (to indicate the render target). The source register may have
2452 an optional swizzle to apply to the returned result
2454 sample_index is an integer scalar indicating which sample position is to
2457 If per-sample shading is not in effect or the source resource or render
2458 target is not multisampled, the result is (0.5, 0.5, undef, undef).
2460 NOTE: no driver has implemented this opcode yet (and no state tracker
2461 emits it). This information is subject to change.
2463 .. opcode:: SAMPLE_INFO
2465 Query the number of samples in a multisampled resource or render target.
2467 Syntax: ``SAMPLE_INFO dst, source``
2469 dst receives int4 (n, 0, 0, 0) where n is the number of samples in a
2470 resource or the render target.
2472 source is either a sampler view (to indicate a shader resource) or temp
2473 register (to indicate the render target). The source register may have
2474 an optional swizzle to apply to the returned result
2476 If per-sample shading is not in effect or the source resource or render
2477 target is not multisampled, the result is (1, 0, 0, 0).
2479 NOTE: no driver has implemented this opcode yet (and no state tracker
2480 emits it). This information is subject to change.
2482 .. _resourceopcodes:
2484 Resource Access Opcodes
2485 ^^^^^^^^^^^^^^^^^^^^^^^
2487 For these opcodes, the resource can be a BUFFER, IMAGE, or MEMORY.
2489 .. opcode:: LOAD - Fetch data from a shader buffer or image
2491 Syntax: ``LOAD dst, resource, address``
2493 Example: ``LOAD TEMP[0], BUFFER[0], TEMP[1]``
2495 Using the provided integer address, LOAD fetches data
2496 from the specified buffer or texture without any
2499 The 'address' is specified as a vector of unsigned
2500 integers. If the 'address' is out of range the result
2503 Only the first mipmap level of a resource can be read
2504 from using this instruction.
2506 For 1D or 2D texture arrays, the array index is
2507 provided as an unsigned integer in address.y or
2508 address.z, respectively. address.yz are ignored for
2509 buffers and 1D textures. address.z is ignored for 1D
2510 texture arrays and 2D textures. address.w is always
2513 A swizzle suffix may be added to the resource argument
2514 this will cause the resource data to be swizzled accordingly.
2516 .. opcode:: STORE - Write data to a shader resource
2518 Syntax: ``STORE resource, address, src``
2520 Example: ``STORE BUFFER[0], TEMP[0], TEMP[1]``
2522 Using the provided integer address, STORE writes data
2523 to the specified buffer or texture.
2525 The 'address' is specified as a vector of unsigned
2526 integers. If the 'address' is out of range the result
2529 Only the first mipmap level of a resource can be
2530 written to using this instruction.
2532 For 1D or 2D texture arrays, the array index is
2533 provided as an unsigned integer in address.y or
2534 address.z, respectively. address.yz are ignored for
2535 buffers and 1D textures. address.z is ignored for 1D
2536 texture arrays and 2D textures. address.w is always
2539 .. opcode:: RESQ - Query information about a resource
2541 Syntax: ``RESQ dst, resource``
2543 Example: ``RESQ TEMP[0], BUFFER[0]``
2545 Returns information about the buffer or image resource. For buffer
2546 resources, the size (in bytes) is returned in the x component. For
2547 image resources, .xyz will contain the width/height/layers of the
2548 image, while .w will contain the number of samples for multi-sampled
2551 .. opcode:: FBFETCH - Load data from framebuffer
2553 Syntax: ``FBFETCH dst, output``
2555 Example: ``FBFETCH TEMP[0], OUT[0]``
2557 This is only valid on ``COLOR`` semantic outputs. Returns the color
2558 of the current position in the framebuffer from before this fragment
2559 shader invocation. May return the same value from multiple calls for
2560 a particular output within a single invocation. Note that result may
2561 be undefined if a fragment is drawn multiple times without a blend
2565 .. _threadsyncopcodes:
2567 Inter-thread synchronization opcodes
2568 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2570 These opcodes are intended for communication between threads running
2571 within the same compute grid. For now they're only valid in compute
2574 .. opcode:: BARRIER - Thread group barrier
2578 This opcode suspends the execution of the current thread until all
2579 the remaining threads in the working group reach the same point of
2580 the program. Results are unspecified if any of the remaining
2581 threads terminates or never reaches an executed BARRIER instruction.
2583 .. opcode:: MEMBAR - Memory barrier
2587 This opcode waits for the completion of all memory accesses based on
2588 the type passed in. The type is an immediate bitfield with the following
2591 Bit 0: Shader storage buffers
2592 Bit 1: Atomic buffers
2594 Bit 3: Shared memory
2597 These may be passed in in any combination. An implementation is free to not
2598 distinguish between these as it sees fit. However these map to all the
2599 possibilities made available by GLSL.
2606 These opcodes provide atomic variants of some common arithmetic and
2607 logical operations. In this context atomicity means that another
2608 concurrent memory access operation that affects the same memory
2609 location is guaranteed to be performed strictly before or after the
2610 entire execution of the atomic operation. The resource may be a BUFFER,
2611 IMAGE, or MEMORY. In the case of an image, the offset works the same as for
2612 ``LOAD`` and ``STORE``, specified above. These atomic operations may
2613 only be used with 32-bit integer image formats.
