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:`Double Opcodes`.
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.
30 ^^^^^^^^^^^^^^^^^^^^^^^^^
32 These opcodes are guaranteed to be available regardless of the driver being
35 .. opcode:: ARL - Address Register Load
39 dst.x = \lfloor src.x\rfloor
41 dst.y = \lfloor src.y\rfloor
43 dst.z = \lfloor src.z\rfloor
45 dst.w = \lfloor src.w\rfloor
48 .. opcode:: MOV - Move
61 .. opcode:: LIT - Light Coefficients
69 dst.z = (src.x > 0) ? max(src.y, 0)^{clamp(src.w, -128, 128))} : 0
74 .. opcode:: RCP - Reciprocal
76 This instruction replicates its result.
83 .. opcode:: RSQ - Reciprocal Square Root
85 This instruction replicates its result.
89 dst = \frac{1}{\sqrt{|src.x|}}
92 .. opcode:: SQRT - Square Root
94 This instruction replicates its result.
101 .. opcode:: EXP - Approximate Exponential Base 2
105 dst.x = 2^{\lfloor src.x\rfloor}
107 dst.y = src.x - \lfloor src.x\rfloor
114 .. opcode:: LOG - Approximate Logarithm Base 2
118 dst.x = \lfloor\log_2{|src.x|}\rfloor
120 dst.y = \frac{|src.x|}{2^{\lfloor\log_2{|src.x|}\rfloor}}
122 dst.z = \log_2{|src.x|}
127 .. opcode:: MUL - Multiply
131 dst.x = src0.x \times src1.x
133 dst.y = src0.y \times src1.y
135 dst.z = src0.z \times src1.z
137 dst.w = src0.w \times src1.w
140 .. opcode:: ADD - Add
144 dst.x = src0.x + src1.x
146 dst.y = src0.y + src1.y
148 dst.z = src0.z + src1.z
150 dst.w = src0.w + src1.w
153 .. opcode:: DP3 - 3-component Dot Product
155 This instruction replicates its result.
159 dst = src0.x \times src1.x + src0.y \times src1.y + src0.z \times src1.z
162 .. opcode:: DP4 - 4-component Dot Product
164 This instruction replicates its result.
168 dst = src0.x \times src1.x + src0.y \times src1.y + src0.z \times src1.z + src0.w \times src1.w
171 .. opcode:: DST - Distance Vector
177 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 : 0
216 dst.y = (src0.y < src1.y) ? 1 : 0
218 dst.z = (src0.z < src1.z) ? 1 : 0
220 dst.w = (src0.w < src1.w) ? 1 : 0
223 .. opcode:: SGE - Set On Greater Equal Than
227 dst.x = (src0.x >= src1.x) ? 1 : 0
229 dst.y = (src0.y >= src1.y) ? 1 : 0
231 dst.z = (src0.z >= src1.z) ? 1 : 0
233 dst.w = (src0.w >= src1.w) ? 1 : 0
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:: RCC - Reciprocal Clamped
409 This instruction replicates its result.
411 XXX cleanup on aisle three
415 dst = (1 / src.x) > 0 ? clamp(1 / src.x, 5.42101e-020, 1.884467e+019) : clamp(1 / src.x, -1.884467e+019, -5.42101e-020)
418 .. opcode:: DPH - Homogeneous Dot Product
420 This instruction replicates its result.
424 dst = src0.x \times src1.x + src0.y \times src1.y + src0.z \times src1.z + src1.w
427 .. opcode:: COS - Cosine
429 This instruction replicates its result.
436 .. opcode:: DDX - Derivative Relative To X
440 dst.x = partialx(src.x)
442 dst.y = partialx(src.y)
444 dst.z = partialx(src.z)
446 dst.w = partialx(src.w)
449 .. opcode:: DDY - Derivative Relative To Y
453 dst.x = partialy(src.x)
455 dst.y = partialy(src.y)
457 dst.z = partialy(src.z)
459 dst.w = partialy(src.w)
462 .. opcode:: KILP - Predicated Discard
467 .. opcode:: PK2H - Pack Two 16-bit Floats
472 .. opcode:: PK2US - Pack Two Unsigned 16-bit Scalars
477 .. opcode:: PK4B - Pack Four Signed 8-bit Scalars
482 .. opcode:: PK4UB - Pack Four Unsigned 8-bit Scalars
487 .. opcode:: RFL - Reflection Vector
491 dst.x = 2 \times (src0.x \times src1.x + src0.y \times src1.y + src0.z \times src1.z) / (src0.x \times src0.x + src0.y \times src0.y + src0.z \times src0.z) \times src0.x - src1.x
493 dst.y = 2 \times (src0.x \times src1.x + src0.y \times src1.y + src0.z \times src1.z) / (src0.x \times src0.x + src0.y \times src0.y + src0.z \times src0.z) \times src0.y - src1.y
495 dst.z = 2 \times (src0.x \times src1.x + src0.y \times src1.y + src0.z \times src1.z) / (src0.x \times src0.x + src0.y \times src0.y + src0.z \times src0.z) \times src0.z - src1.z
