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 = (int) \lfloor src.x\rfloor
53 dst.y = (int) \lfloor src.y\rfloor
55 dst.z = (int) \lfloor src.z\rfloor
57 dst.w = (int) \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:: DP2A - 2-component Dot Product And Add
279 dst.x = src0.x \times src1.x + src0.y \times src1.y + src2.x
281 dst.y = src0.x \times src1.x + src0.y \times src1.y + src2.x
283 dst.z = src0.x \times src1.x + src0.y \times src1.y + src2.x
285 dst.w = src0.x \times src1.x + src0.y \times src1.y + src2.x
288 .. opcode:: FRC - Fraction
292 dst.x = src.x - \lfloor src.x\rfloor
294 dst.y = src.y - \lfloor src.y\rfloor
296 dst.z = src.z - \lfloor src.z\rfloor
298 dst.w = src.w - \lfloor src.w\rfloor
301 .. opcode:: CLAMP - Clamp
305 dst.x = clamp(src0.x, src1.x, src2.x)
307 dst.y = clamp(src0.y, src1.y, src2.y)
309 dst.z = clamp(src0.z, src1.z, src2.z)
311 dst.w = clamp(src0.w, src1.w, src2.w)
314 .. opcode:: FLR - Floor
318 dst.x = \lfloor src.x\rfloor
320 dst.y = \lfloor src.y\rfloor
322 dst.z = \lfloor src.z\rfloor
324 dst.w = \lfloor src.w\rfloor
327 .. opcode:: ROUND - Round
340 .. opcode:: EX2 - Exponential Base 2
342 This instruction replicates its result.
349 .. opcode:: LG2 - Logarithm Base 2
351 This instruction replicates its result.
358 .. opcode:: POW - Power
360 This instruction replicates its result.
364 dst = src0.x^{src1.x}
366 .. opcode:: XPD - Cross Product
370 dst.x = src0.y \times src1.z - src1.y \times src0.z
372 dst.y = src0.z \times src1.x - src1.z \times src0.x
374 dst.z = src0.x \times src1.y - src1.x \times src0.y
379 .. opcode:: ABS - Absolute
392 .. opcode:: DPH - Homogeneous Dot Product
394 This instruction replicates its result.
398 dst = src0.x \times src1.x + src0.y \times src1.y + src0.z \times src1.z + src1.w
401 .. opcode:: COS - Cosine
403 This instruction replicates its result.
410 .. opcode:: DDX, DDX_FINE - Derivative Relative To X
412 The fine variant is only used when ``PIPE_CAP_TGSI_FS_FINE_DERIVATIVE`` is
413 advertised. When it is, the fine version guarantees one derivative per row
414 while DDX is allowed to be the same for the entire 2x2 quad.
418 dst.x = partialx(src.x)
420 dst.y = partialx(src.y)
422 dst.z = partialx(src.z)
424 dst.w = partialx(src.w)
427 .. opcode:: DDY, DDY_FINE - Derivative Relative To Y
429 The fine variant is only used when ``PIPE_CAP_TGSI_FS_FINE_DERIVATIVE`` is
430 advertised. When it is, the fine version guarantees one derivative per column
431 while DDY is allowed to be the same for the entire 2x2 quad.
435 dst.x = partialy(src.x)
437 dst.y = partialy(src.y)
439 dst.z = partialy(src.z)
441 dst.w = partialy(src.w)
444 .. opcode:: PK2H - Pack Two 16-bit Floats
449 .. opcode:: PK2US - Pack Two Unsigned 16-bit Scalars
454 .. opcode:: PK4B - Pack Four Signed 8-bit Scalars
459 .. opcode:: PK4UB - Pack Four Unsigned 8-bit Scalars
464 .. opcode:: SEQ - Set On Equal
468 dst.x = (src0.x == src1.x) ? 1.0F : 0.0F
470 dst.y = (src0.y == src1.y) ? 1.0F : 0.0F
472 dst.z = (src0.z == src1.z) ? 1.0F : 0.0F
474 dst.w = (src0.w == src1.w) ? 1.0F : 0.0F
477 .. opcode:: SGT - Set On Greater Than
481 dst.x = (src0.x > src1.x) ? 1.0F : 0.0F
483 dst.y = (src0.y > src1.y) ? 1.0F : 0.0F
485 dst.z = (src0.z > src1.z) ? 1.0F : 0.0F
487 dst.w = (src0.w > src1.w) ? 1.0F : 0.0F
490 .. opcode:: SIN - Sine
492 This instruction replicates its result.
499 .. opcode:: SLE - Set On Less Equal Than
503 dst.x = (src0.x <= src1.x) ? 1.0F : 0.0F
505 dst.y = (src0.y <= src1.y) ? 1.0F : 0.0F
507 dst.z = (src0.z <= src1.z) ? 1.0F : 0.0F
509 dst.w = (src0.w <= src1.w) ? 1.0F : 0.0F
512 .. opcode:: SNE - Set On Not Equal
516 dst.x = (src0.x != src1.x) ? 1.0F : 0.0F
518 dst.y = (src0.y != src1.y) ? 1.0F : 0.0F
520 dst.z = (src0.z != src1.z) ? 1.0F : 0.0F
522 dst.w = (src0.w != src1.w) ? 1.0F : 0.0F
525 .. opcode:: TEX - Texture Lookup
527 for array textures src0.y contains the slice for 1D,
528 and src0.z contain the slice for 2D.
530 for shadow textures with no arrays (and not cube map),
531 src0.z contains the reference value.
533 for shadow textures with arrays, src0.z contains
534 the reference value for 1D arrays, and src0.w contains
535 the reference value for 2D arrays and cube maps.
537 for cube map array shadow textures, the reference value
538 cannot be passed in src0.w, and TEX2 must be used instead.
544 shadow_ref = src0.z or src0.w (optional)
548 dst = texture\_sample(unit, coord, shadow_ref)
551 .. opcode:: TEX2 - Texture Lookup (for shadow cube map arrays only)
553 this is the same as TEX, but uses another reg to encode the
564 dst = texture\_sample(unit, coord, shadow_ref)
569 .. opcode:: TXD - Texture Lookup with Derivatives
581 dst = texture\_sample\_deriv(unit, coord, ddx, ddy)
584 .. opcode:: TXP - Projective Texture Lookup
588 coord.x = src0.x / src0.w
590 coord.y = src0.y / src0.w
592 coord.z = src0.z / src0.w
598 dst = texture\_sample(unit, coord)
601 .. opcode:: UP2H - Unpack Two 16-Bit Floats
607 Considered for removal.
609 .. opcode:: UP2US - Unpack Two Unsigned 16-Bit Scalars
615 Considered for removal.
617 .. opcode:: UP4B - Unpack Four Signed 8-Bit Values
623 Considered for removal.
625 .. opcode:: UP4UB - Unpack Four Unsigned 8-Bit Scalars
631 Considered for removal.
634 .. opcode:: ARR - Address Register Load With Round
638 dst.x = (int) round(src.x)
640 dst.y = (int) round(src.y)
642 dst.z = (int) round(src.z)
644 dst.w = (int) round(src.w)
647 .. opcode:: SSG - Set Sign
651 dst.x = (src.x > 0) ? 1 : (src.x < 0) ? -1 : 0
653 dst.y = (src.y > 0) ? 1 : (src.y < 0) ? -1 : 0
655 dst.z = (src.z > 0) ? 1 : (src.z < 0) ? -1 : 0
657 dst.w = (src.w > 0) ? 1 : (src.w < 0) ? -1 : 0
660 .. opcode:: CMP - Compare
664 dst.x = (src0.x < 0) ? src1.x : src2.x
666 dst.y = (src0.y < 0) ? src1.y : src2.y
668 dst.z = (src0.z < 0) ? src1.z : src2.z
670 dst.w = (src0.w < 0) ? src1.w : src2.w
673 .. opcode:: KILL_IF - Conditional Discard
675 Conditional discard. Allowed in fragment shaders only.
679 if (src.x < 0 || src.y < 0 || src.z < 0 || src.w < 0)
684 .. opcode:: KILL - Discard
686 Unconditional discard. Allowed in fragment shaders only.
689 .. opcode:: SCS - Sine Cosine
702 .. opcode:: TXB - Texture Lookup With Bias
704 for cube map array textures and shadow cube maps, the bias value
705 cannot be passed in src0.w, and TXB2 must be used instead.
707 if the target is a shadow texture, the reference value is always
708 in src.z (this prevents shadow 3d and shadow 2d arrays from
709 using this instruction, but this is not needed).
725 dst = texture\_sample(unit, coord, bias)
728 .. opcode:: TXB2 - Texture Lookup With Bias (some cube maps only)
730 this is the same as TXB, but uses another reg to encode the
731 lod bias value for cube map arrays and shadow cube maps.
732 Presumably shadow 2d arrays and shadow 3d targets could use
733 this encoding too, but this is not legal.
735 shadow cube map arrays are neither possible nor required.
