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 modifier on instructions).
31 For inputs which have a floating point type, both absolute value and negation
32 modifiers are supported (with absolute value being applied first).
33 TGSI_OPCODE_MOV is considered to have float input type for applying modifiers.
35 For inputs which have signed or unsigned type only the negate modifier is
42 ^^^^^^^^^^^^^^^^^^^^^^^^^
44 These opcodes are guaranteed to be available regardless of the driver being
47 .. opcode:: ARL - Address Register Load
51 dst.x = \lfloor src.x\rfloor
53 dst.y = \lfloor src.y\rfloor
55 dst.z = \lfloor src.z\rfloor
57 dst.w = \lfloor src.w\rfloor
60 .. opcode:: MOV - Move
73 .. opcode:: LIT - Light Coefficients
78 dst.y &= max(src.x, 0) \\
79 dst.z &= (src.x > 0) ? max(src.y, 0)^{clamp(src.w, -128, 128))} : 0 \\
83 .. opcode:: RCP - Reciprocal
85 This instruction replicates its result.
92 .. opcode:: RSQ - Reciprocal Square Root
94 This instruction replicates its result. The results are undefined for src <= 0.
98 dst = \frac{1}{\sqrt{src.x}}
101 .. opcode:: SQRT - Square Root
103 This instruction replicates its result. The results are undefined for src < 0.
110 .. opcode:: EXP - Approximate Exponential Base 2
114 dst.x &= 2^{\lfloor src.x\rfloor} \\
115 dst.y &= src.x - \lfloor src.x\rfloor \\
116 dst.z &= 2^{src.x} \\
120 .. opcode:: LOG - Approximate Logarithm Base 2
124 dst.x &= \lfloor\log_2{|src.x|}\rfloor \\
125 dst.y &= \frac{|src.x|}{2^{\lfloor\log_2{|src.x|}\rfloor}} \\
126 dst.z &= \log_2{|src.x|} \\
130 .. opcode:: MUL - Multiply
134 dst.x = src0.x \times src1.x
136 dst.y = src0.y \times src1.y
138 dst.z = src0.z \times src1.z
140 dst.w = src0.w \times src1.w
143 .. opcode:: ADD - Add
147 dst.x = src0.x + src1.x
149 dst.y = src0.y + src1.y
151 dst.z = src0.z + src1.z
153 dst.w = src0.w + src1.w
156 .. opcode:: DP3 - 3-component Dot Product
158 This instruction replicates its result.
162 dst = src0.x \times src1.x + src0.y \times src1.y + src0.z \times src1.z
165 .. opcode:: DP4 - 4-component Dot Product
167 This instruction replicates its result.
171 dst = src0.x \times src1.x + src0.y \times src1.y + src0.z \times src1.z + src0.w \times src1.w
174 .. opcode:: DST - Distance Vector
179 dst.y &= src0.y \times src1.y\\
184 .. opcode:: MIN - Minimum
188 dst.x = min(src0.x, src1.x)
190 dst.y = min(src0.y, src1.y)
192 dst.z = min(src0.z, src1.z)
194 dst.w = min(src0.w, src1.w)
197 .. opcode:: MAX - Maximum
201 dst.x = max(src0.x, src1.x)
203 dst.y = max(src0.y, src1.y)
205 dst.z = max(src0.z, src1.z)
207 dst.w = max(src0.w, src1.w)
210 .. opcode:: SLT - Set On Less Than
214 dst.x = (src0.x < src1.x) ? 1.0F : 0.0F
216 dst.y = (src0.y < src1.y) ? 1.0F : 0.0F
218 dst.z = (src0.z < src1.z) ? 1.0F : 0.0F
220 dst.w = (src0.w < src1.w) ? 1.0F : 0.0F
223 .. opcode:: SGE - Set On Greater Equal Than
227 dst.x = (src0.x >= src1.x) ? 1.0F : 0.0F
229 dst.y = (src0.y >= src1.y) ? 1.0F : 0.0F
231 dst.z = (src0.z >= src1.z) ? 1.0F : 0.0F
233 dst.w = (src0.w >= src1.w) ? 1.0F : 0.0F
236 .. opcode:: MAD - Multiply And Add
240 dst.x = src0.x \times src1.x + src2.x
242 dst.y = src0.y \times src1.y + src2.y
244 dst.z = src0.z \times src1.z + src2.z
246 dst.w = src0.w \times src1.w + src2.w
249 .. opcode:: SUB - Subtract
253 dst.x = src0.x - src1.x
255 dst.y = src0.y - src1.y
257 dst.z = src0.z - src1.z
259 dst.w = src0.w - src1.w
262 .. opcode:: LRP - Linear Interpolate
266 dst.x = src0.x \times src1.x + (1 - src0.x) \times src2.x
268 dst.y = src0.y \times src1.y + (1 - src0.y) \times src2.y
270 dst.z = src0.z \times src1.z + (1 - src0.z) \times src2.z
272 dst.w = src0.w \times src1.w + (1 - src0.w) \times src2.w
275 .. opcode:: CND - Condition
279 dst.x = (src2.x > 0.5) ? src0.x : src1.x
281 dst.y = (src2.y > 0.5) ? src0.y : src1.y
283 dst.z = (src2.z > 0.5) ? src0.z : src1.z
285 dst.w = (src2.w > 0.5) ? src0.w : src1.w
288 .. opcode:: DP2A - 2-component Dot Product And Add
292 dst.x = src0.x \times src1.x + src0.y \times src1.y + src2.x
294 dst.y = src0.x \times src1.x + src0.y \times src1.y + src2.x
296 dst.z = src0.x \times src1.x + src0.y \times src1.y + src2.x
298 dst.w = src0.x \times src1.x + src0.y \times src1.y + src2.x
301 .. opcode:: FRC - Fraction
305 dst.x = src.x - \lfloor src.x\rfloor
307 dst.y = src.y - \lfloor src.y\rfloor
309 dst.z = src.z - \lfloor src.z\rfloor
311 dst.w = src.w - \lfloor src.w\rfloor
314 .. opcode:: CLAMP - Clamp
318 dst.x = clamp(src0.x, src1.x, src2.x)
320 dst.y = clamp(src0.y, src1.y, src2.y)
322 dst.z = clamp(src0.z, src1.z, src2.z)
324 dst.w = clamp(src0.w, src1.w, src2.w)
327 .. opcode:: FLR - Floor
329 This is identical to :opcode:`ARL`.
333 dst.x = \lfloor src.x\rfloor
335 dst.y = \lfloor src.y\rfloor
337 dst.z = \lfloor src.z\rfloor
339 dst.w = \lfloor src.w\rfloor
342 .. opcode:: ROUND - Round
355 .. opcode:: EX2 - Exponential Base 2
357 This instruction replicates its result.
364 .. opcode:: LG2 - Logarithm Base 2
366 This instruction replicates its result.
373 .. opcode:: POW - Power
375 This instruction replicates its result.
379 dst = src0.x^{src1.x}
381 .. opcode:: XPD - Cross Product
385 dst.x = src0.y \times src1.z - src1.y \times src0.z
387 dst.y = src0.z \times src1.x - src1.z \times src0.x
389 dst.z = src0.x \times src1.y - src1.x \times src0.y
394 .. opcode:: ABS - Absolute
407 .. opcode:: DPH - Homogeneous Dot Product
409 This instruction replicates its result.
413 dst = src0.x \times src1.x + src0.y \times src1.y + src0.z \times src1.z + src1.w
416 .. opcode:: COS - Cosine
418 This instruction replicates its result.
425 .. opcode:: DDX, DDX_FINE - Derivative Relative To X
427 The fine variant is only used when ``PIPE_CAP_TGSI_FS_FINE_DERIVATIVE`` is
428 advertised. When it is, the fine version guarantees one derivative per row
429 while DDX is allowed to be the same for the entire 2x2 quad.
433 dst.x = partialx(src.x)
435 dst.y = partialx(src.y)
437 dst.z = partialx(src.z)
439 dst.w = partialx(src.w)
442 .. opcode:: DDY, DDY_FINE - Derivative Relative To Y
444 The fine variant is only used when ``PIPE_CAP_TGSI_FS_FINE_DERIVATIVE`` is
445 advertised. When it is, the fine version guarantees one derivative per column
446 while DDY is allowed to be the same for the entire 2x2 quad.
450 dst.x = partialy(src.x)
452 dst.y = partialy(src.y)
454 dst.z = partialy(src.z)
456 dst.w = partialy(src.w)
459 .. opcode:: PK2H - Pack Two 16-bit Floats
464 .. opcode:: PK2US - Pack Two Unsigned 16-bit Scalars
469 .. opcode:: PK4B - Pack Four Signed 8-bit Scalars
474 .. opcode:: PK4UB - Pack Four Unsigned 8-bit Scalars
479 .. opcode:: SEQ - Set On Equal
483 dst.x = (src0.x == src1.x) ? 1.0F : 0.0F
485 dst.y = (src0.y == src1.y) ? 1.0F : 0.0F
487 dst.z = (src0.z == src1.z) ? 1.0F : 0.0F
489 dst.w = (src0.w == src1.w) ? 1.0F : 0.0F
492 .. opcode:: SFL - Set On False
494 This instruction replicates its result.
502 Considered for removal.
505 .. opcode:: SGT - Set On Greater Than
509 dst.x = (src0.x > src1.x) ? 1.0F : 0.0F
511 dst.y = (src0.y > src1.y) ? 1.0F : 0.0F
513 dst.z = (src0.z > src1.z) ? 1.0F : 0.0F
515 dst.w = (src0.w > src1.w) ? 1.0F : 0.0F
518 .. opcode:: SIN - Sine
520 This instruction replicates its result.
527 .. opcode:: SLE - Set On Less Equal Than
531 dst.x = (src0.x <= src1.x) ? 1.0F : 0.0F
533 dst.y = (src0.y <= src1.y) ? 1.0F : 0.0F
535 dst.z = (src0.z <= src1.z) ? 1.0F : 0.0F
537 dst.w = (src0.w <= src1.w) ? 1.0F : 0.0F
540 .. opcode:: SNE - Set On Not Equal
544 dst.x = (src0.x != src1.x) ? 1.0F : 0.0F
546 dst.y = (src0.y != src1.y) ? 1.0F : 0.0F
548 dst.z = (src0.z != src1.z) ? 1.0F : 0.0F
550 dst.w = (src0.w != src1.w) ? 1.0F : 0.0F
553 .. opcode:: STR - Set On True
555 This instruction replicates its result.
562 .. opcode:: TEX - Texture Lookup
564 for array textures src0.y contains the slice for 1D,
565 and src0.z contain the slice for 2D.
