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.
29 TGSI supports modifiers on inputs (as well as saturate modifier on instructions).
31 For inputs which have a floating point type, both absolute value and negation
32 modifiers are supported (with absolute value being applied first).
33 TGSI_OPCODE_MOV is considered to have float input type for applying modifiers.
35 For inputs which have signed or unsigned type only the negate modifier is
42 ^^^^^^^^^^^^^^^^^^^^^^^^^
44 These opcodes are guaranteed to be available regardless of the driver being
47 .. opcode:: ARL - Address Register Load
51 dst.x = \lfloor src.x\rfloor
53 dst.y = \lfloor src.y\rfloor
55 dst.z = \lfloor src.z\rfloor
57 dst.w = \lfloor src.w\rfloor
60 .. opcode:: MOV - Move
73 .. opcode:: LIT - Light Coefficients
81 dst.z = (src.x > 0) ? max(src.y, 0)^{clamp(src.w, -128, 128))} : 0
86 .. opcode:: RCP - Reciprocal
88 This instruction replicates its result.
95 .. opcode:: RSQ - Reciprocal Square Root
97 This instruction replicates its result. The results are undefined for src <= 0.
101 dst = \frac{1}{\sqrt{src.x}}
104 .. opcode:: SQRT - Square Root
106 This instruction replicates its result. The results are undefined for src < 0.
113 .. opcode:: EXP - Approximate Exponential Base 2
117 dst.x = 2^{\lfloor src.x\rfloor}
119 dst.y = src.x - \lfloor src.x\rfloor
126 .. opcode:: LOG - Approximate Logarithm Base 2
130 dst.x = \lfloor\log_2{|src.x|}\rfloor
132 dst.y = \frac{|src.x|}{2^{\lfloor\log_2{|src.x|}\rfloor}}
134 dst.z = \log_2{|src.x|}
139 .. opcode:: MUL - Multiply
143 dst.x = src0.x \times src1.x
145 dst.y = src0.y \times src1.y
147 dst.z = src0.z \times src1.z
149 dst.w = src0.w \times src1.w
152 .. opcode:: ADD - Add
156 dst.x = src0.x + src1.x
158 dst.y = src0.y + src1.y
160 dst.z = src0.z + src1.z
162 dst.w = src0.w + src1.w
165 .. opcode:: DP3 - 3-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
174 .. opcode:: DP4 - 4-component Dot Product
176 This instruction replicates its result.
180 dst = src0.x \times src1.x + src0.y \times src1.y + src0.z \times src1.z + src0.w \times src1.w
183 .. opcode:: DST - Distance Vector
189 dst.y = src0.y \times src1.y
196 .. opcode:: MIN - Minimum
200 dst.x = min(src0.x, src1.x)
202 dst.y = min(src0.y, src1.y)
204 dst.z = min(src0.z, src1.z)
206 dst.w = min(src0.w, src1.w)
209 .. opcode:: MAX - Maximum
213 dst.x = max(src0.x, src1.x)
215 dst.y = max(src0.y, src1.y)
217 dst.z = max(src0.z, src1.z)
219 dst.w = max(src0.w, src1.w)
222 .. opcode:: SLT - Set On Less Than
226 dst.x = (src0.x < src1.x) ? 1.0F : 0.0F
228 dst.y = (src0.y < src1.y) ? 1.0F : 0.0F
230 dst.z = (src0.z < src1.z) ? 1.0F : 0.0F
232 dst.w = (src0.w < src1.w) ? 1.0F : 0.0F
235 .. opcode:: SGE - Set On Greater Equal Than
239 dst.x = (src0.x >= src1.x) ? 1.0F : 0.0F
241 dst.y = (src0.y >= src1.y) ? 1.0F : 0.0F
243 dst.z = (src0.z >= src1.z) ? 1.0F : 0.0F
245 dst.w = (src0.w >= src1.w) ? 1.0F : 0.0F
248 .. opcode:: MAD - Multiply And Add
252 dst.x = src0.x \times src1.x + src2.x
254 dst.y = src0.y \times src1.y + src2.y
256 dst.z = src0.z \times src1.z + src2.z
258 dst.w = src0.w \times src1.w + src2.w
261 .. opcode:: SUB - Subtract
265 dst.x = src0.x - src1.x
267 dst.y = src0.y - src1.y
269 dst.z = src0.z - src1.z
271 dst.w = src0.w - src1.w
274 .. opcode:: LRP - Linear Interpolate