745 dst = texture\_sample(unit, coord, bias)
748 .. opcode:: DIV - Divide
752 dst.x = \frac{src0.x}{src1.x}
754 dst.y = \frac{src0.y}{src1.y}
756 dst.z = \frac{src0.z}{src1.z}
758 dst.w = \frac{src0.w}{src1.w}
761 .. opcode:: DP2 - 2-component Dot Product
763 This instruction replicates its result.
767 dst = src0.x \times src1.x + src0.y \times src1.y
770 .. opcode:: TXL - Texture Lookup With explicit LOD
772 for cube map array textures, the explicit lod value
773 cannot be passed in src0.w, and TXL2 must be used instead.
775 if the target is a shadow texture, the reference value is always
776 in src.z (this prevents shadow 3d / 2d array / cube targets from
777 using this instruction, but this is not needed).
793 dst = texture\_sample(unit, coord, lod)
796 .. opcode:: TXL2 - Texture Lookup With explicit LOD (for cube map arrays only)
798 this is the same as TXL, but uses another reg to encode the
800 Presumably shadow 3d / 2d array / cube targets could use
801 this encoding too, but this is not legal.
803 shadow cube map arrays are neither possible nor required.
813 dst = texture\_sample(unit, coord, lod)
816 .. opcode:: PUSHA - Push Address Register On Stack
825 Considered for cleanup.
829 Considered for removal.
831 .. opcode:: POPA - Pop Address Register From Stack
840 Considered for cleanup.
844 Considered for removal.
847 .. opcode:: CALLNZ - Subroutine Call If Not Zero
853 Considered for cleanup.
857 Considered for removal.
861 ^^^^^^^^^^^^^^^^^^^^^^^^
863 These opcodes are primarily provided for special-use computational shaders.
864 Support for these opcodes indicated by a special pipe capability bit (TBD).
866 XXX doesn't look like most of the opcodes really belong here.
868 .. opcode:: CEIL - Ceiling
872 dst.x = \lceil src.x\rceil
874 dst.y = \lceil src.y\rceil
876 dst.z = \lceil src.z\rceil
878 dst.w = \lceil src.w\rceil
881 .. opcode:: TRUNC - Truncate
894 .. opcode:: MOD - Modulus
898 dst.x = src0.x \bmod src1.x
900 dst.y = src0.y \bmod src1.y
902 dst.z = src0.z \bmod src1.z
904 dst.w = src0.w \bmod src1.w
907 .. opcode:: UARL - Integer Address Register Load
909 Moves the contents of the source register, assumed to be an integer, into the
910 destination register, which is assumed to be an address (ADDR) register.
913 .. opcode:: SAD - Sum Of Absolute Differences
917 dst.x = |src0.x - src1.x| + src2.x
919 dst.y = |src0.y - src1.y| + src2.y
921 dst.z = |src0.z - src1.z| + src2.z
923 dst.w = |src0.w - src1.w| + src2.w
926 .. opcode:: TXF - Texel Fetch
928 As per NV_gpu_shader4, extract a single texel from a specified texture
929 image. The source sampler may not be a CUBE or SHADOW. src 0 is a
930 four-component signed integer vector used to identify the single texel
931 accessed. 3 components + level. Just like texture instructions, an optional
932 offset vector is provided, which is subject to various driver restrictions
933 (regarding range, source of offsets).
934 TXF(uint_vec coord, int_vec offset).
937 .. opcode:: TXQ - Texture Size Query
939 As per NV_gpu_program4, retrieve the dimensions of the texture depending on
940 the target. For 1D (width), 2D/RECT/CUBE (width, height), 3D (width, height,
941 depth), 1D array (width, layers), 2D array (width, height, layers).
942 Also return the number of accessible levels (last_level - first_level + 1)
945 For components which don't return a resource dimension, their value
953 dst.x = texture\_width(unit, lod)
955 dst.y = texture\_height(unit, lod)
957 dst.z = texture\_depth(unit, lod)
959 dst.w = texture\_levels(unit)
961 .. opcode:: TG4 - Texture Gather
963 As per ARB_texture_gather, gathers the four texels to be used in a bi-linear
964 filtering operation and packs them into a single register. Only works with
965 2D, 2D array, cubemaps, and cubemaps arrays. For 2D textures, only the
966 addressing modes of the sampler and the top level of any mip pyramid are
967 used. Set W to zero. It behaves like the TEX instruction, but a filtered
968 sample is not generated. The four samples that contribute to filtering are
969 placed into xyzw in clockwise order, starting with the (u,v) texture
970 coordinate delta at the following locations (-, +), (+, +), (+, -), (-, -),
971 where the magnitude of the deltas are half a texel.
973 PIPE_CAP_TEXTURE_SM5 enhances this instruction to support shadow per-sample
974 depth compares, single component selection, and a non-constant offset. It
975 doesn't allow support for the GL independent offset to get i0,j0. This would
976 require another CAP is hw can do it natively. For now we lower that before
985 dst = texture\_gather4 (unit, coord, component)
987 (with SM5 - cube array shadow)
995 dst = texture\_gather (uint, coord, compare)
997 .. opcode:: LODQ - level of detail query
999 Compute the LOD information that the texture pipe would use to access the
1000 texture. The Y component contains the computed LOD lambda_prime. The X
1001 component contains the LOD that will be accessed, based on min/max lod's
1008 dst.xy = lodq(uint, coord);