567 for shadow textures with no arrays (and not cube map),
568 src0.z contains the reference value.
570 for shadow textures with arrays, src0.z contains
571 the reference value for 1D arrays, and src0.w contains
572 the reference value for 2D arrays and cube maps.
574 for cube map array shadow textures, the reference value
575 cannot be passed in src0.w, and TEX2 must be used instead.
581 shadow_ref = src0.z or src0.w (optional)
585 dst = texture\_sample(unit, coord, shadow_ref)
588 .. opcode:: TEX2 - Texture Lookup (for shadow cube map arrays only)
590 this is the same as TEX, but uses another reg to encode the
601 dst = texture\_sample(unit, coord, shadow_ref)
606 .. opcode:: TXD - Texture Lookup with Derivatives
618 dst = texture\_sample\_deriv(unit, coord, ddx, ddy)
621 .. opcode:: TXP - Projective Texture Lookup
625 coord.x = src0.x / src0.w
627 coord.y = src0.y / src0.w
629 coord.z = src0.z / src0.w
635 dst = texture\_sample(unit, coord)
638 .. opcode:: UP2H - Unpack Two 16-Bit Floats
644 Considered for removal.
646 .. opcode:: UP2US - Unpack Two Unsigned 16-Bit Scalars
652 Considered for removal.
654 .. opcode:: UP4B - Unpack Four Signed 8-Bit Values
660 Considered for removal.
662 .. opcode:: UP4UB - Unpack Four Unsigned 8-Bit Scalars
668 Considered for removal.
671 .. opcode:: ARR - Address Register Load With Round
684 .. opcode:: SSG - Set Sign
688 dst.x = (src.x > 0) ? 1 : (src.x < 0) ? -1 : 0
690 dst.y = (src.y > 0) ? 1 : (src.y < 0) ? -1 : 0
692 dst.z = (src.z > 0) ? 1 : (src.z < 0) ? -1 : 0
694 dst.w = (src.w > 0) ? 1 : (src.w < 0) ? -1 : 0
697 .. opcode:: CMP - Compare
701 dst.x = (src0.x < 0) ? src1.x : src2.x
703 dst.y = (src0.y < 0) ? src1.y : src2.y
705 dst.z = (src0.z < 0) ? src1.z : src2.z
707 dst.w = (src0.w < 0) ? src1.w : src2.w
710 .. opcode:: KILL_IF - Conditional Discard
712 Conditional discard. Allowed in fragment shaders only.
716 if (src.x < 0 || src.y < 0 || src.z < 0 || src.w < 0)
721 .. opcode:: KILL - Discard
723 Unconditional discard. Allowed in fragment shaders only.
726 .. opcode:: SCS - Sine Cosine
739 .. opcode:: TXB - Texture Lookup With Bias
741 for cube map array textures and shadow cube maps, the bias value
742 cannot be passed in src0.w, and TXB2 must be used instead.
744 if the target is a shadow texture, the reference value is always
745 in src.z (this prevents shadow 3d and shadow 2d arrays from
746 using this instruction, but this is not needed).
762 dst = texture\_sample(unit, coord, bias)
765 .. opcode:: TXB2 - Texture Lookup With Bias (some cube maps only)
767 this is the same as TXB, but uses another reg to encode the
768 lod bias value for cube map arrays and shadow cube maps.
769 Presumably shadow 2d arrays and shadow 3d targets could use
770 this encoding too, but this is not legal.
772 shadow cube map arrays are neither possible nor required.
782 dst = texture\_sample(unit, coord, bias)
785 .. opcode:: DIV - Divide
789 dst.x = \frac{src0.x}{src1.x}
791 dst.y = \frac{src0.y}{src1.y}
793 dst.z = \frac{src0.z}{src1.z}
795 dst.w = \frac{src0.w}{src1.w}
798 .. opcode:: DP2 - 2-component Dot Product
800 This instruction replicates its result.
804 dst = src0.x \times src1.x + src0.y \times src1.y
807 .. opcode:: TXL - Texture Lookup With explicit LOD
809 for cube map array textures, the explicit lod value
810 cannot be passed in src0.w, and TXL2 must be used instead.
812 if the target is a shadow texture, the reference value is always
813 in src.z (this prevents shadow 3d / 2d array / cube targets from
814 using this instruction, but this is not needed).
830 dst = texture\_sample(unit, coord, lod)
833 .. opcode:: TXL2 - Texture Lookup With explicit LOD (for cube map arrays only)
835 this is the same as TXL, but uses another reg to encode the
837 Presumably shadow 3d / 2d array / cube targets could use
838 this encoding too, but this is not legal.
840 shadow cube map arrays are neither possible nor required.
850 dst = texture\_sample(unit, coord, lod)
853 .. opcode:: PUSHA - Push Address Register On Stack
862 Considered for cleanup.
866 Considered for removal.
868 .. opcode:: POPA - Pop Address Register From Stack
877 Considered for cleanup.
881 Considered for removal.
884 .. opcode:: BRA - Branch
890 Considered for removal.
893 .. opcode:: CALLNZ - Subroutine Call If Not Zero
899 Considered for cleanup.
903 Considered for removal.
907 ^^^^^^^^^^^^^^^^^^^^^^^^
909 These opcodes are primarily provided for special-use computational shaders.
910 Support for these opcodes indicated by a special pipe capability bit (TBD).
912 XXX doesn't look like most of the opcodes really belong here.
914 .. opcode:: CEIL - Ceiling
918 dst.x = \lceil src.x\rceil
920 dst.y = \lceil src.y\rceil
922 dst.z = \lceil src.z\rceil
924 dst.w = \lceil src.w\rceil
927 .. opcode:: TRUNC - Truncate
940 .. opcode:: MOD - Modulus
944 dst.x = src0.x \bmod src1.x
946 dst.y = src0.y \bmod src1.y
948 dst.z = src0.z \bmod src1.z
950 dst.w = src0.w \bmod src1.w
953 .. opcode:: UARL - Integer Address Register Load
955 Moves the contents of the source register, assumed to be an integer, into the
956 destination register, which is assumed to be an address (ADDR) register.
959 .. opcode:: SAD - Sum Of Absolute Differences
963 dst.x = |src0.x - src1.x| + src2.x
965 dst.y = |src0.y - src1.y| + src2.y
967 dst.z = |src0.z - src1.z| + src2.z
969 dst.w = |src0.w - src1.w| + src2.w
972 .. opcode:: TXF - Texel Fetch
974 As per NV_gpu_shader4, extract a single texel from a specified texture
975 image. The source sampler may not be a CUBE or SHADOW. src 0 is a
976 four-component signed integer vector used to identify the single texel
977 accessed. 3 components + level. Just like texture instructions, an optional
978 offset vector is provided, which is subject to various driver restrictions
979 (regarding range, source of offsets).
980 TXF(uint_vec coord, int_vec offset).
983 .. opcode:: TXQ - Texture Size Query
985 As per NV_gpu_program4, retrieve the dimensions of the texture depending on
986 the target. For 1D (width), 2D/RECT/CUBE (width, height), 3D (width, height,
987 depth), 1D array (width, layers), 2D array (width, height, layers).
988 Also return the number of accessible levels (last_level - first_level + 1)
991 For components which don't return a resource dimension, their value
999 dst.x = texture\_width(unit, lod)
1001 dst.y = texture\_height(unit, lod)
1003 dst.z = texture\_depth(unit, lod)
1005 dst.w = texture\_levels(unit)
1007 .. opcode:: TG4 - Texture Gather
1009 As per ARB_texture_gather, gathers the four texels to be used in a bi-linear
1010 filtering operation and packs them into a single register. Only works with
1011 2D, 2D array, cubemaps, and cubemaps arrays. For 2D textures, only the
1012 addressing modes of the sampler and the top level of any mip pyramid are
1013 used. Set W to zero. It behaves like the TEX instruction, but a filtered
1014 sample is not generated. The four samples that contribute to filtering are
1015 placed into xyzw in clockwise order, starting with the (u,v) texture
1016 coordinate delta at the following locations (-, +), (+, +), (+, -), (-, -),
1017 where the magnitude of the deltas are half a texel.