501 Considered for removal.
504 .. opcode:: SEQ - Set On Equal
508 dst.x = (src0.x == src1.x) ? 1 : 0
510 dst.y = (src0.y == src1.y) ? 1 : 0
512 dst.z = (src0.z == src1.z) ? 1 : 0
514 dst.w = (src0.w == src1.w) ? 1 : 0
517 .. opcode:: SFL - Set On False
519 This instruction replicates its result.
527 Considered for removal.
530 .. opcode:: SGT - Set On Greater Than
534 dst.x = (src0.x > src1.x) ? 1 : 0
536 dst.y = (src0.y > src1.y) ? 1 : 0
538 dst.z = (src0.z > src1.z) ? 1 : 0
540 dst.w = (src0.w > src1.w) ? 1 : 0
543 .. opcode:: SIN - Sine
545 This instruction replicates its result.
552 .. opcode:: SLE - Set On Less Equal Than
556 dst.x = (src0.x <= src1.x) ? 1 : 0
558 dst.y = (src0.y <= src1.y) ? 1 : 0
560 dst.z = (src0.z <= src1.z) ? 1 : 0
562 dst.w = (src0.w <= src1.w) ? 1 : 0
565 .. opcode:: SNE - Set On Not Equal
569 dst.x = (src0.x != src1.x) ? 1 : 0
571 dst.y = (src0.y != src1.y) ? 1 : 0
573 dst.z = (src0.z != src1.z) ? 1 : 0
575 dst.w = (src0.w != src1.w) ? 1 : 0
578 .. opcode:: STR - Set On True
580 This instruction replicates its result.
587 .. opcode:: TEX - Texture Lookup
595 dst = texture_sample(unit, coord, bias)
597 for array textures src0.y contains the slice for 1D,
598 and src0.z contain the slice for 2D.
599 for shadow textures with no arrays, src0.z contains
601 for shadow textures with arrays, src0.z contains
602 the reference value for 1D arrays, and src0.w contains
603 the reference value for 2D arrays.
604 There is no way to pass a bias in the .w value for
605 shadow arrays, and GLSL doesn't allow this.
606 GLSL does allow cube shadows maps to take a bias value,
607 and we have to determine how this will look in TGSI.
609 .. opcode:: TXD - Texture Lookup with Derivatives
621 dst = texture_sample_deriv(unit, coord, bias, ddx, ddy)
624 .. opcode:: TXP - Projective Texture Lookup
628 coord.x = src0.x / src.w
630 coord.y = src0.y / src.w
632 coord.z = src0.z / src.w
638 dst = texture_sample(unit, coord, bias)
641 .. opcode:: UP2H - Unpack Two 16-Bit Floats
647 Considered for removal.
649 .. opcode:: UP2US - Unpack Two Unsigned 16-Bit Scalars
655 Considered for removal.
657 .. opcode:: UP4B - Unpack Four Signed 8-Bit Values
663 Considered for removal.
665 .. opcode:: UP4UB - Unpack Four Unsigned 8-Bit Scalars
671 Considered for removal.
673 .. opcode:: X2D - 2D Coordinate Transformation
677 dst.x = src0.x + src1.x \times src2.x + src1.y \times src2.y
679 dst.y = src0.y + src1.x \times src2.z + src1.y \times src2.w
681 dst.z = src0.x + src1.x \times src2.x + src1.y \times src2.y
683 dst.w = src0.y + src1.x \times src2.z + src1.y \times src2.w
687 Considered for removal.
690 .. opcode:: ARA - Address Register Add
696 Considered for removal.
698 .. opcode:: ARR - Address Register Load With Round
711 .. opcode:: BRA - Branch
717 Considered for removal.
719 .. opcode:: CAL - Subroutine Call
725 .. opcode:: RET - Subroutine Call Return
730 .. opcode:: SSG - Set Sign
734 dst.x = (src.x > 0) ? 1 : (src.x < 0) ? -1 : 0
736 dst.y = (src.y > 0) ? 1 : (src.y < 0) ? -1 : 0
738 dst.z = (src.z > 0) ? 1 : (src.z < 0) ? -1 : 0
740 dst.w = (src.w > 0) ? 1 : (src.w < 0) ? -1 : 0
743 .. opcode:: CMP - Compare
747 dst.x = (src0.x < 0) ? src1.x : src2.x
749 dst.y = (src0.y < 0) ? src1.y : src2.y
751 dst.z = (src0.z < 0) ? src1.z : src2.z
753 dst.w = (src0.w < 0) ? src1.w : src2.w
756 .. opcode:: KIL - Conditional Discard
760 if (src.x < 0 || src.y < 0 || src.z < 0 || src.w < 0)
765 .. opcode:: SCS - Sine Cosine
778 .. opcode:: TXB - Texture Lookup With Bias
792 dst = texture_sample(unit, coord, bias)
795 .. opcode:: NRM - 3-component Vector Normalise
799 dst.x = src.x / (src.x \times src.x + src.y \times src.y + src.z \times src.z)
801 dst.y = src.y / (src.x \times src.x + src.y \times src.y + src.z \times src.z)
803 dst.z = src.z / (src.x \times src.x + src.y \times src.y + src.z \times src.z)
808 .. opcode:: DIV - Divide
812 dst.x = \frac{src0.x}{src1.x}
814 dst.y = \frac{src0.y}{src1.y}
816 dst.z = \frac{src0.z}{src1.z}
818 dst.w = \frac{src0.w}{src1.w}
821 .. opcode:: DP2 - 2-component Dot Product
823 This instruction replicates its result.
827 dst = src0.x \times src1.x + src0.y \times src1.y
830 .. opcode:: TXL - Texture Lookup With explicit LOD
844 dst = texture_sample(unit, coord, lod)