2615 .. opcode:: ATOMUADD - Atomic integer addition
2617 Syntax: ``ATOMUADD dst, resource, offset, src``
2619 Example: ``ATOMUADD TEMP[0], BUFFER[0], TEMP[1], TEMP[2]``
2621 The following operation is performed atomically:
2625 dst_x = resource[offset]
2627 resource[offset] = dst_x + src_x
2630 .. opcode:: ATOMXCHG - Atomic exchange
2632 Syntax: ``ATOMXCHG dst, resource, offset, src``
2634 Example: ``ATOMXCHG TEMP[0], BUFFER[0], TEMP[1], TEMP[2]``
2636 The following operation is performed atomically:
2640 dst_x = resource[offset]
2642 resource[offset] = src_x
2645 .. opcode:: ATOMCAS - Atomic compare-and-exchange
2647 Syntax: ``ATOMCAS dst, resource, offset, cmp, src``
2649 Example: ``ATOMCAS TEMP[0], BUFFER[0], TEMP[1], TEMP[2], TEMP[3]``
2651 The following operation is performed atomically:
2655 dst_x = resource[offset]
2657 resource[offset] = (dst_x == cmp_x ? src_x : dst_x)
2660 .. opcode:: ATOMAND - Atomic bitwise And
2662 Syntax: ``ATOMAND dst, resource, offset, src``
2664 Example: ``ATOMAND TEMP[0], BUFFER[0], TEMP[1], TEMP[2]``
2666 The following operation is performed atomically:
2670 dst_x = resource[offset]
2672 resource[offset] = dst_x \& src_x
2675 .. opcode:: ATOMOR - Atomic bitwise Or
2677 Syntax: ``ATOMOR dst, resource, offset, src``
2679 Example: ``ATOMOR TEMP[0], BUFFER[0], TEMP[1], TEMP[2]``
2681 The following operation is performed atomically:
2685 dst_x = resource[offset]
2687 resource[offset] = dst_x | src_x
2690 .. opcode:: ATOMXOR - Atomic bitwise Xor
2692 Syntax: ``ATOMXOR dst, resource, offset, src``
2694 Example: ``ATOMXOR TEMP[0], BUFFER[0], TEMP[1], TEMP[2]``
2696 The following operation is performed atomically:
2700 dst_x = resource[offset]
2702 resource[offset] = dst_x \oplus src_x
2705 .. opcode:: ATOMUMIN - Atomic unsigned minimum
2707 Syntax: ``ATOMUMIN dst, resource, offset, src``
2709 Example: ``ATOMUMIN TEMP[0], BUFFER[0], TEMP[1], TEMP[2]``
2711 The following operation is performed atomically:
2715 dst_x = resource[offset]
2717 resource[offset] = (dst_x < src_x ? dst_x : src_x)
2720 .. opcode:: ATOMUMAX - Atomic unsigned maximum
2722 Syntax: ``ATOMUMAX dst, resource, offset, src``
2724 Example: ``ATOMUMAX TEMP[0], BUFFER[0], TEMP[1], TEMP[2]``
2726 The following operation is performed atomically:
2730 dst_x = resource[offset]
2732 resource[offset] = (dst_x > src_x ? dst_x : src_x)
2735 .. opcode:: ATOMIMIN - Atomic signed minimum
2737 Syntax: ``ATOMIMIN dst, resource, offset, src``
2739 Example: ``ATOMIMIN TEMP[0], BUFFER[0], TEMP[1], TEMP[2]``
2741 The following operation is performed atomically:
2745 dst_x = resource[offset]
2747 resource[offset] = (dst_x < src_x ? dst_x : src_x)
2750 .. opcode:: ATOMIMAX - Atomic signed maximum
2752 Syntax: ``ATOMIMAX dst, resource, offset, src``
2754 Example: ``ATOMIMAX TEMP[0], BUFFER[0], TEMP[1], TEMP[2]``
2756 The following operation is performed atomically:
2760 dst_x = resource[offset]
2762 resource[offset] = (dst_x > src_x ? dst_x : src_x)
2765 .. _interlaneopcodes:
2770 These opcodes reduce the given value across the shader invocations
2771 running in the current SIMD group. Every thread in the subgroup will receive
2772 the same result. The BALLOT operations accept a single-channel argument that
2773 is treated as a boolean and produce a 64-bit value.
2775 .. opcode:: VOTE_ANY - Value is set in any of the active invocations
2777 Syntax: ``VOTE_ANY dst, value``
2779 Example: ``VOTE_ANY TEMP[0].x, TEMP[1].x``
2782 .. opcode:: VOTE_ALL - Value is set in all of the active invocations
2784 Syntax: ``VOTE_ALL dst, value``
2786 Example: ``VOTE_ALL TEMP[0].x, TEMP[1].x``
2789 .. opcode:: VOTE_EQ - Value is the same in all of the active invocations
2791 Syntax: ``VOTE_EQ dst, value``
2793 Example: ``VOTE_EQ TEMP[0].x, TEMP[1].x``
2796 .. opcode:: BALLOT - Lanemask of whether the value is set in each active
2799 Syntax: ``BALLOT dst, value``
2801 Example: ``BALLOT TEMP[0].xy, TEMP[1].x``
2803 When the argument is a constant true, this produces a bitmask of active
2804 invocations. In fragment shaders, this can include helper invocations
2805 (invocations whose outputs and writes to memory are discarded, but which
2806 are used to compute derivatives).
2809 .. opcode:: READ_FIRST - Broadcast the value from the first active
2810 invocation to all active lanes
2812 Syntax: ``READ_FIRST dst, value``
2814 Example: ``READ_FIRST TEMP[0], TEMP[1]``
2817 .. opcode:: READ_INVOC - Retrieve the value from the given invocation
2818 (need not be uniform)
2820 Syntax: ``READ_INVOC dst, value, invocation``
2822 Example: ``READ_INVOC TEMP[0].xy, TEMP[1].xy, TEMP[2].x``
2824 invocation.x controls the invocation number to read from for all channels.