1019 PIPE_CAP_TEXTURE_SM5 enhances this instruction to support shadow per-sample
1020 depth compares, single component selection, and a non-constant offset. It
1021 doesn't allow support for the GL independent offset to get i0,j0. This would
1022 require another CAP is hw can do it natively. For now we lower that before
1031 dst = texture\_gather4 (unit, coord, component)
1033 (with SM5 - cube array shadow)
1041 dst = texture\_gather (uint, coord, compare)
1043 .. opcode:: LODQ - level of detail query
1045 Compute the LOD information that the texture pipe would use to access the
1046 texture. The Y component contains the computed LOD lambda_prime. The X
1047 component contains the LOD that will be accessed, based on min/max lod's
1054 dst.xy = lodq(uint, coord);
1057 ^^^^^^^^^^^^^^^^^^^^^^^^
1058 These opcodes are used for integer operations.
1059 Support for these opcodes indicated by PIPE_SHADER_CAP_INTEGERS (all of them?)
1062 .. opcode:: I2F - Signed Integer To Float
1064 Rounding is unspecified (round to nearest even suggested).
1068 dst.x = (float) src.x
1070 dst.y = (float) src.y
1072 dst.z = (float) src.z
1074 dst.w = (float) src.w
1077 .. opcode:: U2F - Unsigned Integer To Float
1079 Rounding is unspecified (round to nearest even suggested).
1083 dst.x = (float) src.x
1085 dst.y = (float) src.y
1087 dst.z = (float) src.z
1089 dst.w = (float) src.w
1092 .. opcode:: F2I - Float to Signed Integer
1094 Rounding is towards zero (truncate).
1095 Values outside signed range (including NaNs) produce undefined results.
1108 .. opcode:: F2U - Float to Unsigned Integer
1110 Rounding is towards zero (truncate).
1111 Values outside unsigned range (including NaNs) produce undefined results.
1115 dst.x = (unsigned) src.x
1117 dst.y = (unsigned) src.y
1119 dst.z = (unsigned) src.z
1121 dst.w = (unsigned) src.w
1124 .. opcode:: UADD - Integer Add
1126 This instruction works the same for signed and unsigned integers.
1127 The low 32bit of the result is returned.
1131 dst.x = src0.x + src1.x
1133 dst.y = src0.y + src1.y
1135 dst.z = src0.z + src1.z
1137 dst.w = src0.w + src1.w
1140 .. opcode:: UMAD - Integer Multiply And Add
1142 This instruction works the same for signed and unsigned integers.
1143 The multiplication returns the low 32bit (as does the result itself).
1147 dst.x = src0.x \times src1.x + src2.x
1149 dst.y = src0.y \times src1.y + src2.y
1151 dst.z = src0.z \times src1.z + src2.z
1153 dst.w = src0.w \times src1.w + src2.w
1156 .. opcode:: UMUL - Integer Multiply
1158 This instruction works the same for signed and unsigned integers.
1159 The low 32bit of the result is returned.
1163 dst.x = src0.x \times src1.x
1165 dst.y = src0.y \times src1.y
1167 dst.z = src0.z \times src1.z
1169 dst.w = src0.w \times src1.w
1172 .. opcode:: IMUL_HI - Signed Integer Multiply High Bits
1174 The high 32bits of the multiplication of 2 signed integers are returned.
1178 dst.x = (src0.x \times src1.x) >> 32
1180 dst.y = (src0.y \times src1.y) >> 32
1182 dst.z = (src0.z \times src1.z) >> 32
1184 dst.w = (src0.w \times src1.w) >> 32
1187 .. opcode:: UMUL_HI - Unsigned Integer Multiply High Bits
1189 The high 32bits of the multiplication of 2 unsigned integers are returned.
1193 dst.x = (src0.x \times src1.x) >> 32
1195 dst.y = (src0.y \times src1.y) >> 32
1197 dst.z = (src0.z \times src1.z) >> 32
1199 dst.w = (src0.w \times src1.w) >> 32
1202 .. opcode:: IDIV - Signed Integer Division
1204 TBD: behavior for division by zero.
1208 dst.x = src0.x \ src1.x
1210 dst.y = src0.y \ src1.y
1212 dst.z = src0.z \ src1.z
1214 dst.w = src0.w \ src1.w
1217 .. opcode:: UDIV - Unsigned Integer Division
1219 For division by zero, 0xffffffff is returned.
1223 dst.x = src0.x \ src1.x
1225 dst.y = src0.y \ src1.y
1227 dst.z = src0.z \ src1.z
1229 dst.w = src0.w \ src1.w
1232 .. opcode:: UMOD - Unsigned Integer Remainder
1234 If second arg is zero, 0xffffffff is returned.
1238 dst.x = src0.x \ src1.x
1240 dst.y = src0.y \ src1.y
1242 dst.z = src0.z \ src1.z
1244 dst.w = src0.w \ src1.w
1247 .. opcode:: NOT - Bitwise Not
1260 .. opcode:: AND - Bitwise And
1264 dst.x = src0.x \& src1.x
1266 dst.y = src0.y \& src1.y
1268 dst.z = src0.z \& src1.z
1270 dst.w = src0.w \& src1.w
1273 .. opcode:: OR - Bitwise Or
1277 dst.x = src0.x | src1.x
1279 dst.y = src0.y | src1.y
1281 dst.z = src0.z | src1.z
1283 dst.w = src0.w | src1.w
1286 .. opcode:: XOR - Bitwise Xor
1290 dst.x = src0.x \oplus src1.x
1292 dst.y = src0.y \oplus src1.y
1294 dst.z = src0.z \oplus src1.z
1296 dst.w = src0.w \oplus src1.w
1299 .. opcode:: IMAX - Maximum of Signed Integers
1303 dst.x = max(src0.x, src1.x)
1305 dst.y = max(src0.y, src1.y)
1307 dst.z = max(src0.z, src1.z)
1309 dst.w = max(src0.w, src1.w)
1312 .. opcode:: UMAX - Maximum of Unsigned Integers
1316 dst.x = max(src0.x, src1.x)
1318 dst.y = max(src0.y, src1.y)
1320 dst.z = max(src0.z, src1.z)
1322 dst.w = max(src0.w, src1.w)
1325 .. opcode:: IMIN - Minimum of Signed Integers
1329 dst.x = min(src0.x, src1.x)
1331 dst.y = min(src0.y, src1.y)
1333 dst.z = min(src0.z, src1.z)
1335 dst.w = min(src0.w, src1.w)
1338 .. opcode:: UMIN - Minimum of Unsigned Integers
1342 dst.x = min(src0.x, src1.x)
1344 dst.y = min(src0.y, src1.y)
1346 dst.z = min(src0.z, src1.z)
1348 dst.w = min(src0.w, src1.w)
1351 .. opcode:: SHL - Shift Left
1353 The shift count is masked with 0x1f before the shift is applied.
1357 dst.x = src0.x << (0x1f \& src1.x)
1359 dst.y = src0.y << (0x1f \& src1.y)
1361 dst.z = src0.z << (0x1f \& src1.z)
1363 dst.w = src0.w << (0x1f \& src1.w)
1366 .. opcode:: ISHR - Arithmetic Shift Right (of Signed Integer)
1368 The shift count is masked with 0x1f before the shift is applied.
1372 dst.x = src0.x >> (0x1f \& src1.x)
1374 dst.y = src0.y >> (0x1f \& src1.y)
1376 dst.z = src0.z >> (0x1f \& src1.z)
1378 dst.w = src0.w >> (0x1f \& src1.w)