847 .. opcode:: BRK - Break
857 .. opcode:: ELSE - Else
862 .. opcode:: ENDIF - End If
867 .. opcode:: PUSHA - Push Address Register On Stack
876 Considered for cleanup.
880 Considered for removal.
882 .. opcode:: POPA - Pop Address Register From Stack
891 Considered for cleanup.
895 Considered for removal.
899 ^^^^^^^^^^^^^^^^^^^^^^^^
901 These opcodes are primarily provided for special-use computational shaders.
902 Support for these opcodes indicated by a special pipe capability bit (TBD).
904 XXX so let's discuss it, yeah?
906 .. opcode:: CEIL - Ceiling
910 dst.x = \lceil src.x\rceil
912 dst.y = \lceil src.y\rceil
914 dst.z = \lceil src.z\rceil
916 dst.w = \lceil src.w\rceil
919 .. opcode:: I2F - Integer To Float
923 dst.x = (float) src.x
925 dst.y = (float) src.y
927 dst.z = (float) src.z
929 dst.w = (float) src.w
932 .. opcode:: NOT - Bitwise Not
945 .. opcode:: TRUNC - Truncate
958 .. opcode:: SHL - Shift Left
962 dst.x = src0.x << src1.x
964 dst.y = src0.y << src1.x
966 dst.z = src0.z << src1.x
968 dst.w = src0.w << src1.x
971 .. opcode:: SHR - Shift Right
975 dst.x = src0.x >> src1.x
977 dst.y = src0.y >> src1.x
979 dst.z = src0.z >> src1.x
981 dst.w = src0.w >> src1.x
984 .. opcode:: AND - Bitwise And
988 dst.x = src0.x & src1.x
990 dst.y = src0.y & src1.y
992 dst.z = src0.z & src1.z
994 dst.w = src0.w & src1.w
997 .. opcode:: OR - Bitwise Or
1001 dst.x = src0.x | src1.x
1003 dst.y = src0.y | src1.y
1005 dst.z = src0.z | src1.z
1007 dst.w = src0.w | src1.w
1010 .. opcode:: MOD - Modulus
1014 dst.x = src0.x \bmod src1.x
1016 dst.y = src0.y \bmod src1.y
1018 dst.z = src0.z \bmod src1.z
1020 dst.w = src0.w \bmod src1.w
1023 .. opcode:: XOR - Bitwise Xor
1027 dst.x = src0.x \oplus src1.x
1029 dst.y = src0.y \oplus src1.y
1031 dst.z = src0.z \oplus src1.z
1033 dst.w = src0.w \oplus src1.w
1036 .. opcode:: UCMP - Integer Conditional Move
1040 dst.x = src0.x ? src1.x : src2.x
1042 dst.y = src0.y ? src1.y : src2.y
1044 dst.z = src0.z ? src1.z : src2.z
1046 dst.w = src0.w ? src1.w : src2.w
1049 .. opcode:: UARL - Integer Address Register Load
1051 Moves the contents of the source register, assumed to be an integer, into the
1052 destination register, which is assumed to be an address (ADDR) register.
1055 .. opcode:: IABS - Integer Absolute Value
1068 .. opcode:: SAD - Sum Of Absolute Differences
1072 dst.x = |src0.x - src1.x| + src2.x
1074 dst.y = |src0.y - src1.y| + src2.y
1076 dst.z = |src0.z - src1.z| + src2.z
1078 dst.w = |src0.w - src1.w| + src2.w
1081 .. opcode:: TXF - Texel Fetch (as per NV_gpu_shader4), extract a single texel
1082 from a specified texture image. The source sampler may
1083 not be a CUBE or SHADOW.
1084 src 0 is a four-component signed integer vector used to
1085 identify the single texel accessed. 3 components + level.
1086 src 1 is a 3 component constant signed integer vector,
1087 with each component only have a range of
1088 -8..+8 (hw only seems to deal with this range, interface
1089 allows for up to unsigned int).
1090 TXF(uint_vec coord, int_vec offset).
1093 .. opcode:: TXQ - Texture Size Query (as per NV_gpu_program4)
1094 retrieve the dimensions of the texture
1095 depending on the target. For 1D (width), 2D/RECT/CUBE
1096 (width, height), 3D (width, height, depth),
1097 1D array (width, layers), 2D array (width, height, layers)
1103 dst.x = texture_width(unit, lod)
1105 dst.y = texture_height(unit, lod)
1107 dst.z = texture_depth(unit, lod)
1110 .. opcode:: CONT - Continue
1116 Support for CONT is determined by a special capability bit,
1117 ``TGSI_CONT_SUPPORTED``. See :ref:`Screen` for more information.
1121 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1123 These opcodes are only supported in geometry shaders; they have no meaning
1124 in any other type of shader.
1126 .. opcode:: EMIT - Emit
1131 .. opcode:: ENDPRIM - End Primitive
1139 These opcodes are part of :term:`GLSL`'s opcode set. Support for these
1140 opcodes is determined by a special capability bit, ``GLSL``.
1142 .. opcode:: BGNLOOP - Begin a Loop
1147 .. opcode:: BGNSUB - Begin Subroutine
1152 .. opcode:: ENDLOOP - End a Loop
1157 .. opcode:: ENDSUB - End Subroutine
1162 .. opcode:: NOP - No Operation
1167 .. opcode:: NRM4 - 4-component Vector Normalise
1169 This instruction replicates its result.
1173 dst = \frac{src.x}{src.x \times src.x + src.y \times src.y + src.z \times src.z + src.w \times src.w}
1181 .. opcode:: CALLNZ - Subroutine Call If Not Zero
1186 .. opcode:: IFC - If
1191 .. opcode:: BREAKC - Break Conditional
1200 The double-precision opcodes reinterpret four-component vectors into
1201 two-component vectors with doubled precision in each component.
1203 Support for these opcodes is XXX undecided. :T
1205 .. opcode:: DADD - Add
1209 dst.xy = src0.xy + src1.xy
1211 dst.zw = src0.zw + src1.zw
1214 .. opcode:: DDIV - Divide
1218 dst.xy = src0.xy / src1.xy
1220 dst.zw = src0.zw / src1.zw
1222 .. opcode:: DSEQ - Set on Equal
1226 dst.xy = src0.xy == src1.xy ? 1.0F : 0.0F
1228 dst.zw = src0.zw == src1.zw ? 1.0F : 0.0F
1230 .. opcode:: DSLT - Set on Less than
1234 dst.xy = src0.xy < src1.xy ? 1.0F : 0.0F
1236 dst.zw = src0.zw < src1.zw ? 1.0F : 0.0F
1238 .. opcode:: DFRAC - Fraction
1242 dst.xy = src.xy - \lfloor src.xy\rfloor
1244 dst.zw = src.zw - \lfloor src.zw\rfloor
1247 .. opcode:: DFRACEXP - Convert Number to Fractional and Integral Components
1249 Like the ``frexp()`` routine in many math libraries, this opcode stores the
1250 exponent of its source to ``dst0``, and the significand to ``dst1``, such that
1251 :math:`dst1 \times 2^{dst0} = src` .