2825 The invocation number must be the same across all active invocations in a
2826 sub-group; otherwise, the results are undefined.
2829 Explanation of symbols used
2830 ------------------------------
2837 :math:`|x|` Absolute value of `x`.
2839 :math:`\lceil x \rceil` Ceiling of `x`.
2841 clamp(x,y,z) Clamp x between y and z.
2842 (x < y) ? y : (x > z) ? z : x
2844 :math:`\lfloor x\rfloor` Floor of `x`.
2846 :math:`\log_2{x}` Logarithm of `x`, base 2.
2848 max(x,y) Maximum of x and y.
2851 min(x,y) Minimum of x and y.
2854 partialx(x) Derivative of x relative to fragment's X.
2856 partialy(x) Derivative of x relative to fragment's Y.
2858 pop() Pop from stack.
2860 :math:`x^y` `x` to the power `y`.
2862 push(x) Push x on stack.
2866 trunc(x) Truncate x, i.e. drop the fraction bits.
2873 discard Discard fragment.
2877 target Label of target instruction.
2888 Declares a register that is will be referenced as an operand in Instruction
2891 File field contains register file that is being declared and is one
2894 UsageMask field specifies which of the register components can be accessed
2895 and is one of TGSI_WRITEMASK.
2897 The Local flag specifies that a given value isn't intended for
2898 subroutine parameter passing and, as a result, the implementation
2899 isn't required to give any guarantees of it being preserved across
2900 subroutine boundaries. As it's merely a compiler hint, the
2901 implementation is free to ignore it.
2903 If Dimension flag is set to 1, a Declaration Dimension token follows.
2905 If Semantic flag is set to 1, a Declaration Semantic token follows.
2907 If Interpolate flag is set to 1, a Declaration Interpolate token follows.
2909 If file is TGSI_FILE_RESOURCE, a Declaration Resource token follows.
2911 If Array flag is set to 1, a Declaration Array token follows.
2914 ^^^^^^^^^^^^^^^^^^^^^^^^
2916 Declarations can optional have an ArrayID attribute which can be referred by
2917 indirect addressing operands. An ArrayID of zero is reserved and treated as
2918 if no ArrayID is specified.
2920 If an indirect addressing operand refers to a specific declaration by using
2921 an ArrayID only the registers in this declaration are guaranteed to be
2922 accessed, accessing any register outside this declaration results in undefined
2923 behavior. Note that for compatibility the effective index is zero-based and
2924 not relative to the specified declaration
2926 If no ArrayID is specified with an indirect addressing operand the whole
2927 register file might be accessed by this operand. This is strongly discouraged
2928 and will prevent packing of scalar/vec2 arrays and effective alias analysis.
2929 This is only legal for TEMP and CONST register files.
2931 Declaration Semantic
2932 ^^^^^^^^^^^^^^^^^^^^^^^^
2934 Vertex and fragment shader input and output registers may be labeled
2935 with semantic information consisting of a name and index.
2937 Follows Declaration token if Semantic bit is set.
2939 Since its purpose is to link a shader with other stages of the pipeline,
2940 it is valid to follow only those Declaration tokens that declare a register
2941 either in INPUT or OUTPUT file.
2943 SemanticName field contains the semantic name of the register being declared.
2944 There is no default value.
2946 SemanticIndex is an optional subscript that can be used to distinguish
2947 different register declarations with the same semantic name. The default value
2950 The meanings of the individual semantic names are explained in the following
2953 TGSI_SEMANTIC_POSITION
2954 """"""""""""""""""""""
2956 For vertex shaders, TGSI_SEMANTIC_POSITION indicates the vertex shader
2957 output register which contains the homogeneous vertex position in the clip
2958 space coordinate system. After clipping, the X, Y and Z components of the
2959 vertex will be divided by the W value to get normalized device coordinates.
2961 For fragment shaders, TGSI_SEMANTIC_POSITION is used to indicate that
2962 fragment shader input (or system value, depending on which one is
2963 supported by the driver) contains the fragment's window position. The X
2964 component starts at zero and always increases from left to right.
2965 The Y component starts at zero and always increases but Y=0 may either
2966 indicate the top of the window or the bottom depending on the fragment
2967 coordinate origin convention (see TGSI_PROPERTY_FS_COORD_ORIGIN).
2968 The Z coordinate ranges from 0 to 1 to represent depth from the front
2969 to the back of the Z buffer. The W component contains the interpolated
2970 reciprocal of the vertex position W component (corresponding to gl_Fragcoord,
2971 but unlike d3d10 which interpolates the same 1/w but then gives back
2972 the reciprocal of the interpolated value).
2974 Fragment shaders may also declare an output register with
2975 TGSI_SEMANTIC_POSITION. Only the Z component is writable. This allows
2976 the fragment shader to change the fragment's Z position.
2983 For vertex shader outputs or fragment shader inputs/outputs, this
2984 label indicates that the register contains an R,G,B,A color.
2986 Several shader inputs/outputs may contain colors so the semantic index
2987 is used to distinguish them. For example, color[0] may be the diffuse
2988 color while color[1] may be the specular color.
2990 This label is needed so that the flat/smooth shading can be applied
2991 to the right interpolants during rasterization.
2995 TGSI_SEMANTIC_BCOLOR
2996 """"""""""""""""""""
2998 Back-facing colors are only used for back-facing polygons, and are only valid
2999 in vertex shader outputs. After rasterization, all polygons are front-facing
3000 and COLOR and BCOLOR end up occupying the same slots in the fragment shader,
3001 so all BCOLORs effectively become regular COLORs in the fragment shader.