1011 ^^^^^^^^^^^^^^^^^^^^^^^^
1012 These opcodes are used for integer operations.
1013 Support for these opcodes indicated by PIPE_SHADER_CAP_INTEGERS (all of them?)
1016 .. opcode:: I2F - Signed Integer To Float
1018 Rounding is unspecified (round to nearest even suggested).
1022 dst.x = (float) src.x
1024 dst.y = (float) src.y
1026 dst.z = (float) src.z
1028 dst.w = (float) src.w
1031 .. opcode:: U2F - Unsigned Integer To Float
1033 Rounding is unspecified (round to nearest even suggested).
1037 dst.x = (float) src.x
1039 dst.y = (float) src.y
1041 dst.z = (float) src.z
1043 dst.w = (float) src.w
1046 .. opcode:: F2I - Float to Signed Integer
1048 Rounding is towards zero (truncate).
1049 Values outside signed range (including NaNs) produce undefined results.
1062 .. opcode:: F2U - Float to Unsigned Integer
1064 Rounding is towards zero (truncate).
1065 Values outside unsigned range (including NaNs) produce undefined results.
1069 dst.x = (unsigned) src.x
1071 dst.y = (unsigned) src.y
1073 dst.z = (unsigned) src.z
1075 dst.w = (unsigned) src.w
1078 .. opcode:: UADD - Integer Add
1080 This instruction works the same for signed and unsigned integers.
1081 The low 32bit of the result is returned.
1085 dst.x = src0.x + src1.x
1087 dst.y = src0.y + src1.y
1089 dst.z = src0.z + src1.z
1091 dst.w = src0.w + src1.w
1094 .. opcode:: UMAD - Integer Multiply And Add
1096 This instruction works the same for signed and unsigned integers.
1097 The multiplication returns the low 32bit (as does the result itself).
1101 dst.x = src0.x \times src1.x + src2.x
1103 dst.y = src0.y \times src1.y + src2.y
1105 dst.z = src0.z \times src1.z + src2.z
1107 dst.w = src0.w \times src1.w + src2.w
1110 .. opcode:: UMUL - Integer Multiply
1112 This instruction works the same for signed and unsigned integers.
1113 The low 32bit of the result is returned.
1117 dst.x = src0.x \times src1.x
1119 dst.y = src0.y \times src1.y
1121 dst.z = src0.z \times src1.z
1123 dst.w = src0.w \times src1.w
1126 .. opcode:: IMUL_HI - Signed Integer Multiply High Bits
1128 The high 32bits of the multiplication of 2 signed integers are returned.
1132 dst.x = (src0.x \times src1.x) >> 32
1134 dst.y = (src0.y \times src1.y) >> 32
1136 dst.z = (src0.z \times src1.z) >> 32
1138 dst.w = (src0.w \times src1.w) >> 32
1141 .. opcode:: UMUL_HI - Unsigned Integer Multiply High Bits
1143 The high 32bits of the multiplication of 2 unsigned integers are returned.
1147 dst.x = (src0.x \times src1.x) >> 32
1149 dst.y = (src0.y \times src1.y) >> 32
1151 dst.z = (src0.z \times src1.z) >> 32
1153 dst.w = (src0.w \times src1.w) >> 32
1156 .. opcode:: IDIV - Signed Integer Division
1158 TBD: behavior for division by zero.
1162 dst.x = src0.x \ src1.x
1164 dst.y = src0.y \ src1.y
1166 dst.z = src0.z \ src1.z
1168 dst.w = src0.w \ src1.w
1171 .. opcode:: UDIV - Unsigned Integer Division
1173 For division by zero, 0xffffffff is returned.
1177 dst.x = src0.x \ src1.x
1179 dst.y = src0.y \ src1.y
1181 dst.z = src0.z \ src1.z
1183 dst.w = src0.w \ src1.w
1186 .. opcode:: UMOD - Unsigned Integer Remainder
1188 If second arg is zero, 0xffffffff is returned.
1192 dst.x = src0.x \ src1.x
1194 dst.y = src0.y \ src1.y
1196 dst.z = src0.z \ src1.z
1198 dst.w = src0.w \ src1.w
1201 .. opcode:: NOT - Bitwise Not
1214 .. opcode:: AND - Bitwise And
1218 dst.x = src0.x \& src1.x
1220 dst.y = src0.y \& src1.y
1222 dst.z = src0.z \& src1.z
1224 dst.w = src0.w \& src1.w
1227 .. opcode:: OR - Bitwise Or
1231 dst.x = src0.x | src1.x
1233 dst.y = src0.y | src1.y
1235 dst.z = src0.z | src1.z
1237 dst.w = src0.w | src1.w
1240 .. opcode:: XOR - Bitwise Xor
1244 dst.x = src0.x \oplus src1.x
1246 dst.y = src0.y \oplus src1.y
1248 dst.z = src0.z \oplus src1.z
1250 dst.w = src0.w \oplus src1.w
1253 .. opcode:: IMAX - Maximum of Signed Integers
1257 dst.x = max(src0.x, src1.x)
1259 dst.y = max(src0.y, src1.y)
1261 dst.z = max(src0.z, src1.z)
1263 dst.w = max(src0.w, src1.w)
1266 .. opcode:: UMAX - Maximum of Unsigned Integers
1270 dst.x = max(src0.x, src1.x)
1272 dst.y = max(src0.y, src1.y)
1274 dst.z = max(src0.z, src1.z)
1276 dst.w = max(src0.w, src1.w)
1279 .. opcode:: IMIN - Minimum of Signed Integers
1283 dst.x = min(src0.x, src1.x)
1285 dst.y = min(src0.y, src1.y)
1287 dst.z = min(src0.z, src1.z)
1289 dst.w = min(src0.w, src1.w)
1292 .. opcode:: UMIN - Minimum of Unsigned Integers
1296 dst.x = min(src0.x, src1.x)
1298 dst.y = min(src0.y, src1.y)
1300 dst.z = min(src0.z, src1.z)
1302 dst.w = min(src0.w, src1.w)
1305 .. opcode:: SHL - Shift Left
1307 The shift count is masked with 0x1f before the shift is applied.
1311 dst.x = src0.x << (0x1f \& src1.x)
1313 dst.y = src0.y << (0x1f \& src1.y)
1315 dst.z = src0.z << (0x1f \& src1.z)
1317 dst.w = src0.w << (0x1f \& src1.w)
1320 .. opcode:: ISHR - Arithmetic Shift Right (of Signed Integer)
1322 The shift count is masked with 0x1f before the shift is applied.
1326 dst.x = src0.x >> (0x1f \& src1.x)
1328 dst.y = src0.y >> (0x1f \& src1.y)
1330 dst.z = src0.z >> (0x1f \& src1.z)
1332 dst.w = src0.w >> (0x1f \& src1.w)
1335 .. opcode:: USHR - Logical Shift Right
1337 The shift count is masked with 0x1f before the shift is applied.
1341 dst.x = src0.x >> (unsigned) (0x1f \& src1.x)
1343 dst.y = src0.y >> (unsigned) (0x1f \& src1.y)
1345 dst.z = src0.z >> (unsigned) (0x1f \& src1.z)
1347 dst.w = src0.w >> (unsigned) (0x1f \& src1.w)
1350 .. opcode:: UCMP - Integer Conditional Move
1354 dst.x = src0.x ? src1.x : src2.x
1356 dst.y = src0.y ? src1.y : src2.y
1358 dst.z = src0.z ? src1.z : src2.z
1360 dst.w = src0.w ? src1.w : src2.w
1364 .. opcode:: ISSG - Integer Set Sign
1368 dst.x = (src0.x < 0) ? -1 : (src0.x > 0) ? 1 : 0
1370 dst.y = (src0.y < 0) ? -1 : (src0.y > 0) ? 1 : 0
1372 dst.z = (src0.z < 0) ? -1 : (src0.z > 0) ? 1 : 0
1374 dst.w = (src0.w < 0) ? -1 : (src0.w > 0) ? 1 : 0
1378 .. opcode:: FSLT - Float Set On Less Than (ordered)
1380 Same comparison as SLT but returns integer instead of 1.0/0.0 float
1384 dst.x = (src0.x < src1.x) ? \sim 0 : 0
1386 dst.y = (src0.y < src1.y) ? \sim 0 : 0
1388 dst.z = (src0.z < src1.z) ? \sim 0 : 0
1390 dst.w = (src0.w < src1.w) ? \sim 0 : 0
1393 .. opcode:: ISLT - Signed Integer Set On Less Than
1397 dst.x = (src0.x < src1.x) ? \sim 0 : 0
1399 dst.y = (src0.y < src1.y) ? \sim 0 : 0
1401 dst.z = (src0.z < src1.z) ? \sim 0 : 0
1403 dst.w = (src0.w < src1.w) ? \sim 0 : 0
1406 .. opcode:: USLT - Unsigned Integer Set On Less Than
1410 dst.x = (src0.x < src1.x) ? \sim 0 : 0
1412 dst.y = (src0.y < src1.y) ? \sim 0 : 0
1414 dst.z = (src0.z < src1.z) ? \sim 0 : 0
1416 dst.w = (src0.w < src1.w) ? \sim 0 : 0
1419 .. opcode:: FSGE - Float Set On Greater Equal Than (ordered)
1421 Same comparison as SGE but returns integer instead of 1.0/0.0 float
1425 dst.x = (src0.x >= src1.x) ? \sim 0 : 0
1427 dst.y = (src0.y >= src1.y) ? \sim 0 : 0
1429 dst.z = (src0.z >= src1.z) ? \sim 0 : 0
1431 dst.w = (src0.w >= src1.w) ? \sim 0 : 0
1434 .. opcode:: ISGE - Signed Integer Set On Greater Equal Than
1438 dst.x = (src0.x >= src1.x) ? \sim 0 : 0
1440 dst.y = (src0.y >= src1.y) ? \sim 0 : 0
1442 dst.z = (src0.z >= src1.z) ? \sim 0 : 0
1444 dst.w = (src0.w >= src1.w) ? \sim 0 : 0
1447 .. opcode:: USGE - Unsigned Integer Set On Greater Equal Than
1451 dst.x = (src0.x >= src1.x) ? \sim 0 : 0
1453 dst.y = (src0.y >= src1.y) ? \sim 0 : 0
1455 dst.z = (src0.z >= src1.z) ? \sim 0 : 0
1457 dst.w = (src0.w >= src1.w) ? \sim 0 : 0
1460 .. opcode:: FSEQ - Float Set On Equal (ordered)
1462 Same comparison as SEQ but returns integer instead of 1.0/0.0 float
1466 dst.x = (src0.x == src1.x) ? \sim 0 : 0
1468 dst.y = (src0.y == src1.y) ? \sim 0 : 0
1470 dst.z = (src0.z == src1.z) ? \sim 0 : 0
1472 dst.w = (src0.w == src1.w) ? \sim 0 : 0
1475 .. opcode:: USEQ - Integer Set On Equal
1479 dst.x = (src0.x == src1.x) ? \sim 0 : 0
1481 dst.y = (src0.y == src1.y) ? \sim 0 : 0
1483 dst.z = (src0.z == src1.z) ? \sim 0 : 0
1485 dst.w = (src0.w == src1.w) ? \sim 0 : 0
1488 .. opcode:: FSNE - Float Set On Not Equal (unordered)
1490 Same comparison as SNE but returns integer instead of 1.0/0.0 float
1494 dst.x = (src0.x != src1.x) ? \sim 0 : 0
1496 dst.y = (src0.y != src1.y) ? \sim 0 : 0
1498 dst.z = (src0.z != src1.z) ? \sim 0 : 0
1500 dst.w = (src0.w != src1.w) ? \sim 0 : 0
1503 .. opcode:: USNE - Integer Set On Not Equal
1507 dst.x = (src0.x != src1.x) ? \sim 0 : 0
1509 dst.y = (src0.y != src1.y) ? \sim 0 : 0
1511 dst.z = (src0.z != src1.z) ? \sim 0 : 0
1513 dst.w = (src0.w != src1.w) ? \sim 0 : 0
1516 .. opcode:: INEG - Integer Negate
1531 .. opcode:: IABS - Integer Absolute Value
1545 These opcodes are used for bit-level manipulation of integers.
1547 .. opcode:: IBFE - Signed Bitfield Extract
1549 See SM5 instruction of the same name. Extracts a set of bits from the input,
1550 and sign-extends them if the high bit of the extracted window is set.
1554 def ibfe(value, offset, bits):
1555 offset = offset & 0x1f
1557 if bits == 0: return 0
1558 # Note: >> sign-extends
1559 if width + offset < 32:
1560 return (value << (32 - offset - bits)) >> (32 - bits)
1562 return value >> offset
1564 .. opcode:: UBFE - Unsigned Bitfield Extract
1566 See SM5 instruction of the same name. Extracts a set of bits from the input,
1567 without any sign-extension.
1571 def ubfe(value, offset, bits):
1572 offset = offset & 0x1f
1574 if bits == 0: return 0
1575 # Note: >> does not sign-extend
1576 if width + offset < 32:
1577 return (value << (32 - offset - bits)) >> (32 - bits)
1579 return value >> offset
1581 .. opcode:: BFI - Bitfield Insert
1583 See SM5 instruction of the same name. Replaces a bit region of 'base' with
1584 the low bits of 'insert'.