1255 dst0.xy = exp(src.xy)
1257 dst1.xy = frac(src.xy)
1259 dst0.zw = exp(src.zw)
1261 dst1.zw = frac(src.zw)
1263 .. opcode:: DLDEXP - Multiply Number by Integral Power of 2
1265 This opcode is the inverse of :opcode:`DFRACEXP`.
1269 dst.xy = src0.xy \times 2^{src1.xy}
1271 dst.zw = src0.zw \times 2^{src1.zw}
1273 .. opcode:: DMIN - Minimum
1277 dst.xy = min(src0.xy, src1.xy)
1279 dst.zw = min(src0.zw, src1.zw)
1281 .. opcode:: DMAX - Maximum
1285 dst.xy = max(src0.xy, src1.xy)
1287 dst.zw = max(src0.zw, src1.zw)
1289 .. opcode:: DMUL - Multiply
1293 dst.xy = src0.xy \times src1.xy
1295 dst.zw = src0.zw \times src1.zw
1298 .. opcode:: DMAD - Multiply And Add
1302 dst.xy = src0.xy \times src1.xy + src2.xy
1304 dst.zw = src0.zw \times src1.zw + src2.zw
1307 .. opcode:: DRCP - Reciprocal
1311 dst.xy = \frac{1}{src.xy}
1313 dst.zw = \frac{1}{src.zw}
1315 .. opcode:: DSQRT - Square Root
1319 dst.xy = \sqrt{src.xy}
1321 dst.zw = \sqrt{src.zw}
1324 .. _samplingopcodes:
1326 Resource Sampling Opcodes
1327 ^^^^^^^^^^^^^^^^^^^^^^^^^
1329 Those opcodes follow very closely semantics of the respective Direct3D
1330 instructions. If in doubt double check Direct3D documentation.
1332 .. opcode:: SAMPLE - Using provided address, sample data from the
1333 specified texture using the filtering mode identified
1334 by the gven sampler. The source data may come from
1335 any resource type other than buffers.
1336 SAMPLE dst, address, sampler_view, sampler
1338 SAMPLE TEMP[0], TEMP[1], SVIEW[0], SAMP[0]
1340 .. opcode:: SAMPLE_I - Simplified alternative to the SAMPLE instruction.
1341 Using the provided integer address, SAMPLE_I fetches data
1342 from the specified sampler view without any filtering.
1343 The source data may come from any resource type other
1345 SAMPLE_I dst, address, sampler_view
1347 SAMPLE_I TEMP[0], TEMP[1], SVIEW[0]
1348 The 'address' is specified as unsigned integers. If the
1349 'address' is out of range [0...(# texels - 1)] the
1350 result of the fetch is always 0 in all components.
1351 As such the instruction doesn't honor address wrap
1352 modes, in cases where that behavior is desirable
1353 'SAMPLE' instruction should be used.
1354 address.w always provides an unsigned integer mipmap
1355 level. If the value is out of the range then the
1356 instruction always returns 0 in all components.
1357 address.yz are ignored for buffers and 1d textures.
1358 address.z is ignored for 1d texture arrays and 2d
1360 For 1D texture arrays address.y provides the array
1361 index (also as unsigned integer). If the value is
1362 out of the range of available array indices
1363 [0... (array size - 1)] then the opcode always returns
1364 0 in all components.
1365 For 2D texture arrays address.z provides the array
1366 index, otherwise it exhibits the same behavior as in
1367 the case for 1D texture arrays.
1368 The exact semantics of the source address are presented
1370 resource type X Y Z W
1371 ------------- ------------------------
1372 PIPE_BUFFER x ignored
1373 PIPE_TEXTURE_1D x mpl
1374 PIPE_TEXTURE_2D x y mpl
1375 PIPE_TEXTURE_3D x y z mpl
1376 PIPE_TEXTURE_RECT x y mpl
1377 PIPE_TEXTURE_CUBE not allowed as source
1378 PIPE_TEXTURE_1D_ARRAY x idx mpl
1379 PIPE_TEXTURE_2D_ARRAY x y idx mpl
1381 Where 'mpl' is a mipmap level and 'idx' is the
1384 .. opcode:: SAMPLE_I_MS - Just like SAMPLE_I but allows fetch data from
1385 multi-sampled surfaces.
1387 .. opcode:: SAMPLE_B - Just like the SAMPLE instruction with the
1388 exception that an additiona bias is applied to the
1389 level of detail computed as part of the instruction
1391 SAMPLE_B dst, address, sampler_view, sampler, lod_bias
1393 SAMPLE_B TEMP[0], TEMP[1], SVIEW[0], SAMP[0], TEMP[2].x
1395 .. opcode:: SAMPLE_C - Similar to the SAMPLE instruction but it
1396 performs a comparison filter. The operands to SAMPLE_C
1397 are identical to SAMPLE, except that tere is an additional
1398 float32 operand, reference value, which must be a register
1399 with single-component, or a scalar literal.
1400 SAMPLE_C makes the hardware use the current samplers
1401 compare_func (in pipe_sampler_state) to compare
1402 reference value against the red component value for the
1403 surce resource at each texel that the currently configured
1404 texture filter covers based on the provided coordinates.
1405 SAMPLE_C dst, address, sampler_view.r, sampler, ref_value
1407 SAMPLE_C TEMP[0], TEMP[1], SVIEW[0].r, SAMP[0], TEMP[2].x
1409 .. opcode:: SAMPLE_C_LZ - Same as SAMPLE_C, but LOD is 0 and derivatives
1410 are ignored. The LZ stands for level-zero.