3007 Vertex shader inputs and outputs and fragment shader inputs may be
3008 labeled with TGSI_SEMANTIC_FOG to indicate that the register contains
3009 a fog coordinate. Typically, the fragment shader will use the fog coordinate
3010 to compute a fog blend factor which is used to blend the normal fragment color
3011 with a constant fog color. But fog coord really is just an ordinary vec4
3012 register like regular semantics.
3018 Vertex shader input and output registers may be labeled with
3019 TGIS_SEMANTIC_PSIZE to indicate that the register contains a point size
3020 in the form (S, 0, 0, 1). The point size controls the width or diameter
3021 of points for rasterization. This label cannot be used in fragment
3024 When using this semantic, be sure to set the appropriate state in the
3025 :ref:`rasterizer` first.
3028 TGSI_SEMANTIC_TEXCOORD
3029 """"""""""""""""""""""
3031 Only available if PIPE_CAP_TGSI_TEXCOORD is exposed !
3033 Vertex shader outputs and fragment shader inputs may be labeled with
3034 this semantic to make them replaceable by sprite coordinates via the
3035 sprite_coord_enable state in the :ref:`rasterizer`.
3036 The semantic index permitted with this semantic is limited to <= 7.
3038 If the driver does not support TEXCOORD, sprite coordinate replacement
3039 applies to inputs with the GENERIC semantic instead.
3041 The intended use case for this semantic is gl_TexCoord.
3044 TGSI_SEMANTIC_PCOORD
3045 """"""""""""""""""""
3047 Only available if PIPE_CAP_TGSI_TEXCOORD is exposed !
3049 Fragment shader inputs may be labeled with TGSI_SEMANTIC_PCOORD to indicate
3050 that the register contains sprite coordinates in the form (x, y, 0, 1), if
3051 the current primitive is a point and point sprites are enabled. Otherwise,
3052 the contents of the register are undefined.
3054 The intended use case for this semantic is gl_PointCoord.
3057 TGSI_SEMANTIC_GENERIC
3058 """""""""""""""""""""
3060 All vertex/fragment shader inputs/outputs not labeled with any other
3061 semantic label can be considered to be generic attributes. Typical
3062 uses of generic inputs/outputs are texcoords and user-defined values.
3065 TGSI_SEMANTIC_NORMAL
3066 """"""""""""""""""""
3068 Indicates that a vertex shader input is a normal vector. This is
3069 typically only used for legacy graphics APIs.
3075 This label applies to fragment shader inputs (or system values,
3076 depending on which one is supported by the driver) and indicates that
3077 the register contains front/back-face information.
3079 If it is an input, it will be a floating-point vector in the form (F, 0, 0, 1),
3080 where F will be positive when the fragment belongs to a front-facing polygon,
3081 and negative when the fragment belongs to a back-facing polygon.
3083 If it is a system value, it will be an integer vector in the form (F, 0, 0, 1),
3084 where F is 0xffffffff when the fragment belongs to a front-facing polygon and
3085 0 when the fragment belongs to a back-facing polygon.
3088 TGSI_SEMANTIC_EDGEFLAG
3089 """"""""""""""""""""""
3091 For vertex shaders, this sematic label indicates that an input or
3092 output is a boolean edge flag. The register layout is [F, x, x, x]
3093 where F is 0.0 or 1.0 and x = don't care. Normally, the vertex shader
3094 simply copies the edge flag input to the edgeflag output.
3096 Edge flags are used to control which lines or points are actually
3097 drawn when the polygon mode converts triangles/quads/polygons into
3101 TGSI_SEMANTIC_STENCIL
3102 """""""""""""""""""""
3104 For fragment shaders, this semantic label indicates that an output
3105 is a writable stencil reference value. Only the Y component is writable.
3106 This allows the fragment shader to change the fragments stencilref value.
3109 TGSI_SEMANTIC_VIEWPORT_INDEX
3110 """"""""""""""""""""""""""""
3112 For geometry shaders, this semantic label indicates that an output
3113 contains the index of the viewport (and scissor) to use.
3114 This is an integer value, and only the X component is used.
3116 If PIPE_CAP_TGSI_VS_LAYER_VIEWPORT or PIPE_CAP_TGSI_TES_LAYER_VIEWPORT is
3117 supported, then this semantic label can also be used in vertex or
3118 tessellation evaluation shaders, respectively. Only the value written in the
3119 last vertex processing stage is used.
3125 For geometry shaders, this semantic label indicates that an output
3126 contains the layer value to use for the color and depth/stencil surfaces.
3127 This is an integer value, and only the X component is used.
3128 (Also known as rendertarget array index.)
3130 If PIPE_CAP_TGSI_VS_LAYER_VIEWPORT or PIPE_CAP_TGSI_TES_LAYER_VIEWPORT is
3131 supported, then this semantic label can also be used in vertex or
3132 tessellation evaluation shaders, respectively. Only the value written in the
3133 last vertex processing stage is used.