1381 .. opcode:: USHR - Logical Shift Right
1383 The shift count is masked with 0x1f before the shift is applied.
1387 dst.x = src0.x >> (unsigned) (0x1f \& src1.x)
1389 dst.y = src0.y >> (unsigned) (0x1f \& src1.y)
1391 dst.z = src0.z >> (unsigned) (0x1f \& src1.z)
1393 dst.w = src0.w >> (unsigned) (0x1f \& src1.w)
1396 .. opcode:: UCMP - Integer Conditional Move
1400 dst.x = src0.x ? src1.x : src2.x
1402 dst.y = src0.y ? src1.y : src2.y
1404 dst.z = src0.z ? src1.z : src2.z
1406 dst.w = src0.w ? src1.w : src2.w
1410 .. opcode:: ISSG - Integer Set Sign
1414 dst.x = (src0.x < 0) ? -1 : (src0.x > 0) ? 1 : 0
1416 dst.y = (src0.y < 0) ? -1 : (src0.y > 0) ? 1 : 0
1418 dst.z = (src0.z < 0) ? -1 : (src0.z > 0) ? 1 : 0
1420 dst.w = (src0.w < 0) ? -1 : (src0.w > 0) ? 1 : 0
1424 .. opcode:: FSLT - Float Set On Less Than (ordered)
1426 Same comparison as SLT but returns integer instead of 1.0/0.0 float
1430 dst.x = (src0.x < src1.x) ? \sim 0 : 0
1432 dst.y = (src0.y < src1.y) ? \sim 0 : 0
1434 dst.z = (src0.z < src1.z) ? \sim 0 : 0
1436 dst.w = (src0.w < src1.w) ? \sim 0 : 0
1439 .. opcode:: ISLT - Signed Integer Set On Less Than
1443 dst.x = (src0.x < src1.x) ? \sim 0 : 0
1445 dst.y = (src0.y < src1.y) ? \sim 0 : 0
1447 dst.z = (src0.z < src1.z) ? \sim 0 : 0
1449 dst.w = (src0.w < src1.w) ? \sim 0 : 0
1452 .. opcode:: USLT - Unsigned Integer Set On Less Than
1456 dst.x = (src0.x < src1.x) ? \sim 0 : 0
1458 dst.y = (src0.y < src1.y) ? \sim 0 : 0
1460 dst.z = (src0.z < src1.z) ? \sim 0 : 0
1462 dst.w = (src0.w < src1.w) ? \sim 0 : 0
1465 .. opcode:: FSGE - Float Set On Greater Equal Than (ordered)
1467 Same comparison as SGE but returns integer instead of 1.0/0.0 float
1471 dst.x = (src0.x >= src1.x) ? \sim 0 : 0
1473 dst.y = (src0.y >= src1.y) ? \sim 0 : 0
1475 dst.z = (src0.z >= src1.z) ? \sim 0 : 0
1477 dst.w = (src0.w >= src1.w) ? \sim 0 : 0
1480 .. opcode:: ISGE - Signed Integer Set On Greater Equal Than
1484 dst.x = (src0.x >= src1.x) ? \sim 0 : 0
1486 dst.y = (src0.y >= src1.y) ? \sim 0 : 0
1488 dst.z = (src0.z >= src1.z) ? \sim 0 : 0
1490 dst.w = (src0.w >= src1.w) ? \sim 0 : 0
1493 .. opcode:: USGE - Unsigned Integer Set On Greater Equal Than
1497 dst.x = (src0.x >= src1.x) ? \sim 0 : 0
1499 dst.y = (src0.y >= src1.y) ? \sim 0 : 0
1501 dst.z = (src0.z >= src1.z) ? \sim 0 : 0
1503 dst.w = (src0.w >= src1.w) ? \sim 0 : 0
1506 .. opcode:: FSEQ - Float Set On Equal (ordered)
1508 Same comparison as SEQ but returns integer instead of 1.0/0.0 float
1512 dst.x = (src0.x == src1.x) ? \sim 0 : 0
1514 dst.y = (src0.y == src1.y) ? \sim 0 : 0
1516 dst.z = (src0.z == src1.z) ? \sim 0 : 0
1518 dst.w = (src0.w == src1.w) ? \sim 0 : 0
1521 .. opcode:: USEQ - Integer Set On Equal
1525 dst.x = (src0.x == src1.x) ? \sim 0 : 0
1527 dst.y = (src0.y == src1.y) ? \sim 0 : 0
1529 dst.z = (src0.z == src1.z) ? \sim 0 : 0
1531 dst.w = (src0.w == src1.w) ? \sim 0 : 0
1534 .. opcode:: FSNE - Float Set On Not Equal (unordered)
1536 Same comparison as SNE but returns integer instead of 1.0/0.0 float
1540 dst.x = (src0.x != src1.x) ? \sim 0 : 0
1542 dst.y = (src0.y != src1.y) ? \sim 0 : 0
1544 dst.z = (src0.z != src1.z) ? \sim 0 : 0
1546 dst.w = (src0.w != src1.w) ? \sim 0 : 0
1549 .. opcode:: USNE - Integer Set On Not Equal
1553 dst.x = (src0.x != src1.x) ? \sim 0 : 0
1555 dst.y = (src0.y != src1.y) ? \sim 0 : 0
1557 dst.z = (src0.z != src1.z) ? \sim 0 : 0
1559 dst.w = (src0.w != src1.w) ? \sim 0 : 0
1562 .. opcode:: INEG - Integer Negate
1577 .. opcode:: IABS - Integer Absolute Value
1591 These opcodes are used for bit-level manipulation of integers.
1593 .. opcode:: IBFE - Signed Bitfield Extract
1595 See SM5 instruction of the same name. Extracts a set of bits from the input,
1596 and sign-extends them if the high bit of the extracted window is set.
1600 def ibfe(value, offset, bits):
1601 offset = offset & 0x1f
1603 if bits == 0: return 0
1604 # Note: >> sign-extends
1605 if width + offset < 32:
1606 return (value << (32 - offset - bits)) >> (32 - bits)
1608 return value >> offset
1610 .. opcode:: UBFE - Unsigned Bitfield Extract
1612 See SM5 instruction of the same name. Extracts a set of bits from the input,
1613 without any sign-extension.
1617 def ubfe(value, offset, bits):
1618 offset = offset & 0x1f
1620 if bits == 0: return 0
1621 # Note: >> does not sign-extend
1622 if width + offset < 32:
1623 return (value << (32 - offset - bits)) >> (32 - bits)
1625 return value >> offset
1627 .. opcode:: BFI - Bitfield Insert
1629 See SM5 instruction of the same name. Replaces a bit region of 'base' with
1630 the low bits of 'insert'.
1634 def bfi(base, insert, offset, bits):
1635 offset = offset & 0x1f
1637 mask = ((1 << bits) - 1) << offset
1638 return ((insert << offset) & mask) | (base & ~mask)
1640 .. opcode:: BREV - Bitfield Reverse
1642 See SM5 instruction BFREV. Reverses the bits of the argument.
1644 .. opcode:: POPC - Population Count
1646 See SM5 instruction COUNTBITS. Counts the number of set bits in the argument.
1648 .. opcode:: LSB - Index of lowest set bit
1650 See SM5 instruction FIRSTBIT_LO. Computes the 0-based index of the first set
1651 bit of the argument. Returns -1 if none are set.
1653 .. opcode:: IMSB - Index of highest non-sign bit
1655 See SM5 instruction FIRSTBIT_SHI. Computes the 0-based index of the highest
1656 non-sign bit of the argument (i.e. highest 0 bit for negative numbers,
1657 highest 1 bit for positive numbers). Returns -1 if all bits are the same
1658 (i.e. for inputs 0 and -1).
1660 .. opcode:: UMSB - Index of highest set bit
1662 See SM5 instruction FIRSTBIT_HI. Computes the 0-based index of the highest
1663 set bit of the argument. Returns -1 if none are set.
1666 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1668 These opcodes are only supported in geometry shaders; they have no meaning
1669 in any other type of shader.
1671 .. opcode:: EMIT - Emit
1673 Generate a new vertex for the current primitive into the specified vertex
1674 stream using the values in the output registers.
1677 .. opcode:: ENDPRIM - End Primitive
1679 Complete the current primitive in the specified vertex stream (consisting of
1680 the emitted vertices), and start a new one.
1686 These opcodes are part of :term:`GLSL`'s opcode set. Support for these
1687 opcodes is determined by a special capability bit, ``GLSL``.
1688 Some require glsl version 1.30 (UIF/BREAKC/SWITCH/CASE/DEFAULT/ENDSWITCH).
1690 .. opcode:: CAL - Subroutine Call
1696 .. opcode:: RET - Subroutine Call Return
1701 .. opcode:: CONT - Continue
1703 Unconditionally moves the point of execution to the instruction after the
1704 last bgnloop. The instruction must appear within a bgnloop/endloop.
1708 Support for CONT is determined by a special capability bit,
1709 ``TGSI_CONT_SUPPORTED``. See :ref:`Screen` for more information.
1712 .. opcode:: BGNLOOP - Begin a Loop
1714 Start a loop. Must have a matching endloop.
1717 .. opcode:: BGNSUB - Begin Subroutine
1719 Starts definition of a subroutine. Must have a matching endsub.
1722 .. opcode:: ENDLOOP - End a Loop
1724 End a loop started with bgnloop.
1727 .. opcode:: ENDSUB - End Subroutine
1729 Ends definition of a subroutine.
1732 .. opcode:: NOP - No Operation
1737 .. opcode:: BRK - Break
1739 Unconditionally moves the point of execution to the instruction after the
1740 next endloop or endswitch. The instruction must appear within a loop/endloop
1741 or switch/endswitch.
1744 .. opcode:: BREAKC - Break Conditional
1746 Conditionally moves the point of execution to the instruction after the
1747 next endloop or endswitch. The instruction must appear within a loop/endloop
1748 or switch/endswitch.
1749 Condition evaluates to true if src0.x != 0 where src0.x is interpreted
1750 as an integer register.
1754 Considered for removal as it's quite inconsistent wrt other opcodes
1755 (could emulate with UIF/BRK/ENDIF).
1758 .. opcode:: IF - Float If
1760 Start an IF ... ELSE .. ENDIF block. Condition evaluates to true if
1764 where src0.x is interpreted as a floating point register.
1767 .. opcode:: UIF - Bitwise If
1769 Start an UIF ... ELSE .. ENDIF block. Condition evaluates to true if
1773 where src0.x is interpreted as an integer register.
1776 .. opcode:: ELSE - Else
1778 Starts an else block, after an IF or UIF statement.
1781 .. opcode:: ENDIF - End If
1783 Ends an IF or UIF block.
1786 .. opcode:: SWITCH - Switch
1788 Starts a C-style switch expression. The switch consists of one or multiple
1789 CASE statements, and at most one DEFAULT statement. Execution of a statement
1790 ends when a BRK is hit, but just like in C falling through to other cases
1791 without a break is allowed. Similarly, DEFAULT label is allowed anywhere not
1792 just as last statement, and fallthrough is allowed into/from it.
1793 CASE src arguments are evaluated at bit level against the SWITCH src argument.
1799 (some instructions here)
1802 (some instructions here)
1805 (some instructions here)
1810 .. opcode:: CASE - Switch case
1812 This represents a switch case label. The src arg must be an integer immediate.
1815 .. opcode:: DEFAULT - Switch default
1817 This represents the default case in the switch, which is taken if no other
1821 .. opcode:: ENDSWITCH - End of switch
1823 Ends a switch expression.
1829 The interpolation instructions allow an input to be interpolated in a
1830 different way than its declaration. This corresponds to the GLSL 4.00
1831 interpolateAt* functions. The first argument of each of these must come from
1832 ``TGSI_FILE_INPUT``.