1411 SAMPLE_C_LZ dst, address, sampler_view.r, sampler, ref_value
1413 SAMPLE_C_LZ TEMP[0], TEMP[1], SVIEW[0].r, SAMP[0], TEMP[2].x
1416 .. opcode:: SAMPLE_D - SAMPLE_D is identical to the SAMPLE opcode except
1417 that the derivatives for the source address in the x
1418 direction and the y direction are provided by extra
1420 SAMPLE_D dst, address, sampler_view, sampler, der_x, der_y
1422 SAMPLE_D TEMP[0], TEMP[1], SVIEW[0], SAMP[0], TEMP[2], TEMP[3]
1424 .. opcode:: SAMPLE_L - SAMPLE_L is identical to the SAMPLE opcode except
1425 that the LOD is provided directly as a scalar value,
1426 representing no anisotropy.
1427 SAMPLE_L dst, address, sampler_view, sampler, explicit_lod
1429 SAMPLE_L TEMP[0], TEMP[1], SVIEW[0], SAMP[0], TEMP[2].x
1431 .. opcode:: GATHER4 - Gathers the four texels to be used in a bi-linear
1432 filtering operation and packs them into a single register.
1433 Only works with 2D, 2D array, cubemaps, and cubemaps arrays.
1434 For 2D textures, only the addressing modes of the sampler and
1435 the top level of any mip pyramid are used. Set W to zero.
1436 It behaves like the SAMPLE instruction, but a filtered
1437 sample is not generated. The four samples that contribute
1438 to filtering are placed into xyzw in counter-clockwise order,
1439 starting with the (u,v) texture coordinate delta at the
1440 following locations (-, +), (+, +), (+, -), (-, -), where
1441 the magnitude of the deltas are half a texel.
1444 .. opcode:: SVIEWINFO - query the dimensions of a given sampler view.
1445 dst receives width, height, depth or array size and
1446 number of mipmap levels as int4. The dst can have a writemask
1447 which will specify what info is the caller interested
1449 SVIEWINFO dst, src_mip_level, sampler_view
1451 SVIEWINFO TEMP[0], TEMP[1].x, SVIEW[0]
1452 src_mip_level is an unsigned integer scalar. If it's
1453 out of range then returns 0 for width, height and
1454 depth/array size but the total number of mipmap is
1455 still returned correctly for the given sampler view.
1456 The returned width, height and depth values are for
1457 the mipmap level selected by the src_mip_level and
1458 are in the number of texels.
1459 For 1d texture array width is in dst.x, array size
1460 is in dst.y and dst.zw are always 0.
1462 .. opcode:: SAMPLE_POS - query the position of a given sample.
1463 dst receives float4 (x, y, 0, 0) indicated where the
1464 sample is located. If the resource is not a multi-sample
1465 resource and not a render target, the result is 0.
1467 .. opcode:: SAMPLE_INFO - dst receives number of samples in x.
1468 If the resource is not a multi-sample resource and
1469 not a render target, the result is 0.
1472 .. _resourceopcodes:
1474 Resource Access Opcodes
1475 ^^^^^^^^^^^^^^^^^^^^^^^
1477 .. opcode:: LOAD - Fetch data from a shader resource
1479 Syntax: ``LOAD dst, resource, address``
1481 Example: ``LOAD TEMP[0], RES[0], TEMP[1]``
1483 Using the provided integer address, LOAD fetches data
1484 from the specified buffer or texture without any
1487 The 'address' is specified as a vector of unsigned
1488 integers. If the 'address' is out of range the result
1491 Only the first mipmap level of a resource can be read
1492 from using this instruction.
1494 For 1D or 2D texture arrays, the array index is
1495 provided as an unsigned integer in address.y or
1496 address.z, respectively. address.yz are ignored for
1497 buffers and 1D textures. address.z is ignored for 1D
1498 texture arrays and 2D textures. address.w is always
1501 .. opcode:: STORE - Write data to a shader resource
1503 Syntax: ``STORE resource, address, src``
1505 Example: ``STORE RES[0], TEMP[0], TEMP[1]``
1507 Using the provided integer address, STORE writes data
1508 to the specified buffer or texture.
1510 The 'address' is specified as a vector of unsigned
1511 integers. If the 'address' is out of range the result
1514 Only the first mipmap level of a resource can be
1515 written to using this instruction.
1517 For 1D or 2D texture arrays, the array index is
1518 provided as an unsigned integer in address.y or
1519 address.z, respectively. address.yz are ignored for
1520 buffers and 1D textures. address.z is ignored for 1D
1521 texture arrays and 2D textures. address.w is always
1525 .. _threadsyncopcodes:
1527 Inter-thread synchronization opcodes
1528 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1530 These opcodes are intended for communication between threads running
1531 within the same compute grid. For now they're only valid in compute
1534 .. opcode:: MFENCE - Memory fence
1536 Syntax: ``MFENCE resource``
1538 Example: ``MFENCE RES[0]``
1540 This opcode forces strong ordering between any memory access
1541 operations that affect the specified resource. This means that
1542 previous loads and stores (and only those) will be performed and
1543 visible to other threads before the program execution continues.
1546 .. opcode:: LFENCE - Load memory fence
1548 Syntax: ``LFENCE resource``
1550 Example: ``LFENCE RES[0]``
1552 Similar to MFENCE, but it only affects the ordering of memory loads.
1555 .. opcode:: SFENCE - Store memory fence
1557 Syntax: ``SFENCE resource``
1559 Example: ``SFENCE RES[0]``
1561 Similar to MFENCE, but it only affects the ordering of memory stores.
1564 .. opcode:: BARRIER - Thread group barrier
1568 This opcode suspends the execution of the current thread until all
1569 the remaining threads in the working group reach the same point of
1570 the program. Results are unspecified if any of the remaining
1571 threads terminates or never reaches an executed BARRIER instruction.
1579 These opcodes provide atomic variants of some common arithmetic and
1580 logical operations. In this context atomicity means that another
1581 concurrent memory access operation that affects the same memory
1582 location is guaranteed to be performed strictly before or after the
1583 entire execution of the atomic operation.