1588 def bfi(base, insert, offset, bits):
1589 offset = offset & 0x1f
1591 mask = ((1 << bits) - 1) << offset
1592 return ((insert << offset) & mask) | (base & ~mask)
1594 .. opcode:: BREV - Bitfield Reverse
1596 See SM5 instruction BFREV. Reverses the bits of the argument.
1598 .. opcode:: POPC - Population Count
1600 See SM5 instruction COUNTBITS. Counts the number of set bits in the argument.
1602 .. opcode:: LSB - Index of lowest set bit
1604 See SM5 instruction FIRSTBIT_LO. Computes the 0-based index of the first set
1605 bit of the argument. Returns -1 if none are set.
1607 .. opcode:: IMSB - Index of highest non-sign bit
1609 See SM5 instruction FIRSTBIT_SHI. Computes the 0-based index of the highest
1610 non-sign bit of the argument (i.e. highest 0 bit for negative numbers,
1611 highest 1 bit for positive numbers). Returns -1 if all bits are the same
1612 (i.e. for inputs 0 and -1).
1614 .. opcode:: UMSB - Index of highest set bit
1616 See SM5 instruction FIRSTBIT_HI. Computes the 0-based index of the highest
1617 set bit of the argument. Returns -1 if none are set.
1620 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1622 These opcodes are only supported in geometry shaders; they have no meaning
1623 in any other type of shader.
1625 .. opcode:: EMIT - Emit
1627 Generate a new vertex for the current primitive into the specified vertex
1628 stream using the values in the output registers.
1631 .. opcode:: ENDPRIM - End Primitive
1633 Complete the current primitive in the specified vertex stream (consisting of
1634 the emitted vertices), and start a new one.
1640 These opcodes are part of :term:`GLSL`'s opcode set. Support for these
1641 opcodes is determined by a special capability bit, ``GLSL``.
1642 Some require glsl version 1.30 (UIF/BREAKC/SWITCH/CASE/DEFAULT/ENDSWITCH).
1644 .. opcode:: CAL - Subroutine Call
1650 .. opcode:: RET - Subroutine Call Return
1655 .. opcode:: CONT - Continue
1657 Unconditionally moves the point of execution to the instruction after the
1658 last bgnloop. The instruction must appear within a bgnloop/endloop.
1662 Support for CONT is determined by a special capability bit,
1663 ``TGSI_CONT_SUPPORTED``. See :ref:`Screen` for more information.
1666 .. opcode:: BGNLOOP - Begin a Loop
1668 Start a loop. Must have a matching endloop.
1671 .. opcode:: BGNSUB - Begin Subroutine
1673 Starts definition of a subroutine. Must have a matching endsub.
1676 .. opcode:: ENDLOOP - End a Loop
1678 End a loop started with bgnloop.
1681 .. opcode:: ENDSUB - End Subroutine
1683 Ends definition of a subroutine.
1686 .. opcode:: NOP - No Operation
1691 .. opcode:: BRK - Break
1693 Unconditionally moves the point of execution to the instruction after the
1694 next endloop or endswitch. The instruction must appear within a loop/endloop
1695 or switch/endswitch.
1698 .. opcode:: BREAKC - Break Conditional
1700 Conditionally moves the point of execution to the instruction after the
1701 next endloop or endswitch. The instruction must appear within a loop/endloop
1702 or switch/endswitch.
1703 Condition evaluates to true if src0.x != 0 where src0.x is interpreted
1704 as an integer register.
1708 Considered for removal as it's quite inconsistent wrt other opcodes
1709 (could emulate with UIF/BRK/ENDIF).
1712 .. opcode:: IF - Float If
1714 Start an IF ... ELSE .. ENDIF block. Condition evaluates to true if
1718 where src0.x is interpreted as a floating point register.
1721 .. opcode:: UIF - Bitwise If
1723 Start an UIF ... ELSE .. ENDIF block. Condition evaluates to true if
1727 where src0.x is interpreted as an integer register.
1730 .. opcode:: ELSE - Else
1732 Starts an else block, after an IF or UIF statement.
1735 .. opcode:: ENDIF - End If
1737 Ends an IF or UIF block.
1740 .. opcode:: SWITCH - Switch
1742 Starts a C-style switch expression. The switch consists of one or multiple
1743 CASE statements, and at most one DEFAULT statement. Execution of a statement
1744 ends when a BRK is hit, but just like in C falling through to other cases
1745 without a break is allowed. Similarly, DEFAULT label is allowed anywhere not
1746 just as last statement, and fallthrough is allowed into/from it.
1747 CASE src arguments are evaluated at bit level against the SWITCH src argument.
1753 (some instructions here)
1756 (some instructions here)
1759 (some instructions here)
1764 .. opcode:: CASE - Switch case
1766 This represents a switch case label. The src arg must be an integer immediate.
1769 .. opcode:: DEFAULT - Switch default
1771 This represents the default case in the switch, which is taken if no other
1775 .. opcode:: ENDSWITCH - End of switch
1777 Ends a switch expression.
1783 The interpolation instructions allow an input to be interpolated in a
1784 different way than its declaration. This corresponds to the GLSL 4.00
1785 interpolateAt* functions. The first argument of each of these must come from
1786 ``TGSI_FILE_INPUT``.
1788 .. opcode:: INTERP_CENTROID - Interpolate at the centroid
1790 Interpolates the varying specified by src0 at the centroid
1792 .. opcode:: INTERP_SAMPLE - Interpolate at the specified sample
1794 Interpolates the varying specified by src0 at the sample id specified by
1795 src1.x (interpreted as an integer)
1797 .. opcode:: INTERP_OFFSET - Interpolate at the specified offset
1799 Interpolates the varying specified by src0 at the offset src1.xy from the
1800 pixel center (interpreted as floats)
1808 The double-precision opcodes reinterpret four-component vectors into
1809 two-component vectors with doubled precision in each component.
1811 Support for these opcodes is XXX undecided. :T
1813 .. opcode:: DADD - Add
1817 dst.xy = src0.xy + src1.xy
1819 dst.zw = src0.zw + src1.zw
1822 .. opcode:: DDIV - Divide
1826 dst.xy = src0.xy / src1.xy
1828 dst.zw = src0.zw / src1.zw
1830 .. opcode:: DSEQ - Set on Equal
1834 dst.xy = src0.xy == src1.xy ? 1.0F : 0.0F
1836 dst.zw = src0.zw == src1.zw ? 1.0F : 0.0F
1838 .. opcode:: DSLT - Set on Less than
1842 dst.xy = src0.xy < src1.xy ? 1.0F : 0.0F
1844 dst.zw = src0.zw < src1.zw ? 1.0F : 0.0F
1846 .. opcode:: DFRAC - Fraction
1850 dst.xy = src.xy - \lfloor src.xy\rfloor
1852 dst.zw = src.zw - \lfloor src.zw\rfloor
1855 .. opcode:: DFRACEXP - Convert Number to Fractional and Integral Components
1857 Like the ``frexp()`` routine in many math libraries, this opcode stores the
1858 exponent of its source to ``dst0``, and the significand to ``dst1``, such that
1859 :math:`dst1 \times 2^{dst0} = src` .
1863 dst0.xy = exp(src.xy)
1865 dst1.xy = frac(src.xy)
1867 dst0.zw = exp(src.zw)
1869 dst1.zw = frac(src.zw)
1871 .. opcode:: DLDEXP - Multiply Number by Integral Power of 2
1873 This opcode is the inverse of :opcode:`DFRACEXP`.
1877 dst.xy = src0.xy \times 2^{src1.xy}
1879 dst.zw = src0.zw \times 2^{src1.zw}
1881 .. opcode:: DMIN - Minimum
1885 dst.xy = min(src0.xy, src1.xy)
1887 dst.zw = min(src0.zw, src1.zw)
1889 .. opcode:: DMAX - Maximum
1893 dst.xy = max(src0.xy, src1.xy)
1895 dst.zw = max(src0.zw, src1.zw)
1897 .. opcode:: DMUL - Multiply
1901 dst.xy = src0.xy \times src1.xy
1903 dst.zw = src0.zw \times src1.zw
1906 .. opcode:: DMAD - Multiply And Add
1910 dst.xy = src0.xy \times src1.xy + src2.xy
1912 dst.zw = src0.zw \times src1.zw + src2.zw
1915 .. opcode:: DRCP - Reciprocal
1919 dst.xy = \frac{1}{src.xy}
1921 dst.zw = \frac{1}{src.zw}
1923 .. opcode:: DSQRT - Square Root
1927 dst.xy = \sqrt{src.xy}
1929 dst.zw = \sqrt{src.zw}
1932 .. _samplingopcodes:
1934 Resource Sampling Opcodes
1935 ^^^^^^^^^^^^^^^^^^^^^^^^^
1937 Those opcodes follow very closely semantics of the respective Direct3D
1938 instructions. If in doubt double check Direct3D documentation.
1939 Note that the swizzle on SVIEW (src1) determines texel swizzling
1944 Using provided address, sample data from the specified texture using the
1945 filtering mode identified by the gven sampler. The source data may come from
1946 any resource type other than buffers.