1834 .. opcode:: INTERP_CENTROID - Interpolate at the centroid
1836 Interpolates the varying specified by src0 at the centroid
1838 .. opcode:: INTERP_SAMPLE - Interpolate at the specified sample
1840 Interpolates the varying specified by src0 at the sample id specified by
1841 src1.x (interpreted as an integer)
1843 .. opcode:: INTERP_OFFSET - Interpolate at the specified offset
1845 Interpolates the varying specified by src0 at the offset src1.xy from the
1846 pixel center (interpreted as floats)
1854 The double-precision opcodes reinterpret four-component vectors into
1855 two-component vectors with doubled precision in each component.
1857 Support for these opcodes is XXX undecided. :T
1859 .. opcode:: DADD - Add
1863 dst.xy = src0.xy + src1.xy
1865 dst.zw = src0.zw + src1.zw
1868 .. opcode:: DDIV - Divide
1872 dst.xy = src0.xy / src1.xy
1874 dst.zw = src0.zw / src1.zw
1876 .. opcode:: DSEQ - Set on Equal
1880 dst.xy = src0.xy == src1.xy ? 1.0F : 0.0F
1882 dst.zw = src0.zw == src1.zw ? 1.0F : 0.0F
1884 .. opcode:: DSLT - Set on Less than
1888 dst.xy = src0.xy < src1.xy ? 1.0F : 0.0F
1890 dst.zw = src0.zw < src1.zw ? 1.0F : 0.0F
1892 .. opcode:: DFRAC - Fraction
1896 dst.xy = src.xy - \lfloor src.xy\rfloor
1898 dst.zw = src.zw - \lfloor src.zw\rfloor
1901 .. opcode:: DFRACEXP - Convert Number to Fractional and Integral Components
1903 Like the ``frexp()`` routine in many math libraries, this opcode stores the
1904 exponent of its source to ``dst0``, and the significand to ``dst1``, such that
1905 :math:`dst1 \times 2^{dst0} = src` .
1909 dst0.xy = exp(src.xy)
1911 dst1.xy = frac(src.xy)
1913 dst0.zw = exp(src.zw)
1915 dst1.zw = frac(src.zw)
1917 .. opcode:: DLDEXP - Multiply Number by Integral Power of 2
1919 This opcode is the inverse of :opcode:`DFRACEXP`.
1923 dst.xy = src0.xy \times 2^{src1.xy}
1925 dst.zw = src0.zw \times 2^{src1.zw}
1927 .. opcode:: DMIN - Minimum
1931 dst.xy = min(src0.xy, src1.xy)
1933 dst.zw = min(src0.zw, src1.zw)
1935 .. opcode:: DMAX - Maximum
1939 dst.xy = max(src0.xy, src1.xy)
1941 dst.zw = max(src0.zw, src1.zw)
1943 .. opcode:: DMUL - Multiply
1947 dst.xy = src0.xy \times src1.xy
1949 dst.zw = src0.zw \times src1.zw
1952 .. opcode:: DMAD - Multiply And Add
1956 dst.xy = src0.xy \times src1.xy + src2.xy
1958 dst.zw = src0.zw \times src1.zw + src2.zw
1961 .. opcode:: DRCP - Reciprocal
1965 dst.xy = \frac{1}{src.xy}
1967 dst.zw = \frac{1}{src.zw}
1969 .. opcode:: DSQRT - Square Root
1973 dst.xy = \sqrt{src.xy}
1975 dst.zw = \sqrt{src.zw}
1978 .. _samplingopcodes:
1980 Resource Sampling Opcodes
1981 ^^^^^^^^^^^^^^^^^^^^^^^^^
1983 Those opcodes follow very closely semantics of the respective Direct3D
1984 instructions. If in doubt double check Direct3D documentation.
1985 Note that the swizzle on SVIEW (src1) determines texel swizzling
1990 Using provided address, sample data from the specified texture using the
1991 filtering mode identified by the gven sampler. The source data may come from
1992 any resource type other than buffers.
1994 Syntax: ``SAMPLE dst, address, sampler_view, sampler``
1996 Example: ``SAMPLE TEMP[0], TEMP[1], SVIEW[0], SAMP[0]``
1998 .. opcode:: SAMPLE_I
2000 Simplified alternative to the SAMPLE instruction. Using the provided
2001 integer address, SAMPLE_I fetches data from the specified sampler view
2002 without any filtering. The source data may come from any resource type
2005 Syntax: ``SAMPLE_I dst, address, sampler_view``
2007 Example: ``SAMPLE_I TEMP[0], TEMP[1], SVIEW[0]``
2009 The 'address' is specified as unsigned integers. If the 'address' is out of
2010 range [0...(# texels - 1)] the result of the fetch is always 0 in all
2011 components. As such the instruction doesn't honor address wrap modes, in
2012 cases where that behavior is desirable 'SAMPLE' instruction should be used.
2013 address.w always provides an unsigned integer mipmap level. If the value is
2014 out of the range then the instruction always returns 0 in all components.
2015 address.yz are ignored for buffers and 1d textures. address.z is ignored
2016 for 1d texture arrays and 2d textures.
2018 For 1D texture arrays address.y provides the array index (also as unsigned
2019 integer). If the value is out of the range of available array indices
2020 [0... (array size - 1)] then the opcode always returns 0 in all components.
2021 For 2D texture arrays address.z provides the array index, otherwise it
2022 exhibits the same behavior as in the case for 1D texture arrays. The exact
2023 semantics of the source address are presented in the table below:
2025 +---------------------------+----+-----+-----+---------+
2026 | resource type | X | Y | Z | W |
2027 +===========================+====+=====+=====+=========+
2028 | ``PIPE_BUFFER`` | x | | | ignored |
2029 +---------------------------+----+-----+-----+---------+
2030 | ``PIPE_TEXTURE_1D`` | x | | | mpl |
2031 +---------------------------+----+-----+-----+---------+
2032 | ``PIPE_TEXTURE_2D`` | x | y | | mpl |
2033 +---------------------------+----+-----+-----+---------+
2034 | ``PIPE_TEXTURE_3D`` | x | y | z | mpl |
2035 +---------------------------+----+-----+-----+---------+
2036 | ``PIPE_TEXTURE_RECT`` | x | y | | mpl |
2037 +---------------------------+----+-----+-----+---------+
2038 | ``PIPE_TEXTURE_CUBE`` | not allowed as source |
2039 +---------------------------+----+-----+-----+---------+
2040 | ``PIPE_TEXTURE_1D_ARRAY`` | x | idx | | mpl |
2041 +---------------------------+----+-----+-----+---------+
2042 | ``PIPE_TEXTURE_2D_ARRAY`` | x | y | idx | mpl |
2043 +---------------------------+----+-----+-----+---------+
2045 Where 'mpl' is a mipmap level and 'idx' is the array index.
2047 .. opcode:: SAMPLE_I_MS
2049 Just like SAMPLE_I but allows fetch data from multi-sampled surfaces.
2051 Syntax: ``SAMPLE_I_MS dst, address, sampler_view, sample``
2053 .. opcode:: SAMPLE_B
2055 Just like the SAMPLE instruction with the exception that an additional bias
2056 is applied to the level of detail computed as part of the instruction
2059 Syntax: ``SAMPLE_B dst, address, sampler_view, sampler, lod_bias``
2061 Example: ``SAMPLE_B TEMP[0], TEMP[1], SVIEW[0], SAMP[0], TEMP[2].x``
2063 .. opcode:: SAMPLE_C
2065 Similar to the SAMPLE instruction but it performs a comparison filter. The
2066 operands to SAMPLE_C are identical to SAMPLE, except that there is an
2067 additional float32 operand, reference value, which must be a register with
2068 single-component, or a scalar literal. SAMPLE_C makes the hardware use the
2069 current samplers compare_func (in pipe_sampler_state) to compare reference
2070 value against the red component value for the surce resource at each texel
2071 that the currently configured texture filter covers based on the provided
2074 Syntax: ``SAMPLE_C dst, address, sampler_view.r, sampler, ref_value``
2076 Example: ``SAMPLE_C TEMP[0], TEMP[1], SVIEW[0].r, SAMP[0], TEMP[2].x``
2078 .. opcode:: SAMPLE_C_LZ
2080 Same as SAMPLE_C, but LOD is 0 and derivatives are ignored. The LZ stands
2083 Syntax: ``SAMPLE_C_LZ dst, address, sampler_view.r, sampler, ref_value``
2085 Example: ``SAMPLE_C_LZ TEMP[0], TEMP[1], SVIEW[0].r, SAMP[0], TEMP[2].x``
2088 .. opcode:: SAMPLE_D
2090 SAMPLE_D is identical to the SAMPLE opcode except that the derivatives for
2091 the source address in the x direction and the y direction are provided by
2094 Syntax: ``SAMPLE_D dst, address, sampler_view, sampler, der_x, der_y``
2096 Example: ``SAMPLE_D TEMP[0], TEMP[1], SVIEW[0], SAMP[0], TEMP[2], TEMP[3]``
2098 .. opcode:: SAMPLE_L
2100 SAMPLE_L is identical to the SAMPLE opcode except that the LOD is provided
2101 directly as a scalar value, representing no anisotropy.
2103 Syntax: ``SAMPLE_L dst, address, sampler_view, sampler, explicit_lod``
2105 Example: ``SAMPLE_L TEMP[0], TEMP[1], SVIEW[0], SAMP[0], TEMP[2].x``
2109 Gathers the four texels to be used in a bi-linear filtering operation and
2110 packs them into a single register. Only works with 2D, 2D array, cubemaps,
2111 and cubemaps arrays. For 2D textures, only the addressing modes of the
2112 sampler and the top level of any mip pyramid are used. Set W to zero. It
2113 behaves like the SAMPLE instruction, but a filtered sample is not
2114 generated. The four samples that contribute to filtering are placed into
2115 xyzw in counter-clockwise order, starting with the (u,v) texture coordinate
2116 delta at the following locations (-, +), (+, +), (+, -), (-, -), where the
2117 magnitude of the deltas are half a texel.