1585 For the moment they're only valid in compute programs.
1587 .. opcode:: ATOMUADD - Atomic integer addition
1589 Syntax: ``ATOMUADD dst, resource, offset, src``
1591 Example: ``ATOMUADD TEMP[0], RES[0], TEMP[1], TEMP[2]``
1593 The following operation is performed atomically on each component:
1597 dst_i = resource[offset]_i
1599 resource[offset]_i = dst_i + src_i
1602 .. opcode:: ATOMXCHG - Atomic exchange
1604 Syntax: ``ATOMXCHG dst, resource, offset, src``
1606 Example: ``ATOMXCHG TEMP[0], RES[0], TEMP[1], TEMP[2]``
1608 The following operation is performed atomically on each component:
1612 dst_i = resource[offset]_i
1614 resource[offset]_i = src_i
1617 .. opcode:: ATOMCAS - Atomic compare-and-exchange
1619 Syntax: ``ATOMCAS dst, resource, offset, cmp, src``
1621 Example: ``ATOMCAS TEMP[0], RES[0], TEMP[1], TEMP[2], TEMP[3]``
1623 The following operation is performed atomically on each component:
1627 dst_i = resource[offset]_i
1629 resource[offset]_i = (dst_i == cmp_i ? src_i : dst_i)
1632 .. opcode:: ATOMAND - Atomic bitwise And
1634 Syntax: ``ATOMAND dst, resource, offset, src``
1636 Example: ``ATOMAND TEMP[0], RES[0], TEMP[1], TEMP[2]``
1638 The following operation is performed atomically on each component:
1642 dst_i = resource[offset]_i
1644 resource[offset]_i = dst_i \& src_i
1647 .. opcode:: ATOMOR - Atomic bitwise Or
1649 Syntax: ``ATOMOR dst, resource, offset, src``
1651 Example: ``ATOMOR TEMP[0], RES[0], TEMP[1], TEMP[2]``
1653 The following operation is performed atomically on each component:
1657 dst_i = resource[offset]_i
1659 resource[offset]_i = dst_i | src_i
1662 .. opcode:: ATOMXOR - Atomic bitwise Xor
1664 Syntax: ``ATOMXOR dst, resource, offset, src``
1666 Example: ``ATOMXOR TEMP[0], RES[0], TEMP[1], TEMP[2]``
1668 The following operation is performed atomically on each component:
1672 dst_i = resource[offset]_i
1674 resource[offset]_i = dst_i \oplus src_i
1677 .. opcode:: ATOMUMIN - Atomic unsigned minimum
1679 Syntax: ``ATOMUMIN dst, resource, offset, src``
1681 Example: ``ATOMUMIN TEMP[0], RES[0], TEMP[1], TEMP[2]``
1683 The following operation is performed atomically on each component:
1687 dst_i = resource[offset]_i
1689 resource[offset]_i = (dst_i < src_i ? dst_i : src_i)
1692 .. opcode:: ATOMUMAX - Atomic unsigned maximum
1694 Syntax: ``ATOMUMAX dst, resource, offset, src``
1696 Example: ``ATOMUMAX TEMP[0], RES[0], TEMP[1], TEMP[2]``
1698 The following operation is performed atomically on each component:
1702 dst_i = resource[offset]_i
1704 resource[offset]_i = (dst_i > src_i ? dst_i : src_i)
1707 .. opcode:: ATOMIMIN - Atomic signed minimum
1709 Syntax: ``ATOMIMIN dst, resource, offset, src``
1711 Example: ``ATOMIMIN TEMP[0], RES[0], TEMP[1], TEMP[2]``
1713 The following operation is performed atomically on each component:
1717 dst_i = resource[offset]_i
1719 resource[offset]_i = (dst_i < src_i ? dst_i : src_i)
1722 .. opcode:: ATOMIMAX - Atomic signed maximum
1724 Syntax: ``ATOMIMAX dst, resource, offset, src``
1726 Example: ``ATOMIMAX TEMP[0], RES[0], TEMP[1], TEMP[2]``
1728 The following operation is performed atomically on each component:
1732 dst_i = resource[offset]_i
1734 resource[offset]_i = (dst_i > src_i ? dst_i : src_i)
1738 Explanation of symbols used
1739 ------------------------------
1746 :math:`|x|` Absolute value of `x`.
1748 :math:`\lceil x \rceil` Ceiling of `x`.
1750 clamp(x,y,z) Clamp x between y and z.
1751 (x < y) ? y : (x > z) ? z : x
1753 :math:`\lfloor x\rfloor` Floor of `x`.
1755 :math:`\log_2{x}` Logarithm of `x`, base 2.
1757 max(x,y) Maximum of x and y.
1760 min(x,y) Minimum of x and y.
1763 partialx(x) Derivative of x relative to fragment's X.
1765 partialy(x) Derivative of x relative to fragment's Y.
1767 pop() Pop from stack.
1769 :math:`x^y` `x` to the power `y`.
1771 push(x) Push x on stack.
1775 trunc(x) Truncate x, i.e. drop the fraction bits.
1782 discard Discard fragment.
1786 target Label of target instruction.
1797 Declares a register that is will be referenced as an operand in Instruction
1800 File field contains register file that is being declared and is one
1803 UsageMask field specifies which of the register components can be accessed
1804 and is one of TGSI_WRITEMASK.
1806 The Local flag specifies that a given value isn't intended for
1807 subroutine parameter passing and, as a result, the implementation
1808 isn't required to give any guarantees of it being preserved across
1809 subroutine boundaries. As it's merely a compiler hint, the
1810 implementation is free to ignore it.
1812 If Dimension flag is set to 1, a Declaration Dimension token follows.
1814 If Semantic flag is set to 1, a Declaration Semantic token follows.
1816 If Interpolate flag is set to 1, a Declaration Interpolate token follows.
1818 If file is TGSI_FILE_RESOURCE, a Declaration Resource token follows.
1821 Declaration Semantic
1822 ^^^^^^^^^^^^^^^^^^^^^^^^
1824 Vertex and fragment shader input and output registers may be labeled
1825 with semantic information consisting of a name and index.