3136 TGSI_SEMANTIC_CULLDIST
3137 """"""""""""""""""""""
3139 Used as distance to plane for performing application-defined culling
3140 of individual primitives against a plane. When components of vertex
3141 elements are given this label, these values are assumed to be a
3142 float32 signed distance to a plane. Primitives will be completely
3143 discarded if the plane distance for all of the vertices in the
3144 primitive are < 0. If a vertex has a cull distance of NaN, that
3145 vertex counts as "out" (as if its < 0);
3146 The limits on both clip and cull distances are bound
3147 by the PIPE_MAX_CLIP_OR_CULL_DISTANCE_COUNT define which defines
3148 the maximum number of components that can be used to hold the
3149 distances and by the PIPE_MAX_CLIP_OR_CULL_DISTANCE_ELEMENT_COUNT
3150 which specifies the maximum number of registers which can be
3151 annotated with those semantics.
3154 TGSI_SEMANTIC_CLIPDIST
3155 """"""""""""""""""""""
3157 Note this covers clipping and culling distances.
3159 When components of vertex elements are identified this way, these
3160 values are each assumed to be a float32 signed distance to a plane.
3163 Primitive setup only invokes rasterization on pixels for which
3164 the interpolated plane distances are >= 0.
3167 Primitives will be completely discarded if the plane distance
3168 for all of the vertices in the primitive are < 0.
3169 If a vertex has a cull distance of NaN, that vertex counts as "out"
3172 Multiple clip/cull planes can be implemented simultaneously, by
3173 annotating multiple components of one or more vertex elements with
3174 the above specified semantic.
3175 The limits on both clip and cull distances are bound
3176 by the PIPE_MAX_CLIP_OR_CULL_DISTANCE_COUNT define which defines
3177 the maximum number of components that can be used to hold the
3178 distances and by the PIPE_MAX_CLIP_OR_CULL_DISTANCE_ELEMENT_COUNT
3179 which specifies the maximum number of registers which can be
3180 annotated with those semantics.
3181 The properties NUM_CLIPDIST_ENABLED and NUM_CULLDIST_ENABLED
3182 are used to divide up the 2 x vec4 space between clipping and culling.
3184 TGSI_SEMANTIC_SAMPLEID
3185 """"""""""""""""""""""
3187 For fragment shaders, this semantic label indicates that a system value
3188 contains the current sample id (i.e. gl_SampleID) as an unsigned int.
3189 Only the X component is used. If per-sample shading is not enabled,
3190 the result is (0, undef, undef, undef).
3192 Note that if the fragment shader uses this system value, the fragment
3193 shader is automatically executed at per sample frequency.
3195 TGSI_SEMANTIC_SAMPLEPOS
3196 """""""""""""""""""""""
3198 For fragment shaders, this semantic label indicates that a system
3199 value contains the current sample's position as float4(x, y, undef, undef)
3200 in the render target (i.e. gl_SamplePosition) when per-fragment shading
3201 is in effect. Position values are in the range [0, 1] where 0.5 is
3202 the center of the fragment.
3204 Note that if the fragment shader uses this system value, the fragment
3205 shader is automatically executed at per sample frequency.
3207 TGSI_SEMANTIC_SAMPLEMASK
3208 """"""""""""""""""""""""
3210 For fragment shaders, this semantic label can be applied to either a
3211 shader system value input or output.
3213 For a system value, the sample mask indicates the set of samples covered by
3214 the current primitive. If MSAA is not enabled, the value is (1, 0, 0, 0).
3216 For an output, the sample mask is used to disable further sample processing.
3218 For both, the register type is uint[4] but only the X component is used
3219 (i.e. gl_SampleMask[0]). Each bit corresponds to one sample position (up
3220 to 32x MSAA is supported).
3222 TGSI_SEMANTIC_INVOCATIONID
3223 """"""""""""""""""""""""""
3225 For geometry shaders, this semantic label indicates that a system value
3226 contains the current invocation id (i.e. gl_InvocationID).
3227 This is an integer value, and only the X component is used.
3229 TGSI_SEMANTIC_INSTANCEID
3230 """"""""""""""""""""""""
3232 For vertex shaders, this semantic label indicates that a system value contains
3233 the current instance id (i.e. gl_InstanceID). It does not include the base
3234 instance. This is an integer value, and only the X component is used.
3236 TGSI_SEMANTIC_VERTEXID
3237 """"""""""""""""""""""
3239 For vertex shaders, this semantic label indicates that a system value contains
3240 the current vertex id (i.e. gl_VertexID). It does (unlike in d3d10) include the
3241 base vertex. This is an integer value, and only the X component is used.
3243 TGSI_SEMANTIC_VERTEXID_NOBASE
3244 """""""""""""""""""""""""""""""
3246 For vertex shaders, this semantic label indicates that a system value contains
3247 the current vertex id without including the base vertex (this corresponds to
3248 d3d10 vertex id, so TGSI_SEMANTIC_VERTEXID_NOBASE + TGSI_SEMANTIC_BASEVERTEX
3249 == TGSI_SEMANTIC_VERTEXID). This is an integer value, and only the X component
3252 TGSI_SEMANTIC_BASEVERTEX
3253 """"""""""""""""""""""""
3255 For vertex shaders, this semantic label indicates that a system value contains
3256 the base vertex (i.e. gl_BaseVertex). Note that for non-indexed draw calls,
3257 this contains the first (or start) value instead.
3258 This is an integer value, and only the X component is used.
3260 TGSI_SEMANTIC_PRIMID
3261 """"""""""""""""""""
3263 For geometry and fragment shaders, this semantic label indicates the value
3264 contains the primitive id (i.e. gl_PrimitiveID). This is an integer value,
3265 and only the X component is used.