1948 Syntax: ``SAMPLE dst, address, sampler_view, sampler``
1950 Example: ``SAMPLE TEMP[0], TEMP[1], SVIEW[0], SAMP[0]``
1952 .. opcode:: SAMPLE_I
1954 Simplified alternative to the SAMPLE instruction. Using the provided
1955 integer address, SAMPLE_I fetches data from the specified sampler view
1956 without any filtering. The source data may come from any resource type
1959 Syntax: ``SAMPLE_I dst, address, sampler_view``
1961 Example: ``SAMPLE_I TEMP[0], TEMP[1], SVIEW[0]``
1963 The 'address' is specified as unsigned integers. If the 'address' is out of
1964 range [0...(# texels - 1)] the result of the fetch is always 0 in all
1965 components. As such the instruction doesn't honor address wrap modes, in
1966 cases where that behavior is desirable 'SAMPLE' instruction should be used.
1967 address.w always provides an unsigned integer mipmap level. If the value is
1968 out of the range then the instruction always returns 0 in all components.
1969 address.yz are ignored for buffers and 1d textures. address.z is ignored
1970 for 1d texture arrays and 2d textures.
1972 For 1D texture arrays address.y provides the array index (also as unsigned
1973 integer). If the value is out of the range of available array indices
1974 [0... (array size - 1)] then the opcode always returns 0 in all components.
1975 For 2D texture arrays address.z provides the array index, otherwise it
1976 exhibits the same behavior as in the case for 1D texture arrays. The exact
1977 semantics of the source address are presented in the table below:
1979 +---------------------------+----+-----+-----+---------+
1980 | resource type | X | Y | Z | W |
1981 +===========================+====+=====+=====+=========+
1982 | ``PIPE_BUFFER`` | x | | | ignored |
1983 +---------------------------+----+-----+-----+---------+
1984 | ``PIPE_TEXTURE_1D`` | x | | | mpl |
1985 +---------------------------+----+-----+-----+---------+
1986 | ``PIPE_TEXTURE_2D`` | x | y | | mpl |
1987 +---------------------------+----+-----+-----+---------+
1988 | ``PIPE_TEXTURE_3D`` | x | y | z | mpl |
1989 +---------------------------+----+-----+-----+---------+
1990 | ``PIPE_TEXTURE_RECT`` | x | y | | mpl |
1991 +---------------------------+----+-----+-----+---------+
1992 | ``PIPE_TEXTURE_CUBE`` | not allowed as source |
1993 +---------------------------+----+-----+-----+---------+
1994 | ``PIPE_TEXTURE_1D_ARRAY`` | x | idx | | mpl |
1995 +---------------------------+----+-----+-----+---------+
1996 | ``PIPE_TEXTURE_2D_ARRAY`` | x | y | idx | mpl |
1997 +---------------------------+----+-----+-----+---------+
1999 Where 'mpl' is a mipmap level and 'idx' is the array index.
2001 .. opcode:: SAMPLE_I_MS
2003 Just like SAMPLE_I but allows fetch data from multi-sampled surfaces.
2005 Syntax: ``SAMPLE_I_MS dst, address, sampler_view, sample``
2007 .. opcode:: SAMPLE_B
2009 Just like the SAMPLE instruction with the exception that an additional bias
2010 is applied to the level of detail computed as part of the instruction
2013 Syntax: ``SAMPLE_B dst, address, sampler_view, sampler, lod_bias``
2015 Example: ``SAMPLE_B TEMP[0], TEMP[1], SVIEW[0], SAMP[0], TEMP[2].x``
2017 .. opcode:: SAMPLE_C
2019 Similar to the SAMPLE instruction but it performs a comparison filter. The
2020 operands to SAMPLE_C are identical to SAMPLE, except that there is an
2021 additional float32 operand, reference value, which must be a register with
2022 single-component, or a scalar literal. SAMPLE_C makes the hardware use the
2023 current samplers compare_func (in pipe_sampler_state) to compare reference
2024 value against the red component value for the surce resource at each texel
2025 that the currently configured texture filter covers based on the provided
2028 Syntax: ``SAMPLE_C dst, address, sampler_view.r, sampler, ref_value``
2030 Example: ``SAMPLE_C TEMP[0], TEMP[1], SVIEW[0].r, SAMP[0], TEMP[2].x``
2032 .. opcode:: SAMPLE_C_LZ
2034 Same as SAMPLE_C, but LOD is 0 and derivatives are ignored. The LZ stands
2037 Syntax: ``SAMPLE_C_LZ dst, address, sampler_view.r, sampler, ref_value``
2039 Example: ``SAMPLE_C_LZ TEMP[0], TEMP[1], SVIEW[0].r, SAMP[0], TEMP[2].x``
2042 .. opcode:: SAMPLE_D
2044 SAMPLE_D is identical to the SAMPLE opcode except that the derivatives for
2045 the source address in the x direction and the y direction are provided by
2048 Syntax: ``SAMPLE_D dst, address, sampler_view, sampler, der_x, der_y``
2050 Example: ``SAMPLE_D TEMP[0], TEMP[1], SVIEW[0], SAMP[0], TEMP[2], TEMP[3]``
2052 .. opcode:: SAMPLE_L
2054 SAMPLE_L is identical to the SAMPLE opcode except that the LOD is provided
2055 directly as a scalar value, representing no anisotropy.
2057 Syntax: ``SAMPLE_L dst, address, sampler_view, sampler, explicit_lod``
2059 Example: ``SAMPLE_L TEMP[0], TEMP[1], SVIEW[0], SAMP[0], TEMP[2].x``
2063 Gathers the four texels to be used in a bi-linear filtering operation and
2064 packs them into a single register. Only works with 2D, 2D array, cubemaps,
2065 and cubemaps arrays. For 2D textures, only the addressing modes of the
2066 sampler and the top level of any mip pyramid are used. Set W to zero. It
2067 behaves like the SAMPLE instruction, but a filtered sample is not
2068 generated. The four samples that contribute to filtering are placed into
2069 xyzw in counter-clockwise order, starting with the (u,v) texture coordinate
2070 delta at the following locations (-, +), (+, +), (+, -), (-, -), where the
2071 magnitude of the deltas are half a texel.
2074 .. opcode:: SVIEWINFO
2076 Query the dimensions of a given sampler view. dst receives width, height,
2077 depth or array size and number of mipmap levels as int4. The dst can have a
2078 writemask which will specify what info is the caller interested in.
2080 Syntax: ``SVIEWINFO dst, src_mip_level, sampler_view``
2082 Example: ``SVIEWINFO TEMP[0], TEMP[1].x, SVIEW[0]``
2084 src_mip_level is an unsigned integer scalar. If it's out of range then
2085 returns 0 for width, height and depth/array size but the total number of
2086 mipmap is still returned correctly for the given sampler view. The returned
2087 width, height and depth values are for the mipmap level selected by the
2088 src_mip_level and are in the number of texels. For 1d texture array width
2089 is in dst.x, array size is in dst.y and dst.z is 0. The number of mipmaps is
2090 still in dst.w. In contrast to d3d10 resinfo, there's no way in the tgsi
2091 instruction encoding to specify the return type (float/rcpfloat/uint), hence
2092 always using uint. Also, unlike the SAMPLE instructions, the swizzle on src1
2093 resinfo allowing swizzling dst values is ignored (due to the interaction
2094 with rcpfloat modifier which requires some swizzle handling in the state
2097 .. opcode:: SAMPLE_POS
2099 Query the position of a given sample. dst receives float4 (x, y, 0, 0)
2100 indicated where the sample is located. If the resource is not a multi-sample
2101 resource and not a render target, the result is 0.
2103 .. opcode:: SAMPLE_INFO
2105 dst receives number of samples in x. If the resource is not a multi-sample
2106 resource and not a render target, the result is 0.
2109 .. _resourceopcodes:
2111 Resource Access Opcodes
2112 ^^^^^^^^^^^^^^^^^^^^^^^
2114 .. opcode:: LOAD - Fetch data from a shader resource
2116 Syntax: ``LOAD dst, resource, address``
2118 Example: ``LOAD TEMP[0], RES[0], TEMP[1]``
2120 Using the provided integer address, LOAD fetches data
2121 from the specified buffer or texture without any
2124 The 'address' is specified as a vector of unsigned
2125 integers. If the 'address' is out of range the result
2128 Only the first mipmap level of a resource can be read
2129 from using this instruction.
2131 For 1D or 2D texture arrays, the array index is
2132 provided as an unsigned integer in address.y or
2133 address.z, respectively. address.yz are ignored for
2134 buffers and 1D textures. address.z is ignored for 1D
2135 texture arrays and 2D textures. address.w is always
2138 .. opcode:: STORE - Write data to a shader resource
2140 Syntax: ``STORE resource, address, src``
2142 Example: ``STORE RES[0], TEMP[0], TEMP[1]``
2144 Using the provided integer address, STORE writes data
2145 to the specified buffer or texture.
2147 The 'address' is specified as a vector of unsigned
2148 integers. If the 'address' is out of range the result
2151 Only the first mipmap level of a resource can be
2152 written to using this instruction.