1827 Follows Declaration token if Semantic bit is set.
1829 Since its purpose is to link a shader with other stages of the pipeline,
1830 it is valid to follow only those Declaration tokens that declare a register
1831 either in INPUT or OUTPUT file.
1833 SemanticName field contains the semantic name of the register being declared.
1834 There is no default value.
1836 SemanticIndex is an optional subscript that can be used to distinguish
1837 different register declarations with the same semantic name. The default value
1840 The meanings of the individual semantic names are explained in the following
1843 TGSI_SEMANTIC_POSITION
1844 """"""""""""""""""""""
1846 For vertex shaders, TGSI_SEMANTIC_POSITION indicates the vertex shader
1847 output register which contains the homogeneous vertex position in the clip
1848 space coordinate system. After clipping, the X, Y and Z components of the
1849 vertex will be divided by the W value to get normalized device coordinates.
1851 For fragment shaders, TGSI_SEMANTIC_POSITION is used to indicate that
1852 fragment shader input contains the fragment's window position. The X
1853 component starts at zero and always increases from left to right.
1854 The Y component starts at zero and always increases but Y=0 may either
1855 indicate the top of the window or the bottom depending on the fragment
1856 coordinate origin convention (see TGSI_PROPERTY_FS_COORD_ORIGIN).
1857 The Z coordinate ranges from 0 to 1 to represent depth from the front
1858 to the back of the Z buffer. The W component contains the reciprocol
1859 of the interpolated vertex position W component.
1861 Fragment shaders may also declare an output register with
1862 TGSI_SEMANTIC_POSITION. Only the Z component is writable. This allows
1863 the fragment shader to change the fragment's Z position.
1870 For vertex shader outputs or fragment shader inputs/outputs, this
1871 label indicates that the resister contains an R,G,B,A color.
1873 Several shader inputs/outputs may contain colors so the semantic index
1874 is used to distinguish them. For example, color[0] may be the diffuse
1875 color while color[1] may be the specular color.
1877 This label is needed so that the flat/smooth shading can be applied
1878 to the right interpolants during rasterization.
1882 TGSI_SEMANTIC_BCOLOR
1883 """"""""""""""""""""
1885 Back-facing colors are only used for back-facing polygons, and are only valid
1886 in vertex shader outputs. After rasterization, all polygons are front-facing
1887 and COLOR and BCOLOR end up occupying the same slots in the fragment shader,
1888 so all BCOLORs effectively become regular COLORs in the fragment shader.
1894 Vertex shader inputs and outputs and fragment shader inputs may be
1895 labeled with TGSI_SEMANTIC_FOG to indicate that the register contains
1896 a fog coordinate in the form (F, 0, 0, 1). Typically, the fragment
1897 shader will use the fog coordinate to compute a fog blend factor which
1898 is used to blend the normal fragment color with a constant fog color.
1900 Only the first component matters when writing from the vertex shader;
1901 the driver will ensure that the coordinate is in this format when used
1902 as a fragment shader input.
1908 Vertex shader input and output registers may be labeled with
1909 TGIS_SEMANTIC_PSIZE to indicate that the register contains a point size
1910 in the form (S, 0, 0, 1). The point size controls the width or diameter
1911 of points for rasterization. This label cannot be used in fragment
1914 When using this semantic, be sure to set the appropriate state in the
1915 :ref:`rasterizer` first.
1918 TGSI_SEMANTIC_GENERIC
1919 """""""""""""""""""""
1921 All vertex/fragment shader inputs/outputs not labeled with any other
1922 semantic label can be considered to be generic attributes. Typical
1923 uses of generic inputs/outputs are texcoords and user-defined values.
1926 TGSI_SEMANTIC_NORMAL
1927 """"""""""""""""""""
1929 Indicates that a vertex shader input is a normal vector. This is
1930 typically only used for legacy graphics APIs.
1936 This label applies to fragment shader inputs only and indicates that
1937 the register contains front/back-face information of the form (F, 0,
1938 0, 1). The first component will be positive when the fragment belongs
1939 to a front-facing polygon, and negative when the fragment belongs to a
1940 back-facing polygon.
1943 TGSI_SEMANTIC_EDGEFLAG
1944 """"""""""""""""""""""
1946 For vertex shaders, this sematic label indicates that an input or
1947 output is a boolean edge flag. The register layout is [F, x, x, x]
1948 where F is 0.0 or 1.0 and x = don't care. Normally, the vertex shader
1949 simply copies the edge flag input to the edgeflag output.
1951 Edge flags are used to control which lines or points are actually
1952 drawn when the polygon mode converts triangles/quads/polygons into
1955 TGSI_SEMANTIC_STENCIL
1956 """"""""""""""""""""""
1958 For fragment shaders, this semantic label indicates than an output
1959 is a writable stencil reference value. Only the Y component is writable.
1960 This allows the fragment shader to change the fragments stencilref value.
1963 Declaration Interpolate
1964 ^^^^^^^^^^^^^^^^^^^^^^^
1966 This token is only valid for fragment shader INPUT declarations.
1968 The Interpolate field specifes the way input is being interpolated by
1969 the rasteriser and is one of TGSI_INTERPOLATE_*.
1971 The CylindricalWrap bitfield specifies which register components
1972 should be subject to cylindrical wrapping when interpolating by the
1973 rasteriser. If TGSI_CYLINDRICAL_WRAP_X is set to 1, the X component
1974 should be interpolated according to cylindrical wrapping rules.
1977 Declaration Sampler View
1978 ^^^^^^^^^^^^^^^^^^^^^^^^
1980 Follows Declaration token if file is TGSI_FILE_SAMPLER_VIEW.
1982 DCL SVIEW[#], resource, type(s)
1984 Declares a shader input sampler view and assigns it to a SVIEW[#]
1987 resource can be one of BUFFER, 1D, 2D, 3D, 1DArray and 2DArray.
1989 type must be 1 or 4 entries (if specifying on a per-component
1990 level) out of UNORM, SNORM, SINT, UINT and FLOAT.