3266 FIXME: This right now can be either a ordinary input or a system value...
3272 For tessellation evaluation/control shaders, this semantic label indicates a
3273 generic per-patch attribute. Such semantics will not implicitly be per-vertex
3276 TGSI_SEMANTIC_TESSCOORD
3277 """""""""""""""""""""""
3279 For tessellation evaluation shaders, this semantic label indicates the
3280 coordinates of the vertex being processed. This is available in XYZ; W is
3283 TGSI_SEMANTIC_TESSOUTER
3284 """""""""""""""""""""""
3286 For tessellation evaluation/control shaders, this semantic label indicates the
3287 outer tessellation levels of the patch. Isoline tessellation will only have XY
3288 defined, triangle will have XYZ and quads will have XYZW defined. This
3289 corresponds to gl_TessLevelOuter.
3291 TGSI_SEMANTIC_TESSINNER
3292 """""""""""""""""""""""
3294 For tessellation evaluation/control shaders, this semantic label indicates the
3295 inner tessellation levels of the patch. The X value is only defined for
3296 triangle tessellation, while quads will have XY defined. This is entirely
3297 undefined for isoline tessellation.
3299 TGSI_SEMANTIC_VERTICESIN
3300 """"""""""""""""""""""""
3302 For tessellation evaluation/control shaders, this semantic label indicates the
3303 number of vertices provided in the input patch. Only the X value is defined.
3305 TGSI_SEMANTIC_HELPER_INVOCATION
3306 """""""""""""""""""""""""""""""
3308 For fragment shaders, this semantic indicates whether the current
3309 invocation is covered or not. Helper invocations are created in order
3310 to properly compute derivatives, however it may be desirable to skip
3311 some of the logic in those cases. See ``gl_HelperInvocation`` documentation.
3313 TGSI_SEMANTIC_BASEINSTANCE
3314 """"""""""""""""""""""""""
3316 For vertex shaders, the base instance argument supplied for this
3317 draw. This is an integer value, and only the X component is used.
3319 TGSI_SEMANTIC_DRAWID
3320 """"""""""""""""""""
3322 For vertex shaders, the zero-based index of the current draw in a
3323 ``glMultiDraw*`` invocation. This is an integer value, and only the X
3327 TGSI_SEMANTIC_WORK_DIM
3328 """"""""""""""""""""""
3330 For compute shaders started via opencl this retrieves the work_dim
3331 parameter to the clEnqueueNDRangeKernel call with which the shader
3335 TGSI_SEMANTIC_GRID_SIZE
3336 """""""""""""""""""""""
3338 For compute shaders, this semantic indicates the maximum (x, y, z) dimensions
3339 of a grid of thread blocks.
3342 TGSI_SEMANTIC_BLOCK_ID
3343 """"""""""""""""""""""
3345 For compute shaders, this semantic indicates the (x, y, z) coordinates of the
3346 current block inside of the grid.
3349 TGSI_SEMANTIC_BLOCK_SIZE
3350 """"""""""""""""""""""""
3352 For compute shaders, this semantic indicates the maximum (x, y, z) dimensions
3353 of a block in threads.
3356 TGSI_SEMANTIC_THREAD_ID
3357 """""""""""""""""""""""
3359 For compute shaders, this semantic indicates the (x, y, z) coordinates of the
3360 current thread inside of the block.
3363 TGSI_SEMANTIC_SUBGROUP_SIZE
3364 """""""""""""""""""""""""""
3366 This semantic indicates the subgroup size for the current invocation. This is
3367 an integer of at most 64, as it indicates the width of lanemasks. It does not
3368 depend on the number of invocations that are active.
3371 TGSI_SEMANTIC_SUBGROUP_INVOCATION
3372 """""""""""""""""""""""""""""""""
3374 The index of the current invocation within its subgroup.
3377 TGSI_SEMANTIC_SUBGROUP_EQ_MASK
3378 """"""""""""""""""""""""""""""
3380 A bit mask of ``bit index == TGSI_SEMANTIC_SUBGROUP_INVOCATION``, i.e.
3381 ``1 << subgroup_invocation`` in arbitrary precision arithmetic.
3384 TGSI_SEMANTIC_SUBGROUP_GE_MASK
3385 """"""""""""""""""""""""""""""
3387 A bit mask of ``bit index >= TGSI_SEMANTIC_SUBGROUP_INVOCATION``, i.e.
3388 ``((1 << (subgroup_size - subgroup_invocation)) - 1) << subgroup_invocation``
3389 in arbitrary precision arithmetic.
3392 TGSI_SEMANTIC_SUBGROUP_GT_MASK
3393 """"""""""""""""""""""""""""""
3395 A bit mask of ``bit index > TGSI_SEMANTIC_SUBGROUP_INVOCATION``, i.e.
3396 ``((1 << (subgroup_size - subgroup_invocation - 1)) - 1) << (subgroup_invocation + 1)``
3397 in arbitrary precision arithmetic.
3400 TGSI_SEMANTIC_SUBGROUP_LE_MASK
3401 """"""""""""""""""""""""""""""
3403 A bit mask of ``bit index <= TGSI_SEMANTIC_SUBGROUP_INVOCATION``, i.e.
3404 ``(1 << (subgroup_invocation + 1)) - 1`` in arbitrary precision arithmetic.
3407 TGSI_SEMANTIC_SUBGROUP_LT_MASK
3408 """"""""""""""""""""""""""""""
3410 A bit mask of ``bit index < TGSI_SEMANTIC_SUBGROUP_INVOCATION``, i.e.
3411 ``(1 << subgroup_invocation) - 1`` in arbitrary precision arithmetic.
3414 Declaration Interpolate
3415 ^^^^^^^^^^^^^^^^^^^^^^^
3417 This token is only valid for fragment shader INPUT declarations.
3419 The Interpolate field specifes the way input is being interpolated by
3420 the rasteriser and is one of TGSI_INTERPOLATE_*.