2154 For 1D or 2D texture arrays, the array index is
2155 provided as an unsigned integer in address.y or
2156 address.z, respectively. address.yz are ignored for
2157 buffers and 1D textures. address.z is ignored for 1D
2158 texture arrays and 2D textures. address.w is always
2162 .. _threadsyncopcodes:
2164 Inter-thread synchronization opcodes
2165 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2167 These opcodes are intended for communication between threads running
2168 within the same compute grid. For now they're only valid in compute
2171 .. opcode:: MFENCE - Memory fence
2173 Syntax: ``MFENCE resource``
2175 Example: ``MFENCE RES[0]``
2177 This opcode forces strong ordering between any memory access
2178 operations that affect the specified resource. This means that
2179 previous loads and stores (and only those) will be performed and
2180 visible to other threads before the program execution continues.
2183 .. opcode:: LFENCE - Load memory fence
2185 Syntax: ``LFENCE resource``
2187 Example: ``LFENCE RES[0]``
2189 Similar to MFENCE, but it only affects the ordering of memory loads.
2192 .. opcode:: SFENCE - Store memory fence
2194 Syntax: ``SFENCE resource``
2196 Example: ``SFENCE RES[0]``
2198 Similar to MFENCE, but it only affects the ordering of memory stores.
2201 .. opcode:: BARRIER - Thread group barrier
2205 This opcode suspends the execution of the current thread until all
2206 the remaining threads in the working group reach the same point of
2207 the program. Results are unspecified if any of the remaining
2208 threads terminates or never reaches an executed BARRIER instruction.
2216 These opcodes provide atomic variants of some common arithmetic and
2217 logical operations. In this context atomicity means that another
2218 concurrent memory access operation that affects the same memory
2219 location is guaranteed to be performed strictly before or after the
2220 entire execution of the atomic operation.
2222 For the moment they're only valid in compute programs.
2224 .. opcode:: ATOMUADD - Atomic integer addition
2226 Syntax: ``ATOMUADD dst, resource, offset, src``
2228 Example: ``ATOMUADD TEMP[0], RES[0], TEMP[1], TEMP[2]``
2230 The following operation is performed atomically on each component:
2234 dst_i = resource[offset]_i
2236 resource[offset]_i = dst_i + src_i
2239 .. opcode:: ATOMXCHG - Atomic exchange
2241 Syntax: ``ATOMXCHG dst, resource, offset, src``
2243 Example: ``ATOMXCHG TEMP[0], RES[0], TEMP[1], TEMP[2]``
2245 The following operation is performed atomically on each component:
2249 dst_i = resource[offset]_i
2251 resource[offset]_i = src_i
2254 .. opcode:: ATOMCAS - Atomic compare-and-exchange
2256 Syntax: ``ATOMCAS dst, resource, offset, cmp, src``
2258 Example: ``ATOMCAS TEMP[0], RES[0], TEMP[1], TEMP[2], TEMP[3]``
2260 The following operation is performed atomically on each component:
2264 dst_i = resource[offset]_i
2266 resource[offset]_i = (dst_i == cmp_i ? src_i : dst_i)
2269 .. opcode:: ATOMAND - Atomic bitwise And
2271 Syntax: ``ATOMAND dst, resource, offset, src``
2273 Example: ``ATOMAND TEMP[0], RES[0], TEMP[1], TEMP[2]``
2275 The following operation is performed atomically on each component:
2279 dst_i = resource[offset]_i
2281 resource[offset]_i = dst_i \& src_i
2284 .. opcode:: ATOMOR - Atomic bitwise Or
2286 Syntax: ``ATOMOR dst, resource, offset, src``
2288 Example: ``ATOMOR TEMP[0], RES[0], TEMP[1], TEMP[2]``
2290 The following operation is performed atomically on each component:
2294 dst_i = resource[offset]_i
2296 resource[offset]_i = dst_i | src_i
2299 .. opcode:: ATOMXOR - Atomic bitwise Xor
2301 Syntax: ``ATOMXOR dst, resource, offset, src``
2303 Example: ``ATOMXOR TEMP[0], RES[0], TEMP[1], TEMP[2]``
2305 The following operation is performed atomically on each component:
2309 dst_i = resource[offset]_i
2311 resource[offset]_i = dst_i \oplus src_i
2314 .. opcode:: ATOMUMIN - Atomic unsigned minimum
2316 Syntax: ``ATOMUMIN dst, resource, offset, src``
2318 Example: ``ATOMUMIN TEMP[0], RES[0], TEMP[1], TEMP[2]``
2320 The following operation is performed atomically on each component:
2324 dst_i = resource[offset]_i
2326 resource[offset]_i = (dst_i < src_i ? dst_i : src_i)
2329 .. opcode:: ATOMUMAX - Atomic unsigned maximum
2331 Syntax: ``ATOMUMAX dst, resource, offset, src``
2333 Example: ``ATOMUMAX TEMP[0], RES[0], TEMP[1], TEMP[2]``
2335 The following operation is performed atomically on each component:
2339 dst_i = resource[offset]_i
2341 resource[offset]_i = (dst_i > src_i ? dst_i : src_i)
2344 .. opcode:: ATOMIMIN - Atomic signed minimum
2346 Syntax: ``ATOMIMIN dst, resource, offset, src``
2348 Example: ``ATOMIMIN TEMP[0], RES[0], TEMP[1], TEMP[2]``
2350 The following operation is performed atomically on each component:
2354 dst_i = resource[offset]_i
2356 resource[offset]_i = (dst_i < src_i ? dst_i : src_i)
2359 .. opcode:: ATOMIMAX - Atomic signed maximum
2361 Syntax: ``ATOMIMAX dst, resource, offset, src``
2363 Example: ``ATOMIMAX TEMP[0], RES[0], TEMP[1], TEMP[2]``
2365 The following operation is performed atomically on each component:
2369 dst_i = resource[offset]_i
2371 resource[offset]_i = (dst_i > src_i ? dst_i : src_i)
2375 Explanation of symbols used
2376 ------------------------------
2383 :math:`|x|` Absolute value of `x`.
2385 :math:`\lceil x \rceil` Ceiling of `x`.
2387 clamp(x,y,z) Clamp x between y and z.
2388 (x < y) ? y : (x > z) ? z : x
2390 :math:`\lfloor x\rfloor` Floor of `x`.
2392 :math:`\log_2{x}` Logarithm of `x`, base 2.
2394 max(x,y) Maximum of x and y.
2397 min(x,y) Minimum of x and y.
2400 partialx(x) Derivative of x relative to fragment's X.
2402 partialy(x) Derivative of x relative to fragment's Y.
2404 pop() Pop from stack.
2406 :math:`x^y` `x` to the power `y`.
2408 push(x) Push x on stack.
2412 trunc(x) Truncate x, i.e. drop the fraction bits.
2419 discard Discard fragment.
2423 target Label of target instruction.
2434 Declares a register that is will be referenced as an operand in Instruction
2437 File field contains register file that is being declared and is one
2440 UsageMask field specifies which of the register components can be accessed
2441 and is one of TGSI_WRITEMASK.
2443 The Local flag specifies that a given value isn't intended for
2444 subroutine parameter passing and, as a result, the implementation
2445 isn't required to give any guarantees of it being preserved across
2446 subroutine boundaries. As it's merely a compiler hint, the
2447 implementation is free to ignore it.
2449 If Dimension flag is set to 1, a Declaration Dimension token follows.
2451 If Semantic flag is set to 1, a Declaration Semantic token follows.
2453 If Interpolate flag is set to 1, a Declaration Interpolate token follows.
2455 If file is TGSI_FILE_RESOURCE, a Declaration Resource token follows.
2457 If Array flag is set to 1, a Declaration Array token follows.
2460 ^^^^^^^^^^^^^^^^^^^^^^^^
2462 Declarations can optional have an ArrayID attribute which can be referred by
2463 indirect addressing operands. An ArrayID of zero is reserved and treaded as
2464 if no ArrayID is specified.
2466 If an indirect addressing operand refers to a specific declaration by using
2467 an ArrayID only the registers in this declaration are guaranteed to be
2468 accessed, accessing any register outside this declaration results in undefined
2469 behavior. Note that for compatibility the effective index is zero-based and
2470 not relative to the specified declaration
2472 If no ArrayID is specified with an indirect addressing operand the whole
2473 register file might be accessed by this operand. This is strongly discouraged
2474 and will prevent packing of scalar/vec2 arrays and effective alias analysis.
2476 Declaration Semantic
2477 ^^^^^^^^^^^^^^^^^^^^^^^^
2479 Vertex and fragment shader input and output registers may be labeled
2480 with semantic information consisting of a name and index.
2482 Follows Declaration token if Semantic bit is set.
2484 Since its purpose is to link a shader with other stages of the pipeline,
2485 it is valid to follow only those Declaration tokens that declare a register
2486 either in INPUT or OUTPUT file.
2488 SemanticName field contains the semantic name of the register being declared.
2489 There is no default value.
2491 SemanticIndex is an optional subscript that can be used to distinguish
2492 different register declarations with the same semantic name. The default value
2495 The meanings of the individual semantic names are explained in the following
2498 TGSI_SEMANTIC_POSITION
2499 """"""""""""""""""""""
2501 For vertex shaders, TGSI_SEMANTIC_POSITION indicates the vertex shader
2502 output register which contains the homogeneous vertex position in the clip
2503 space coordinate system. After clipping, the X, Y and Z components of the
2504 vertex will be divided by the W value to get normalized device coordinates.