2120 .. opcode:: SVIEWINFO
2122 Query the dimensions of a given sampler view. dst receives width, height,
2123 depth or array size and number of mipmap levels as int4. The dst can have a
2124 writemask which will specify what info is the caller interested in.
2126 Syntax: ``SVIEWINFO dst, src_mip_level, sampler_view``
2128 Example: ``SVIEWINFO TEMP[0], TEMP[1].x, SVIEW[0]``
2130 src_mip_level is an unsigned integer scalar. If it's out of range then
2131 returns 0 for width, height and depth/array size but the total number of
2132 mipmap is still returned correctly for the given sampler view. The returned
2133 width, height and depth values are for the mipmap level selected by the
2134 src_mip_level and are in the number of texels. For 1d texture array width
2135 is in dst.x, array size is in dst.y and dst.z is 0. The number of mipmaps is
2136 still in dst.w. In contrast to d3d10 resinfo, there's no way in the tgsi
2137 instruction encoding to specify the return type (float/rcpfloat/uint), hence
2138 always using uint. Also, unlike the SAMPLE instructions, the swizzle on src1
2139 resinfo allowing swizzling dst values is ignored (due to the interaction
2140 with rcpfloat modifier which requires some swizzle handling in the state
2143 .. opcode:: SAMPLE_POS
2145 Query the position of a given sample. dst receives float4 (x, y, 0, 0)
2146 indicated where the sample is located. If the resource is not a multi-sample
2147 resource and not a render target, the result is 0.
2149 .. opcode:: SAMPLE_INFO
2151 dst receives number of samples in x. If the resource is not a multi-sample
2152 resource and not a render target, the result is 0.
2155 .. _resourceopcodes:
2157 Resource Access Opcodes
2158 ^^^^^^^^^^^^^^^^^^^^^^^
2160 .. opcode:: LOAD - Fetch data from a shader resource
2162 Syntax: ``LOAD dst, resource, address``
2164 Example: ``LOAD TEMP[0], RES[0], TEMP[1]``
2166 Using the provided integer address, LOAD fetches data
2167 from the specified buffer or texture without any
2170 The 'address' is specified as a vector of unsigned
2171 integers. If the 'address' is out of range the result
2174 Only the first mipmap level of a resource can be read
2175 from using this instruction.
2177 For 1D or 2D texture arrays, the array index is
2178 provided as an unsigned integer in address.y or
2179 address.z, respectively. address.yz are ignored for
2180 buffers and 1D textures. address.z is ignored for 1D
2181 texture arrays and 2D textures. address.w is always
2184 .. opcode:: STORE - Write data to a shader resource
2186 Syntax: ``STORE resource, address, src``
2188 Example: ``STORE RES[0], TEMP[0], TEMP[1]``
2190 Using the provided integer address, STORE writes data
2191 to the specified buffer or texture.
2193 The 'address' is specified as a vector of unsigned
2194 integers. If the 'address' is out of range the result
2197 Only the first mipmap level of a resource can be
2198 written to using this instruction.
2200 For 1D or 2D texture arrays, the array index is
2201 provided as an unsigned integer in address.y or
2202 address.z, respectively. address.yz are ignored for
2203 buffers and 1D textures. address.z is ignored for 1D
2204 texture arrays and 2D textures. address.w is always
2208 .. _threadsyncopcodes:
2210 Inter-thread synchronization opcodes
2211 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2213 These opcodes are intended for communication between threads running
2214 within the same compute grid. For now they're only valid in compute
2217 .. opcode:: MFENCE - Memory fence
2219 Syntax: ``MFENCE resource``
2221 Example: ``MFENCE RES[0]``
2223 This opcode forces strong ordering between any memory access
2224 operations that affect the specified resource. This means that
2225 previous loads and stores (and only those) will be performed and
2226 visible to other threads before the program execution continues.
2229 .. opcode:: LFENCE - Load memory fence
2231 Syntax: ``LFENCE resource``
2233 Example: ``LFENCE RES[0]``
2235 Similar to MFENCE, but it only affects the ordering of memory loads.
2238 .. opcode:: SFENCE - Store memory fence
2240 Syntax: ``SFENCE resource``
2242 Example: ``SFENCE RES[0]``
2244 Similar to MFENCE, but it only affects the ordering of memory stores.
2247 .. opcode:: BARRIER - Thread group barrier
2251 This opcode suspends the execution of the current thread until all
2252 the remaining threads in the working group reach the same point of
2253 the program. Results are unspecified if any of the remaining
2254 threads terminates or never reaches an executed BARRIER instruction.
2262 These opcodes provide atomic variants of some common arithmetic and
2263 logical operations. In this context atomicity means that another
2264 concurrent memory access operation that affects the same memory
2265 location is guaranteed to be performed strictly before or after the
2266 entire execution of the atomic operation.
2268 For the moment they're only valid in compute programs.
2270 .. opcode:: ATOMUADD - Atomic integer addition
2272 Syntax: ``ATOMUADD dst, resource, offset, src``
2274 Example: ``ATOMUADD TEMP[0], RES[0], TEMP[1], TEMP[2]``
2276 The following operation is performed atomically on each component:
2280 dst_i = resource[offset]_i
2282 resource[offset]_i = dst_i + src_i
2285 .. opcode:: ATOMXCHG - Atomic exchange
2287 Syntax: ``ATOMXCHG dst, resource, offset, src``
2289 Example: ``ATOMXCHG TEMP[0], RES[0], TEMP[1], TEMP[2]``
2291 The following operation is performed atomically on each component:
2295 dst_i = resource[offset]_i
2297 resource[offset]_i = src_i
2300 .. opcode:: ATOMCAS - Atomic compare-and-exchange
2302 Syntax: ``ATOMCAS dst, resource, offset, cmp, src``
2304 Example: ``ATOMCAS TEMP[0], RES[0], TEMP[1], TEMP[2], TEMP[3]``
2306 The following operation is performed atomically on each component:
2310 dst_i = resource[offset]_i
2312 resource[offset]_i = (dst_i == cmp_i ? src_i : dst_i)
2315 .. opcode:: ATOMAND - Atomic bitwise And
2317 Syntax: ``ATOMAND dst, resource, offset, src``
2319 Example: ``ATOMAND TEMP[0], RES[0], TEMP[1], TEMP[2]``
2321 The following operation is performed atomically on each component:
2325 dst_i = resource[offset]_i
2327 resource[offset]_i = dst_i \& src_i
2330 .. opcode:: ATOMOR - Atomic bitwise Or
2332 Syntax: ``ATOMOR dst, resource, offset, src``
2334 Example: ``ATOMOR TEMP[0], RES[0], TEMP[1], TEMP[2]``
2336 The following operation is performed atomically on each component:
2340 dst_i = resource[offset]_i
2342 resource[offset]_i = dst_i | src_i
2345 .. opcode:: ATOMXOR - Atomic bitwise Xor
2347 Syntax: ``ATOMXOR dst, resource, offset, src``
2349 Example: ``ATOMXOR TEMP[0], RES[0], TEMP[1], TEMP[2]``
2351 The following operation is performed atomically on each component:
2355 dst_i = resource[offset]_i
2357 resource[offset]_i = dst_i \oplus src_i
2360 .. opcode:: ATOMUMIN - Atomic unsigned minimum
2362 Syntax: ``ATOMUMIN dst, resource, offset, src``
2364 Example: ``ATOMUMIN TEMP[0], RES[0], TEMP[1], TEMP[2]``
2366 The following operation is performed atomically on each component:
2370 dst_i = resource[offset]_i
2372 resource[offset]_i = (dst_i < src_i ? dst_i : src_i)
2375 .. opcode:: ATOMUMAX - Atomic unsigned maximum
2377 Syntax: ``ATOMUMAX dst, resource, offset, src``
2379 Example: ``ATOMUMAX TEMP[0], RES[0], TEMP[1], TEMP[2]``
2381 The following operation is performed atomically on each component:
2385 dst_i = resource[offset]_i
2387 resource[offset]_i = (dst_i > src_i ? dst_i : src_i)
2390 .. opcode:: ATOMIMIN - Atomic signed minimum
2392 Syntax: ``ATOMIMIN dst, resource, offset, src``
2394 Example: ``ATOMIMIN TEMP[0], RES[0], TEMP[1], TEMP[2]``
2396 The following operation is performed atomically on each component:
2400 dst_i = resource[offset]_i
2402 resource[offset]_i = (dst_i < src_i ? dst_i : src_i)
2405 .. opcode:: ATOMIMAX - Atomic signed maximum
2407 Syntax: ``ATOMIMAX dst, resource, offset, src``
2409 Example: ``ATOMIMAX TEMP[0], RES[0], TEMP[1], TEMP[2]``
2411 The following operation is performed atomically on each component:
2415 dst_i = resource[offset]_i
2417 resource[offset]_i = (dst_i > src_i ? dst_i : src_i)
2421 Explanation of symbols used
2422 ------------------------------
2429 :math:`|x|` Absolute value of `x`.
2431 :math:`\lceil x \rceil` Ceiling of `x`.
2433 clamp(x,y,z) Clamp x between y and z.
2434 (x < y) ? y : (x > z) ? z : x
2436 :math:`\lfloor x\rfloor` Floor of `x`.
2438 :math:`\log_2{x}` Logarithm of `x`, base 2.
2440 max(x,y) Maximum of x and y.
2443 min(x,y) Minimum of x and y.