3422 The Location field specifies the location inside the pixel that the
3423 interpolation should be done at, one of ``TGSI_INTERPOLATE_LOC_*``. Note that
3424 when per-sample shading is enabled, the implementation may choose to
3425 interpolate at the sample irrespective of the Location field.
3427 The CylindricalWrap bitfield specifies which register components
3428 should be subject to cylindrical wrapping when interpolating by the
3429 rasteriser. If TGSI_CYLINDRICAL_WRAP_X is set to 1, the X component
3430 should be interpolated according to cylindrical wrapping rules.
3433 Declaration Sampler View
3434 ^^^^^^^^^^^^^^^^^^^^^^^^
3436 Follows Declaration token if file is TGSI_FILE_SAMPLER_VIEW.
3438 DCL SVIEW[#], resource, type(s)
3440 Declares a shader input sampler view and assigns it to a SVIEW[#]
3443 resource can be one of BUFFER, 1D, 2D, 3D, 1DArray and 2DArray.
3445 type must be 1 or 4 entries (if specifying on a per-component
3446 level) out of UNORM, SNORM, SINT, UINT and FLOAT.
3448 For TEX\* style texture sample opcodes (as opposed to SAMPLE\* opcodes
3449 which take an explicit SVIEW[#] source register), there may be optionally
3450 SVIEW[#] declarations. In this case, the SVIEW index is implied by the
3451 SAMP index, and there must be a corresponding SVIEW[#] declaration for
3452 each SAMP[#] declaration. Drivers are free to ignore this if they wish.
3453 But note in particular that some drivers need to know the sampler type
3454 (float/int/unsigned) in order to generate the correct code, so cases
3455 where integer textures are sampled, SVIEW[#] declarations should be
3458 NOTE: It is NOT legal to mix SAMPLE\* style opcodes and TEX\* opcodes
3461 Declaration Resource
3462 ^^^^^^^^^^^^^^^^^^^^
3464 Follows Declaration token if file is TGSI_FILE_RESOURCE.
3466 DCL RES[#], resource [, WR] [, RAW]
3468 Declares a shader input resource and assigns it to a RES[#]
3471 resource can be one of BUFFER, 1D, 2D, 3D, CUBE, 1DArray and
3474 If the RAW keyword is not specified, the texture data will be
3475 subject to conversion, swizzling and scaling as required to yield
3476 the specified data type from the physical data format of the bound
3479 If the RAW keyword is specified, no channel conversion will be
3480 performed: the values read for each of the channels (X,Y,Z,W) will
3481 correspond to consecutive words in the same order and format
3482 they're found in memory. No element-to-address conversion will be
3483 performed either: the value of the provided X coordinate will be
3484 interpreted in byte units instead of texel units. The result of
3485 accessing a misaligned address is undefined.
3487 Usage of the STORE opcode is only allowed if the WR (writable) flag
3492 ^^^^^^^^^^^^^^^^^^^^^^^^
3494 Properties are general directives that apply to the whole TGSI program.
3499 Specifies the fragment shader TGSI_SEMANTIC_POSITION coordinate origin.
3500 The default value is UPPER_LEFT.
3502 If UPPER_LEFT, the position will be (0,0) at the upper left corner and
3503 increase downward and rightward.
3504 If LOWER_LEFT, the position will be (0,0) at the lower left corner and
3505 increase upward and rightward.
3507 OpenGL defaults to LOWER_LEFT, and is configurable with the
3508 GL_ARB_fragment_coord_conventions extension.
3510 DirectX 9/10 use UPPER_LEFT.
3512 FS_COORD_PIXEL_CENTER
3513 """""""""""""""""""""
3515 Specifies the fragment shader TGSI_SEMANTIC_POSITION pixel center convention.
3516 The default value is HALF_INTEGER.
3518 If HALF_INTEGER, the fractionary part of the position will be 0.5
3519 If INTEGER, the fractionary part of the position will be 0.0
3521 Note that this does not affect the set of fragments generated by
3522 rasterization, which is instead controlled by half_pixel_center in the
3525 OpenGL defaults to HALF_INTEGER, and is configurable with the
3526 GL_ARB_fragment_coord_conventions extension.
3528 DirectX 9 uses INTEGER.
3529 DirectX 10 uses HALF_INTEGER.
3531 FS_COLOR0_WRITES_ALL_CBUFS
3532 """"""""""""""""""""""""""
3533 Specifies that writes to the fragment shader color 0 are replicated to all
3534 bound cbufs. This facilitates OpenGL's fragColor output vs fragData[0] where
3535 fragData is directed to a single color buffer, but fragColor is broadcast.
3538 """"""""""""""""""""""""""
3539 If this property is set on the program bound to the shader stage before the
3540 fragment shader, user clip planes should have no effect (be disabled) even if
3541 that shader does not write to any clip distance outputs and the rasterizer's
3542 clip_plane_enable is non-zero.
3543 This property is only supported by drivers that also support shader clip
3545 This is useful for APIs that don't have UCPs and where clip distances written
3546 by a shader cannot be disabled.
3551 Specifies the number of times a geometry shader should be executed for each
3552 input primitive. Each invocation will have a different
3553 TGSI_SEMANTIC_INVOCATIONID system value set. If not specified, assumed to
3556 VS_WINDOW_SPACE_POSITION
3557 """"""""""""""""""""""""""
3558 If this property is set on the vertex shader, the TGSI_SEMANTIC_POSITION output
3559 is assumed to contain window space coordinates.