2506 For fragment shaders, TGSI_SEMANTIC_POSITION is used to indicate that
2507 fragment shader input contains the fragment's window position. The X
2508 component starts at zero and always increases from left to right.
2509 The Y component starts at zero and always increases but Y=0 may either
2510 indicate the top of the window or the bottom depending on the fragment
2511 coordinate origin convention (see TGSI_PROPERTY_FS_COORD_ORIGIN).
2512 The Z coordinate ranges from 0 to 1 to represent depth from the front
2513 to the back of the Z buffer. The W component contains the interpolated
2514 reciprocal of the vertex position W component (corresponding to gl_Fragcoord,
2515 but unlike d3d10 which interpolates the same 1/w but then gives back
2516 the reciprocal of the interpolated value).
2518 Fragment shaders may also declare an output register with
2519 TGSI_SEMANTIC_POSITION. Only the Z component is writable. This allows
2520 the fragment shader to change the fragment's Z position.
2527 For vertex shader outputs or fragment shader inputs/outputs, this
2528 label indicates that the resister contains an R,G,B,A color.
2530 Several shader inputs/outputs may contain colors so the semantic index
2531 is used to distinguish them. For example, color[0] may be the diffuse
2532 color while color[1] may be the specular color.
2534 This label is needed so that the flat/smooth shading can be applied
2535 to the right interpolants during rasterization.
2539 TGSI_SEMANTIC_BCOLOR
2540 """"""""""""""""""""
2542 Back-facing colors are only used for back-facing polygons, and are only valid
2543 in vertex shader outputs. After rasterization, all polygons are front-facing
2544 and COLOR and BCOLOR end up occupying the same slots in the fragment shader,
2545 so all BCOLORs effectively become regular COLORs in the fragment shader.
2551 Vertex shader inputs and outputs and fragment shader inputs may be
2552 labeled with TGSI_SEMANTIC_FOG to indicate that the register contains
2553 a fog coordinate. Typically, the fragment shader will use the fog coordinate
2554 to compute a fog blend factor which is used to blend the normal fragment color
2555 with a constant fog color. But fog coord really is just an ordinary vec4
2556 register like regular semantics.
2562 Vertex shader input and output registers may be labeled with
2563 TGIS_SEMANTIC_PSIZE to indicate that the register contains a point size
2564 in the form (S, 0, 0, 1). The point size controls the width or diameter
2565 of points for rasterization. This label cannot be used in fragment
2568 When using this semantic, be sure to set the appropriate state in the
2569 :ref:`rasterizer` first.
2572 TGSI_SEMANTIC_TEXCOORD
2573 """"""""""""""""""""""
2575 Only available if PIPE_CAP_TGSI_TEXCOORD is exposed !
2577 Vertex shader outputs and fragment shader inputs may be labeled with
2578 this semantic to make them replaceable by sprite coordinates via the
2579 sprite_coord_enable state in the :ref:`rasterizer`.
2580 The semantic index permitted with this semantic is limited to <= 7.
2582 If the driver does not support TEXCOORD, sprite coordinate replacement
2583 applies to inputs with the GENERIC semantic instead.
2585 The intended use case for this semantic is gl_TexCoord.
2588 TGSI_SEMANTIC_PCOORD
2589 """"""""""""""""""""
2591 Only available if PIPE_CAP_TGSI_TEXCOORD is exposed !
2593 Fragment shader inputs may be labeled with TGSI_SEMANTIC_PCOORD to indicate
2594 that the register contains sprite coordinates in the form (x, y, 0, 1), if
2595 the current primitive is a point and point sprites are enabled. Otherwise,
2596 the contents of the register are undefined.
2598 The intended use case for this semantic is gl_PointCoord.
2601 TGSI_SEMANTIC_GENERIC
2602 """""""""""""""""""""
2604 All vertex/fragment shader inputs/outputs not labeled with any other
2605 semantic label can be considered to be generic attributes. Typical
2606 uses of generic inputs/outputs are texcoords and user-defined values.
2609 TGSI_SEMANTIC_NORMAL
2610 """"""""""""""""""""
2612 Indicates that a vertex shader input is a normal vector. This is
2613 typically only used for legacy graphics APIs.
2619 This label applies to fragment shader inputs only and indicates that
2620 the register contains front/back-face information of the form (F, 0,
2621 0, 1). The first component will be positive when the fragment belongs
2622 to a front-facing polygon, and negative when the fragment belongs to a
2623 back-facing polygon.
2626 TGSI_SEMANTIC_EDGEFLAG
2627 """"""""""""""""""""""
2629 For vertex shaders, this sematic label indicates that an input or
2630 output is a boolean edge flag. The register layout is [F, x, x, x]
2631 where F is 0.0 or 1.0 and x = don't care. Normally, the vertex shader
2632 simply copies the edge flag input to the edgeflag output.
2634 Edge flags are used to control which lines or points are actually
2635 drawn when the polygon mode converts triangles/quads/polygons into
2639 TGSI_SEMANTIC_STENCIL
2640 """""""""""""""""""""
2642 For fragment shaders, this semantic label indicates that an output
2643 is a writable stencil reference value. Only the Y component is writable.
2644 This allows the fragment shader to change the fragments stencilref value.
2647 TGSI_SEMANTIC_VIEWPORT_INDEX
2648 """"""""""""""""""""""""""""
2650 For geometry shaders, this semantic label indicates that an output
2651 contains the index of the viewport (and scissor) to use.
2652 This is an integer value, and only the X component is used.
2658 For geometry shaders, this semantic label indicates that an output
2659 contains the layer value to use for the color and depth/stencil surfaces.
2660 This is an integer value, and only the X component is used.
2661 (Also known as rendertarget array index.)
2664 TGSI_SEMANTIC_CULLDIST
2665 """"""""""""""""""""""
2667 Used as distance to plane for performing application-defined culling
2668 of individual primitives against a plane. When components of vertex
2669 elements are given this label, these values are assumed to be a
2670 float32 signed distance to a plane. Primitives will be completely
2671 discarded if the plane distance for all of the vertices in the
2672 primitive are < 0. If a vertex has a cull distance of NaN, that
2673 vertex counts as "out" (as if its < 0);
2674 The limits on both clip and cull distances are bound
2675 by the PIPE_MAX_CLIP_OR_CULL_DISTANCE_COUNT define which defines
2676 the maximum number of components that can be used to hold the
2677 distances and by the PIPE_MAX_CLIP_OR_CULL_DISTANCE_ELEMENT_COUNT
2678 which specifies the maximum number of registers which can be
2679 annotated with those semantics.
2682 TGSI_SEMANTIC_CLIPDIST
2683 """"""""""""""""""""""
2685 When components of vertex elements are identified this way, these
2686 values are each assumed to be a float32 signed distance to a plane.
2687 Primitive setup only invokes rasterization on pixels for which
2688 the interpolated plane distances are >= 0. Multiple clip planes
2689 can be implemented simultaneously, by annotating multiple
2690 components of one or more vertex elements with the above specified
2691 semantic. The limits on both clip and cull distances are bound
2692 by the PIPE_MAX_CLIP_OR_CULL_DISTANCE_COUNT define which defines
2693 the maximum number of components that can be used to hold the
2694 distances and by the PIPE_MAX_CLIP_OR_CULL_DISTANCE_ELEMENT_COUNT
2695 which specifies the maximum number of registers which can be
2696 annotated with those semantics.
2698 TGSI_SEMANTIC_SAMPLEID
2699 """"""""""""""""""""""
2701 For fragment shaders, this semantic label indicates that a system value
2702 contains the current sample id (i.e. gl_SampleID).
2703 This is an integer value, and only the X component is used.
2705 TGSI_SEMANTIC_SAMPLEPOS
2706 """""""""""""""""""""""
2708 For fragment shaders, this semantic label indicates that a system value
2709 contains the current sample's position (i.e. gl_SamplePosition). Only the X
2710 and Y values are used.
2712 TGSI_SEMANTIC_SAMPLEMASK
2713 """"""""""""""""""""""""
2715 For fragment shaders, this semantic label indicates that an output contains
2716 the sample mask used to disable further sample processing
2717 (i.e. gl_SampleMask). Only the X value is used, up to 32x MS.
2719 TGSI_SEMANTIC_INVOCATIONID
2720 """"""""""""""""""""""""""
2722 For geometry shaders, this semantic label indicates that a system value
2723 contains the current invocation id (i.e. gl_InvocationID).
2724 This is an integer value, and only the X component is used.
2726 TGSI_SEMANTIC_INSTANCEID
2727 """"""""""""""""""""""""
2729 For vertex shaders, this semantic label indicates that a system value contains
2730 the current instance id (i.e. gl_InstanceID). It does not include the base
2731 instance. This is an integer value, and only the X component is used.
2733 TGSI_SEMANTIC_VERTEXID
2734 """"""""""""""""""""""
2736 For vertex shaders, this semantic label indicates that a system value contains
2737 the current vertex id (i.e. gl_VertexID). It does (unlike in d3d10) include the
2738 base vertex. This is an integer value, and only the X component is used.