1993 Declaration Resource
1994 ^^^^^^^^^^^^^^^^^^^^
1996 Follows Declaration token if file is TGSI_FILE_RESOURCE.
1998 DCL RES[#], resource [, WR] [, RAW]
2000 Declares a shader input resource and assigns it to a RES[#]
2003 resource can be one of BUFFER, 1D, 2D, 3D, CUBE, 1DArray and
2006 If the RAW keyword is not specified, the texture data will be
2007 subject to conversion, swizzling and scaling as required to yield
2008 the specified data type from the physical data format of the bound
2011 If the RAW keyword is specified, no channel conversion will be
2012 performed: the values read for each of the channels (X,Y,Z,W) will
2013 correspond to consecutive words in the same order and format
2014 they're found in memory. No element-to-address conversion will be
2015 performed either: the value of the provided X coordinate will be
2016 interpreted in byte units instead of texel units. The result of
2017 accessing a misaligned address is undefined.
2019 Usage of the STORE opcode is only allowed if the WR (writable) flag
2024 ^^^^^^^^^^^^^^^^^^^^^^^^
2027 Properties are general directives that apply to the whole TGSI program.
2032 Specifies the fragment shader TGSI_SEMANTIC_POSITION coordinate origin.
2033 The default value is UPPER_LEFT.
2035 If UPPER_LEFT, the position will be (0,0) at the upper left corner and
2036 increase downward and rightward.
2037 If LOWER_LEFT, the position will be (0,0) at the lower left corner and
2038 increase upward and rightward.
2040 OpenGL defaults to LOWER_LEFT, and is configurable with the
2041 GL_ARB_fragment_coord_conventions extension.
2043 DirectX 9/10 use UPPER_LEFT.
2045 FS_COORD_PIXEL_CENTER
2046 """""""""""""""""""""
2048 Specifies the fragment shader TGSI_SEMANTIC_POSITION pixel center convention.
2049 The default value is HALF_INTEGER.
2051 If HALF_INTEGER, the fractionary part of the position will be 0.5
2052 If INTEGER, the fractionary part of the position will be 0.0
2054 Note that this does not affect the set of fragments generated by
2055 rasterization, which is instead controlled by gl_rasterization_rules in the
2058 OpenGL defaults to HALF_INTEGER, and is configurable with the
2059 GL_ARB_fragment_coord_conventions extension.
2061 DirectX 9 uses INTEGER.
2062 DirectX 10 uses HALF_INTEGER.
2064 FS_COLOR0_WRITES_ALL_CBUFS
2065 """"""""""""""""""""""""""
2066 Specifies that writes to the fragment shader color 0 are replicated to all
2067 bound cbufs. This facilitates OpenGL's fragColor output vs fragData[0] where
2068 fragData is directed to a single color buffer, but fragColor is broadcast.
2071 """"""""""""""""""""""""""
2072 If this property is set on the program bound to the shader stage before the
2073 fragment shader, user clip planes should have no effect (be disabled) even if
2074 that shader does not write to any clip distance outputs and the rasterizer's
2075 clip_plane_enable is non-zero.
2076 This property is only supported by drivers that also support shader clip
2078 This is useful for APIs that don't have UCPs and where clip distances written
2079 by a shader cannot be disabled.
2082 Texture Sampling and Texture Formats
2083 ------------------------------------
2085 This table shows how texture image components are returned as (x,y,z,w) tuples
2086 by TGSI texture instructions, such as :opcode:`TEX`, :opcode:`TXD`, and
2087 :opcode:`TXP`. For reference, OpenGL and Direct3D conventions are shown as
2090 +--------------------+--------------+--------------------+--------------+
2091 | Texture Components | Gallium | OpenGL | Direct3D 9 |
2092 +====================+==============+====================+==============+
2093 | R | (r, 0, 0, 1) | (r, 0, 0, 1) | (r, 1, 1, 1) |
2094 +--------------------+--------------+--------------------+--------------+
2095 | RG | (r, g, 0, 1) | (r, g, 0, 1) | (r, g, 1, 1) |
2096 +--------------------+--------------+--------------------+--------------+
2097 | RGB | (r, g, b, 1) | (r, g, b, 1) | (r, g, b, 1) |
2098 +--------------------+--------------+--------------------+--------------+
2099 | RGBA | (r, g, b, a) | (r, g, b, a) | (r, g, b, a) |
2100 +--------------------+--------------+--------------------+--------------+
2101 | A | (0, 0, 0, a) | (0, 0, 0, a) | (0, 0, 0, a) |
2102 +--------------------+--------------+--------------------+--------------+
2103 | L | (l, l, l, 1) | (l, l, l, 1) | (l, l, l, 1) |
2104 +--------------------+--------------+--------------------+--------------+
2105 | LA | (l, l, l, a) | (l, l, l, a) | (l, l, l, a) |
2106 +--------------------+--------------+--------------------+--------------+
2107 | I | (i, i, i, i) | (i, i, i, i) | N/A |
2108 +--------------------+--------------+--------------------+--------------+
2109 | UV | XXX TBD | (0, 0, 0, 1) | (u, v, 1, 1) |
2110 | | | [#envmap-bumpmap]_ | |
2111 +--------------------+--------------+--------------------+--------------+
2112 | Z | XXX TBD | (z, z, z, 1) | (0, z, 0, 1) |
2113 | | | [#depth-tex-mode]_ | |
2114 +--------------------+--------------+--------------------+--------------+
2115 | S | (s, s, s, s) | unknown | unknown |
2116 +--------------------+--------------+--------------------+--------------+
2118 .. [#envmap-bumpmap] http://www.opengl.org/registry/specs/ATI/envmap_bumpmap.txt
2119 .. [#depth-tex-mode] the default is (z, z, z, 1) but may also be (0, 0, 0, z)
2120 or (z, z, z, z) depending on the value of GL_DEPTH_TEXTURE_MODE.