2446 partialx(x) Derivative of x relative to fragment's X.
2448 partialy(x) Derivative of x relative to fragment's Y.
2450 pop() Pop from stack.
2452 :math:`x^y` `x` to the power `y`.
2454 push(x) Push x on stack.
2458 trunc(x) Truncate x, i.e. drop the fraction bits.
2465 discard Discard fragment.
2469 target Label of target instruction.
2480 Declares a register that is will be referenced as an operand in Instruction
2483 File field contains register file that is being declared and is one
2486 UsageMask field specifies which of the register components can be accessed
2487 and is one of TGSI_WRITEMASK.
2489 The Local flag specifies that a given value isn't intended for
2490 subroutine parameter passing and, as a result, the implementation
2491 isn't required to give any guarantees of it being preserved across
2492 subroutine boundaries. As it's merely a compiler hint, the
2493 implementation is free to ignore it.
2495 If Dimension flag is set to 1, a Declaration Dimension token follows.
2497 If Semantic flag is set to 1, a Declaration Semantic token follows.
2499 If Interpolate flag is set to 1, a Declaration Interpolate token follows.
2501 If file is TGSI_FILE_RESOURCE, a Declaration Resource token follows.
2503 If Array flag is set to 1, a Declaration Array token follows.
2506 ^^^^^^^^^^^^^^^^^^^^^^^^
2508 Declarations can optional have an ArrayID attribute which can be referred by
2509 indirect addressing operands. An ArrayID of zero is reserved and treaded as
2510 if no ArrayID is specified.
2512 If an indirect addressing operand refers to a specific declaration by using
2513 an ArrayID only the registers in this declaration are guaranteed to be
2514 accessed, accessing any register outside this declaration results in undefined
2515 behavior. Note that for compatibility the effective index is zero-based and
2516 not relative to the specified declaration
2518 If no ArrayID is specified with an indirect addressing operand the whole
2519 register file might be accessed by this operand. This is strongly discouraged
2520 and will prevent packing of scalar/vec2 arrays and effective alias analysis.
2522 Declaration Semantic
2523 ^^^^^^^^^^^^^^^^^^^^^^^^
2525 Vertex and fragment shader input and output registers may be labeled
2526 with semantic information consisting of a name and index.
2528 Follows Declaration token if Semantic bit is set.
2530 Since its purpose is to link a shader with other stages of the pipeline,
2531 it is valid to follow only those Declaration tokens that declare a register
2532 either in INPUT or OUTPUT file.
2534 SemanticName field contains the semantic name of the register being declared.
2535 There is no default value.
2537 SemanticIndex is an optional subscript that can be used to distinguish
2538 different register declarations with the same semantic name. The default value
2541 The meanings of the individual semantic names are explained in the following
2544 TGSI_SEMANTIC_POSITION
2545 """"""""""""""""""""""
2547 For vertex shaders, TGSI_SEMANTIC_POSITION indicates the vertex shader
2548 output register which contains the homogeneous vertex position in the clip
2549 space coordinate system. After clipping, the X, Y and Z components of the
2550 vertex will be divided by the W value to get normalized device coordinates.
2552 For fragment shaders, TGSI_SEMANTIC_POSITION is used to indicate that
2553 fragment shader input contains the fragment's window position. The X
2554 component starts at zero and always increases from left to right.
2555 The Y component starts at zero and always increases but Y=0 may either
2556 indicate the top of the window or the bottom depending on the fragment
2557 coordinate origin convention (see TGSI_PROPERTY_FS_COORD_ORIGIN).
2558 The Z coordinate ranges from 0 to 1 to represent depth from the front
2559 to the back of the Z buffer. The W component contains the reciprocol
2560 of the interpolated vertex position W component.
2562 Fragment shaders may also declare an output register with
2563 TGSI_SEMANTIC_POSITION. Only the Z component is writable. This allows
2564 the fragment shader to change the fragment's Z position.
2571 For vertex shader outputs or fragment shader inputs/outputs, this
2572 label indicates that the resister contains an R,G,B,A color.
2574 Several shader inputs/outputs may contain colors so the semantic index
2575 is used to distinguish them. For example, color[0] may be the diffuse
2576 color while color[1] may be the specular color.
2578 This label is needed so that the flat/smooth shading can be applied
2579 to the right interpolants during rasterization.
2583 TGSI_SEMANTIC_BCOLOR
2584 """"""""""""""""""""
2586 Back-facing colors are only used for back-facing polygons, and are only valid
2587 in vertex shader outputs. After rasterization, all polygons are front-facing
2588 and COLOR and BCOLOR end up occupying the same slots in the fragment shader,
2589 so all BCOLORs effectively become regular COLORs in the fragment shader.
2595 Vertex shader inputs and outputs and fragment shader inputs may be
2596 labeled with TGSI_SEMANTIC_FOG to indicate that the register contains
2597 a fog coordinate. Typically, the fragment shader will use the fog coordinate
2598 to compute a fog blend factor which is used to blend the normal fragment color
2599 with a constant fog color. But fog coord really is just an ordinary vec4
2600 register like regular semantics.
2606 Vertex shader input and output registers may be labeled with
2607 TGIS_SEMANTIC_PSIZE to indicate that the register contains a point size
2608 in the form (S, 0, 0, 1). The point size controls the width or diameter
2609 of points for rasterization. This label cannot be used in fragment
2612 When using this semantic, be sure to set the appropriate state in the
2613 :ref:`rasterizer` first.
2616 TGSI_SEMANTIC_TEXCOORD
2617 """"""""""""""""""""""
2619 Only available if PIPE_CAP_TGSI_TEXCOORD is exposed !
2621 Vertex shader outputs and fragment shader inputs may be labeled with
2622 this semantic to make them replaceable by sprite coordinates via the
2623 sprite_coord_enable state in the :ref:`rasterizer`.
2624 The semantic index permitted with this semantic is limited to <= 7.
2626 If the driver does not support TEXCOORD, sprite coordinate replacement
2627 applies to inputs with the GENERIC semantic instead.
2629 The intended use case for this semantic is gl_TexCoord.
2632 TGSI_SEMANTIC_PCOORD
2633 """"""""""""""""""""
2635 Only available if PIPE_CAP_TGSI_TEXCOORD is exposed !
2637 Fragment shader inputs may be labeled with TGSI_SEMANTIC_PCOORD to indicate
2638 that the register contains sprite coordinates in the form (x, y, 0, 1), if
2639 the current primitive is a point and point sprites are enabled. Otherwise,
2640 the contents of the register are undefined.
2642 The intended use case for this semantic is gl_PointCoord.
2645 TGSI_SEMANTIC_GENERIC
2646 """""""""""""""""""""
2648 All vertex/fragment shader inputs/outputs not labeled with any other
2649 semantic label can be considered to be generic attributes. Typical
2650 uses of generic inputs/outputs are texcoords and user-defined values.
2653 TGSI_SEMANTIC_NORMAL
2654 """"""""""""""""""""
2656 Indicates that a vertex shader input is a normal vector. This is
2657 typically only used for legacy graphics APIs.
2663 This label applies to fragment shader inputs only and indicates that
2664 the register contains front/back-face information of the form (F, 0,
2665 0, 1). The first component will be positive when the fragment belongs
2666 to a front-facing polygon, and negative when the fragment belongs to a
2667 back-facing polygon.
2670 TGSI_SEMANTIC_EDGEFLAG
2671 """"""""""""""""""""""
2673 For vertex shaders, this sematic label indicates that an input or
2674 output is a boolean edge flag. The register layout is [F, x, x, x]
2675 where F is 0.0 or 1.0 and x = don't care. Normally, the vertex shader
2676 simply copies the edge flag input to the edgeflag output.
2678 Edge flags are used to control which lines or points are actually
2679 drawn when the polygon mode converts triangles/quads/polygons into
2683 TGSI_SEMANTIC_STENCIL
2684 """""""""""""""""""""
2686 For fragment shaders, this semantic label indicates that an output
2687 is a writable stencil reference value. Only the Y component is writable.
2688 This allows the fragment shader to change the fragments stencilref value.
2691 TGSI_SEMANTIC_VIEWPORT_INDEX
2692 """"""""""""""""""""""""""""
2694 For geometry shaders, this semantic label indicates that an output
2695 contains the index of the viewport (and scissor) to use.
2696 Only the X value is used.
2702 For geometry shaders, this semantic label indicates that an output
2703 contains the layer value to use for the color and depth/stencil surfaces.
2704 Only the X value is used. (Also known as rendertarget array index.)
2707 TGSI_SEMANTIC_CULLDIST
2708 """"""""""""""""""""""
2710 Used as distance to plane for performing application-defined culling
2711 of individual primitives against a plane. When components of vertex
2712 elements are given this label, these values are assumed to be a
2713 float32 signed distance to a plane. Primitives will be completely
2714 discarded if the plane distance for all of the vertices in the
2715 primitive are < 0. If a vertex has a cull distance of NaN, that
2716 vertex counts as "out" (as if its < 0);
2717 The limits on both clip and cull distances are bound
2718 by the PIPE_MAX_CLIP_OR_CULL_DISTANCE_COUNT define which defines
2719 the maximum number of components that can be used to hold the
2720 distances and by the PIPE_MAX_CLIP_OR_CULL_DISTANCE_ELEMENT_COUNT
2721 which specifies the maximum number of registers which can be
2722 annotated with those semantics.
2725 TGSI_SEMANTIC_CLIPDIST
2726 """"""""""""""""""""""
2728 When components of vertex elements are identified this way, these
2729 values are each assumed to be a float32 signed distance to a plane.