2740 TGSI_SEMANTIC_VERTEXID_NOBASE
2741 """""""""""""""""""""""""""""""
2743 For vertex shaders, this semantic label indicates that a system value contains
2744 the current vertex id without including the base vertex (this corresponds to
2745 d3d10 vertex id, so TGSI_SEMANTIC_VERTEXID_NOBASE + TGSI_SEMANTIC_BASEVERTEX
2746 == TGSI_SEMANTIC_VERTEXID). This is an integer value, and only the X component
2749 TGSI_SEMANTIC_BASEVERTEX
2750 """"""""""""""""""""""""
2752 For vertex shaders, this semantic label indicates that a system value contains
2753 the base vertex (i.e. gl_BaseVertex). Note that for non-indexed draw calls,
2754 this contains the first (or start) value instead.
2755 This is an integer value, and only the X component is used.
2757 TGSI_SEMANTIC_PRIMID
2758 """"""""""""""""""""
2760 For geometry and fragment shaders, this semantic label indicates the value
2761 contains the primitive id (i.e. gl_PrimitiveID). This is an integer value,
2762 and only the X component is used.
2763 FIXME: This right now can be either a ordinary input or a system value...
2766 Declaration Interpolate
2767 ^^^^^^^^^^^^^^^^^^^^^^^
2769 This token is only valid for fragment shader INPUT declarations.
2771 The Interpolate field specifes the way input is being interpolated by
2772 the rasteriser and is one of TGSI_INTERPOLATE_*.
2774 The Location field specifies the location inside the pixel that the
2775 interpolation should be done at, one of ``TGSI_INTERPOLATE_LOC_*``. Note that
2776 when per-sample shading is enabled, the implementation may choose to
2777 interpolate at the sample irrespective of the Location field.
2779 The CylindricalWrap bitfield specifies which register components
2780 should be subject to cylindrical wrapping when interpolating by the
2781 rasteriser. If TGSI_CYLINDRICAL_WRAP_X is set to 1, the X component
2782 should be interpolated according to cylindrical wrapping rules.
2785 Declaration Sampler View
2786 ^^^^^^^^^^^^^^^^^^^^^^^^
2788 Follows Declaration token if file is TGSI_FILE_SAMPLER_VIEW.
2790 DCL SVIEW[#], resource, type(s)
2792 Declares a shader input sampler view and assigns it to a SVIEW[#]
2795 resource can be one of BUFFER, 1D, 2D, 3D, 1DArray and 2DArray.
2797 type must be 1 or 4 entries (if specifying on a per-component
2798 level) out of UNORM, SNORM, SINT, UINT and FLOAT.
2801 Declaration Resource
2802 ^^^^^^^^^^^^^^^^^^^^
2804 Follows Declaration token if file is TGSI_FILE_RESOURCE.
2806 DCL RES[#], resource [, WR] [, RAW]
2808 Declares a shader input resource and assigns it to a RES[#]
2811 resource can be one of BUFFER, 1D, 2D, 3D, CUBE, 1DArray and
2814 If the RAW keyword is not specified, the texture data will be
2815 subject to conversion, swizzling and scaling as required to yield
2816 the specified data type from the physical data format of the bound
2819 If the RAW keyword is specified, no channel conversion will be
2820 performed: the values read for each of the channels (X,Y,Z,W) will
2821 correspond to consecutive words in the same order and format
2822 they're found in memory. No element-to-address conversion will be
2823 performed either: the value of the provided X coordinate will be
2824 interpreted in byte units instead of texel units. The result of
2825 accessing a misaligned address is undefined.
2827 Usage of the STORE opcode is only allowed if the WR (writable) flag
2832 ^^^^^^^^^^^^^^^^^^^^^^^^
2834 Properties are general directives that apply to the whole TGSI program.
2839 Specifies the fragment shader TGSI_SEMANTIC_POSITION coordinate origin.
2840 The default value is UPPER_LEFT.
2842 If UPPER_LEFT, the position will be (0,0) at the upper left corner and
2843 increase downward and rightward.
2844 If LOWER_LEFT, the position will be (0,0) at the lower left corner and
2845 increase upward and rightward.
2847 OpenGL defaults to LOWER_LEFT, and is configurable with the
2848 GL_ARB_fragment_coord_conventions extension.
2850 DirectX 9/10 use UPPER_LEFT.
2852 FS_COORD_PIXEL_CENTER
2853 """""""""""""""""""""
2855 Specifies the fragment shader TGSI_SEMANTIC_POSITION pixel center convention.
2856 The default value is HALF_INTEGER.
2858 If HALF_INTEGER, the fractionary part of the position will be 0.5
2859 If INTEGER, the fractionary part of the position will be 0.0
2861 Note that this does not affect the set of fragments generated by
2862 rasterization, which is instead controlled by half_pixel_center in the
2865 OpenGL defaults to HALF_INTEGER, and is configurable with the
2866 GL_ARB_fragment_coord_conventions extension.
2868 DirectX 9 uses INTEGER.
2869 DirectX 10 uses HALF_INTEGER.
2871 FS_COLOR0_WRITES_ALL_CBUFS
2872 """"""""""""""""""""""""""
2873 Specifies that writes to the fragment shader color 0 are replicated to all
2874 bound cbufs. This facilitates OpenGL's fragColor output vs fragData[0] where
2875 fragData is directed to a single color buffer, but fragColor is broadcast.
2878 """"""""""""""""""""""""""
2879 If this property is set on the program bound to the shader stage before the
2880 fragment shader, user clip planes should have no effect (be disabled) even if
2881 that shader does not write to any clip distance outputs and the rasterizer's
2882 clip_plane_enable is non-zero.
2883 This property is only supported by drivers that also support shader clip
2885 This is useful for APIs that don't have UCPs and where clip distances written
2886 by a shader cannot be disabled.
2891 Specifies the number of times a geometry shader should be executed for each
2892 input primitive. Each invocation will have a different
2893 TGSI_SEMANTIC_INVOCATIONID system value set. If not specified, assumed to
2896 VS_WINDOW_SPACE_POSITION
2897 """"""""""""""""""""""""""
2898 If this property is set on the vertex shader, the TGSI_SEMANTIC_POSITION output
2899 is assumed to contain window space coordinates.
2900 Division of X,Y,Z by W and the viewport transformation are disabled, and 1/W is
2901 directly taken from the 4-th component of the shader output.
2902 Naturally, clipping is not performed on window coordinates either.
2903 The effect of this property is undefined if a geometry or tessellation shader
2906 Texture Sampling and Texture Formats
2907 ------------------------------------
2909 This table shows how texture image components are returned as (x,y,z,w) tuples
2910 by TGSI texture instructions, such as :opcode:`TEX`, :opcode:`TXD`, and
2911 :opcode:`TXP`. For reference, OpenGL and Direct3D conventions are shown as
2914 +--------------------+--------------+--------------------+--------------+
2915 | Texture Components | Gallium | OpenGL | Direct3D 9 |
2916 +====================+==============+====================+==============+
2917 | R | (r, 0, 0, 1) | (r, 0, 0, 1) | (r, 1, 1, 1) |
2918 +--------------------+--------------+--------------------+--------------+
2919 | RG | (r, g, 0, 1) | (r, g, 0, 1) | (r, g, 1, 1) |
2920 +--------------------+--------------+--------------------+--------------+
2921 | RGB | (r, g, b, 1) | (r, g, b, 1) | (r, g, b, 1) |
2922 +--------------------+--------------+--------------------+--------------+
2923 | RGBA | (r, g, b, a) | (r, g, b, a) | (r, g, b, a) |
2924 +--------------------+--------------+--------------------+--------------+
2925 | A | (0, 0, 0, a) | (0, 0, 0, a) | (0, 0, 0, a) |
2926 +--------------------+--------------+--------------------+--------------+
2927 | L | (l, l, l, 1) | (l, l, l, 1) | (l, l, l, 1) |
2928 +--------------------+--------------+--------------------+--------------+
2929 | LA | (l, l, l, a) | (l, l, l, a) | (l, l, l, a) |
2930 +--------------------+--------------+--------------------+--------------+
2931 | I | (i, i, i, i) | (i, i, i, i) | N/A |
2932 +--------------------+--------------+--------------------+--------------+
2933 | UV | XXX TBD | (0, 0, 0, 1) | (u, v, 1, 1) |
2934 | | | [#envmap-bumpmap]_ | |
2935 +--------------------+--------------+--------------------+--------------+
2936 | Z | XXX TBD | (z, z, z, 1) | (0, z, 0, 1) |
2937 | | | [#depth-tex-mode]_ | |
2938 +--------------------+--------------+--------------------+--------------+
2939 | S | (s, s, s, s) | unknown | unknown |
2940 +--------------------+--------------+--------------------+--------------+
2942 .. [#envmap-bumpmap] http://www.opengl.org/registry/specs/ATI/envmap_bumpmap.txt
2943 .. [#depth-tex-mode] the default is (z, z, z, 1) but may also be (0, 0, 0, z)
2944 or (z, z, z, z) depending on the value of GL_DEPTH_TEXTURE_MODE.