3560 Division of X,Y,Z by W and the viewport transformation are disabled, and 1/W is
3561 directly taken from the 4-th component of the shader output.
3562 Naturally, clipping is not performed on window coordinates either.
3563 The effect of this property is undefined if a geometry or tessellation shader
3569 The number of vertices written by the tessellation control shader. This
3570 effectively defines the patch input size of the tessellation evaluation shader
3576 This sets the tessellation primitive mode, one of ``PIPE_PRIM_TRIANGLES``,
3577 ``PIPE_PRIM_QUADS``, or ``PIPE_PRIM_LINES``. (Unlike in GL, there is no
3578 separate isolines settings, the regular lines is assumed to mean isolines.)
3583 This sets the spacing mode of the tessellation generator, one of
3584 ``PIPE_TESS_SPACING_*``.
3589 This sets the vertex order to be clockwise if the value is 1, or
3590 counter-clockwise if set to 0.
3595 If set to a non-zero value, this turns on point mode for the tessellator,
3596 which means that points will be generated instead of primitives.
3598 NUM_CLIPDIST_ENABLED
3599 """"""""""""""""""""
3601 How many clip distance scalar outputs are enabled.
3603 NUM_CULLDIST_ENABLED
3604 """"""""""""""""""""
3606 How many cull distance scalar outputs are enabled.
3608 FS_EARLY_DEPTH_STENCIL
3609 """"""""""""""""""""""
3611 Whether depth test, stencil test, and occlusion query should run before
3612 the fragment shader (regardless of fragment shader side effects). Corresponds
3613 to GLSL early_fragment_tests.
3618 Which shader stage will MOST LIKELY follow after this shader when the shader
3619 is bound. This is only a hint to the driver and doesn't have to be precise.
3620 Only set for VS and TES.
3622 CS_FIXED_BLOCK_WIDTH / HEIGHT / DEPTH
3623 """""""""""""""""""""""""""""""""""""
3625 Threads per block in each dimension, if known at compile time. If the block size
3626 is known all three should be at least 1. If it is unknown they should all be set
3632 The MUL TGSI operation (FP32 multiplication) will return 0 if either
3633 of the operands are equal to 0. That means that 0 * Inf = 0. This
3634 should be set the same way for an entire pipeline. Note that this
3635 applies not only to the literal MUL TGSI opcode, but all FP32
3636 multiplications implied by other operations, such as MAD, FMA, DP2,
3637 DP3, DP4, DST, LOG, LRP, and possibly others. If there is a
3638 mismatch between shaders, then it is unspecified whether this behavior
3641 FS_POST_DEPTH_COVERAGE
3642 """"""""""""""""""""""
3644 When enabled, the input for TGSI_SEMANTIC_SAMPLEMASK will exclude samples
3645 that have failed the depth/stencil tests. This is only valid when
3646 FS_EARLY_DEPTH_STENCIL is also specified.
3649 Texture Sampling and Texture Formats
3650 ------------------------------------
3652 This table shows how texture image components are returned as (x,y,z,w) tuples
3653 by TGSI texture instructions, such as :opcode:`TEX`, :opcode:`TXD`, and
3654 :opcode:`TXP`. For reference, OpenGL and Direct3D conventions are shown as
3657 +--------------------+--------------+--------------------+--------------+
3658 | Texture Components | Gallium | OpenGL | Direct3D 9 |
3659 +====================+==============+====================+==============+
3660 | R | (r, 0, 0, 1) | (r, 0, 0, 1) | (r, 1, 1, 1) |
3661 +--------------------+--------------+--------------------+--------------+
3662 | RG | (r, g, 0, 1) | (r, g, 0, 1) | (r, g, 1, 1) |
3663 +--------------------+--------------+--------------------+--------------+
3664 | RGB | (r, g, b, 1) | (r, g, b, 1) | (r, g, b, 1) |
3665 +--------------------+--------------+--------------------+--------------+
3666 | RGBA | (r, g, b, a) | (r, g, b, a) | (r, g, b, a) |
3667 +--------------------+--------------+--------------------+--------------+
3668 | A | (0, 0, 0, a) | (0, 0, 0, a) | (0, 0, 0, a) |
3669 +--------------------+--------------+--------------------+--------------+
3670 | L | (l, l, l, 1) | (l, l, l, 1) | (l, l, l, 1) |
3671 +--------------------+--------------+--------------------+--------------+
3672 | LA | (l, l, l, a) | (l, l, l, a) | (l, l, l, a) |
3673 +--------------------+--------------+--------------------+--------------+
3674 | I | (i, i, i, i) | (i, i, i, i) | N/A |
3675 +--------------------+--------------+--------------------+--------------+
3676 | UV | XXX TBD | (0, 0, 0, 1) | (u, v, 1, 1) |
3677 | | | [#envmap-bumpmap]_ | |
3678 +--------------------+--------------+--------------------+--------------+
3679 | Z | XXX TBD | (z, z, z, 1) | (0, z, 0, 1) |
3680 | | | [#depth-tex-mode]_ | |
3681 +--------------------+--------------+--------------------+--------------+
3682 | S | (s, s, s, s) | unknown | unknown |
3683 +--------------------+--------------+--------------------+--------------+
3685 .. [#envmap-bumpmap] http://www.opengl.org/registry/specs/ATI/envmap_bumpmap.txt
3686 .. [#depth-tex-mode] the default is (z, z, z, 1) but may also be (0, 0, 0, z)
3687 or (z, z, z, z) depending on the value of GL_DEPTH_TEXTURE_MODE.