2730 Primitive setup only invokes rasterization on pixels for which
2731 the interpolated plane distances are >= 0. Multiple clip planes
2732 can be implemented simultaneously, by annotating multiple
2733 components of one or more vertex elements with the above specified
2734 semantic. The limits on both clip and cull distances are bound
2735 by the PIPE_MAX_CLIP_OR_CULL_DISTANCE_COUNT define which defines
2736 the maximum number of components that can be used to hold the
2737 distances and by the PIPE_MAX_CLIP_OR_CULL_DISTANCE_ELEMENT_COUNT
2738 which specifies the maximum number of registers which can be
2739 annotated with those semantics.
2741 TGSI_SEMANTIC_SAMPLEID
2742 """"""""""""""""""""""
2744 For fragment shaders, this semantic label indicates that a system value
2745 contains the current sample id (i.e. gl_SampleID). Only the X value is used.
2747 TGSI_SEMANTIC_SAMPLEPOS
2748 """""""""""""""""""""""
2750 For fragment shaders, this semantic label indicates that a system value
2751 contains the current sample's position (i.e. gl_SamplePosition). Only the X
2752 and Y values are used.
2754 TGSI_SEMANTIC_SAMPLEMASK
2755 """"""""""""""""""""""""
2757 For fragment shaders, this semantic label indicates that an output contains
2758 the sample mask used to disable further sample processing
2759 (i.e. gl_SampleMask). Only the X value is used, up to 32x MS.
2761 TGSI_SEMANTIC_INVOCATIONID
2762 """"""""""""""""""""""""""
2764 For geometry shaders, this semantic label indicates that a system value
2765 contains the current invocation id (i.e. gl_InvocationID). Only the X value is
2768 Declaration Interpolate
2769 ^^^^^^^^^^^^^^^^^^^^^^^
2771 This token is only valid for fragment shader INPUT declarations.
2773 The Interpolate field specifes the way input is being interpolated by
2774 the rasteriser and is one of TGSI_INTERPOLATE_*.
2776 The Location field specifies the location inside the pixel that the
2777 interpolation should be done at, one of ``TGSI_INTERPOLATE_LOC_*``. Note that
2778 when per-sample shading is enabled, the implementation may choose to
2779 interpolate at the sample irrespective of the Location field.
2781 The CylindricalWrap bitfield specifies which register components
2782 should be subject to cylindrical wrapping when interpolating by the
2783 rasteriser. If TGSI_CYLINDRICAL_WRAP_X is set to 1, the X component
2784 should be interpolated according to cylindrical wrapping rules.
2787 Declaration Sampler View
2788 ^^^^^^^^^^^^^^^^^^^^^^^^
2790 Follows Declaration token if file is TGSI_FILE_SAMPLER_VIEW.
2792 DCL SVIEW[#], resource, type(s)
2794 Declares a shader input sampler view and assigns it to a SVIEW[#]
2797 resource can be one of BUFFER, 1D, 2D, 3D, 1DArray and 2DArray.
2799 type must be 1 or 4 entries (if specifying on a per-component
2800 level) out of UNORM, SNORM, SINT, UINT and FLOAT.
2803 Declaration Resource
2804 ^^^^^^^^^^^^^^^^^^^^
2806 Follows Declaration token if file is TGSI_FILE_RESOURCE.
2808 DCL RES[#], resource [, WR] [, RAW]
2810 Declares a shader input resource and assigns it to a RES[#]
2813 resource can be one of BUFFER, 1D, 2D, 3D, CUBE, 1DArray and
2816 If the RAW keyword is not specified, the texture data will be
2817 subject to conversion, swizzling and scaling as required to yield
2818 the specified data type from the physical data format of the bound
2821 If the RAW keyword is specified, no channel conversion will be
2822 performed: the values read for each of the channels (X,Y,Z,W) will
2823 correspond to consecutive words in the same order and format
2824 they're found in memory. No element-to-address conversion will be
2825 performed either: the value of the provided X coordinate will be
2826 interpreted in byte units instead of texel units. The result of
2827 accessing a misaligned address is undefined.
2829 Usage of the STORE opcode is only allowed if the WR (writable) flag
2834 ^^^^^^^^^^^^^^^^^^^^^^^^
2836 Properties are general directives that apply to the whole TGSI program.
2841 Specifies the fragment shader TGSI_SEMANTIC_POSITION coordinate origin.
2842 The default value is UPPER_LEFT.
2844 If UPPER_LEFT, the position will be (0,0) at the upper left corner and
2845 increase downward and rightward.
2846 If LOWER_LEFT, the position will be (0,0) at the lower left corner and
2847 increase upward and rightward.
2849 OpenGL defaults to LOWER_LEFT, and is configurable with the
2850 GL_ARB_fragment_coord_conventions extension.
2852 DirectX 9/10 use UPPER_LEFT.
2854 FS_COORD_PIXEL_CENTER
2855 """""""""""""""""""""
2857 Specifies the fragment shader TGSI_SEMANTIC_POSITION pixel center convention.
2858 The default value is HALF_INTEGER.
2860 If HALF_INTEGER, the fractionary part of the position will be 0.5
2861 If INTEGER, the fractionary part of the position will be 0.0
2863 Note that this does not affect the set of fragments generated by
2864 rasterization, which is instead controlled by half_pixel_center in the
2867 OpenGL defaults to HALF_INTEGER, and is configurable with the
2868 GL_ARB_fragment_coord_conventions extension.
2870 DirectX 9 uses INTEGER.
2871 DirectX 10 uses HALF_INTEGER.
2873 FS_COLOR0_WRITES_ALL_CBUFS
2874 """"""""""""""""""""""""""
2875 Specifies that writes to the fragment shader color 0 are replicated to all
2876 bound cbufs. This facilitates OpenGL's fragColor output vs fragData[0] where
2877 fragData is directed to a single color buffer, but fragColor is broadcast.
2880 """"""""""""""""""""""""""
2881 If this property is set on the program bound to the shader stage before the
2882 fragment shader, user clip planes should have no effect (be disabled) even if
2883 that shader does not write to any clip distance outputs and the rasterizer's
2884 clip_plane_enable is non-zero.
2885 This property is only supported by drivers that also support shader clip
2887 This is useful for APIs that don't have UCPs and where clip distances written
2888 by a shader cannot be disabled.
2893 Specifies the number of times a geometry shader should be executed for each
2894 input primitive. Each invocation will have a different
2895 TGSI_SEMANTIC_INVOCATIONID system value set. If not specified, assumed to
2898 VS_WINDOW_SPACE_POSITION
2899 """"""""""""""""""""""""""
2900 If this property is set on the vertex shader, the TGSI_SEMANTIC_POSITION output
2901 is assumed to contain window space coordinates.
2902 Division of X,Y,Z by W and the viewport transformation are disabled, and 1/W is
2903 directly taken from the 4-th component of the shader output.
2904 Naturally, clipping is not performed on window coordinates either.
2905 The effect of this property is undefined if a geometry or tessellation shader
2908 Texture Sampling and Texture Formats
2909 ------------------------------------
2911 This table shows how texture image components are returned as (x,y,z,w) tuples
2912 by TGSI texture instructions, such as :opcode:`TEX`, :opcode:`TXD`, and
2913 :opcode:`TXP`. For reference, OpenGL and Direct3D conventions are shown as
2916 +--------------------+--------------+--------------------+--------------+
2917 | Texture Components | Gallium | OpenGL | Direct3D 9 |
2918 +====================+==============+====================+==============+
2919 | R | (r, 0, 0, 1) | (r, 0, 0, 1) | (r, 1, 1, 1) |
2920 +--------------------+--------------+--------------------+--------------+
2921 | RG | (r, g, 0, 1) | (r, g, 0, 1) | (r, g, 1, 1) |
2922 +--------------------+--------------+--------------------+--------------+
2923 | RGB | (r, g, b, 1) | (r, g, b, 1) | (r, g, b, 1) |
2924 +--------------------+--------------+--------------------+--------------+
2925 | RGBA | (r, g, b, a) | (r, g, b, a) | (r, g, b, a) |
2926 +--------------------+--------------+--------------------+--------------+
2927 | A | (0, 0, 0, a) | (0, 0, 0, a) | (0, 0, 0, a) |
2928 +--------------------+--------------+--------------------+--------------+
2929 | L | (l, l, l, 1) | (l, l, l, 1) | (l, l, l, 1) |
2930 +--------------------+--------------+--------------------+--------------+
2931 | LA | (l, l, l, a) | (l, l, l, a) | (l, l, l, a) |
2932 +--------------------+--------------+--------------------+--------------+
2933 | I | (i, i, i, i) | (i, i, i, i) | N/A |
2934 +--------------------+--------------+--------------------+--------------+
2935 | UV | XXX TBD | (0, 0, 0, 1) | (u, v, 1, 1) |
2936 | | | [#envmap-bumpmap]_ | |
2937 +--------------------+--------------+--------------------+--------------+
2938 | Z | XXX TBD | (z, z, z, 1) | (0, z, 0, 1) |
2939 | | | [#depth-tex-mode]_ | |
2940 +--------------------+--------------+--------------------+--------------+
2941 | S | (s, s, s, s) | unknown | unknown |
2942 +--------------------+--------------+--------------------+--------------+
2944 .. [#envmap-bumpmap] http://www.opengl.org/registry/specs/ATI/envmap_bumpmap.txt
2945 .. [#depth-tex-mode] the default is (z, z, z, 1) but may also be (0, 0, 0, z)
2946 or (z, z, z, z) depending on the value of GL_DEPTH_TEXTURE_MODE.