278 dst.x = src0.x \times src1.x + (1 - src0.x) \times src2.x
280 dst.y = src0.y \times src1.y + (1 - src0.y) \times src2.y
282 dst.z = src0.z \times src1.z + (1 - src0.z) \times src2.z
284 dst.w = src0.w \times src1.w + (1 - src0.w) \times src2.w
287 .. opcode:: CND - Condition
291 dst.x = (src2.x > 0.5) ? src0.x : src1.x
293 dst.y = (src2.y > 0.5) ? src0.y : src1.y
295 dst.z = (src2.z > 0.5) ? src0.z : src1.z
297 dst.w = (src2.w > 0.5) ? src0.w : src1.w
300 .. opcode:: DP2A - 2-component Dot Product And Add
304 dst.x = src0.x \times src1.x + src0.y \times src1.y + src2.x
306 dst.y = src0.x \times src1.x + src0.y \times src1.y + src2.x
308 dst.z = src0.x \times src1.x + src0.y \times src1.y + src2.x
310 dst.w = src0.x \times src1.x + src0.y \times src1.y + src2.x
313 .. opcode:: FRC - Fraction
317 dst.x = src.x - \lfloor src.x\rfloor
319 dst.y = src.y - \lfloor src.y\rfloor
321 dst.z = src.z - \lfloor src.z\rfloor
323 dst.w = src.w - \lfloor src.w\rfloor
326 .. opcode:: CLAMP - Clamp
330 dst.x = clamp(src0.x, src1.x, src2.x)
332 dst.y = clamp(src0.y, src1.y, src2.y)
334 dst.z = clamp(src0.z, src1.z, src2.z)
336 dst.w = clamp(src0.w, src1.w, src2.w)
339 .. opcode:: FLR - Floor
341 This is identical to :opcode:`ARL`.
345 dst.x = \lfloor src.x\rfloor
347 dst.y = \lfloor src.y\rfloor
349 dst.z = \lfloor src.z\rfloor
351 dst.w = \lfloor src.w\rfloor
354 .. opcode:: ROUND - Round
367 .. opcode:: EX2 - Exponential Base 2
369 This instruction replicates its result.
376 .. opcode:: LG2 - Logarithm Base 2
378 This instruction replicates its result.
385 .. opcode:: POW - Power
387 This instruction replicates its result.
391 dst = src0.x^{src1.x}
393 .. opcode:: XPD - Cross Product
397 dst.x = src0.y \times src1.z - src1.y \times src0.z
399 dst.y = src0.z \times src1.x - src1.z \times src0.x
401 dst.z = src0.x \times src1.y - src1.x \times src0.y
406 .. opcode:: ABS - Absolute
419 .. opcode:: RCC - Reciprocal Clamped
421 This instruction replicates its result.
423 XXX cleanup on aisle three
427 dst = (1 / src.x) > 0 ? clamp(1 / src.x, 5.42101e-020, 1.84467e+019) : clamp(1 / src.x, -1.84467e+019, -5.42101e-020)
430 .. opcode:: DPH - Homogeneous Dot Product
432 This instruction replicates its result.
436 dst = src0.x \times src1.x + src0.y \times src1.y + src0.z \times src1.z + src1.w
439 .. opcode:: COS - Cosine
441 This instruction replicates its result.
448 .. opcode:: DDX - Derivative Relative To X
452 dst.x = partialx(src.x)
454 dst.y = partialx(src.y)
456 dst.z = partialx(src.z)
458 dst.w = partialx(src.w)
461 .. opcode:: DDY - Derivative Relative To Y
465 dst.x = partialy(src.x)
467 dst.y = partialy(src.y)
469 dst.z = partialy(src.z)
471 dst.w = partialy(src.w)
474 .. opcode:: PK2H - Pack Two 16-bit Floats
479 .. opcode:: PK2US - Pack Two Unsigned 16-bit Scalars
484 .. opcode:: PK4B - Pack Four Signed 8-bit Scalars
489 .. opcode:: PK4UB - Pack Four Unsigned 8-bit Scalars
494 .. opcode:: RFL - Reflection Vector
498 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
500 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
502 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
508 Considered for removal.
511 .. opcode:: SEQ - Set On Equal
515 dst.x = (src0.x == src1.x) ? 1.0F : 0.0F
517 dst.y = (src0.y == src1.y) ? 1.0F : 0.0F
519 dst.z = (src0.z == src1.z) ? 1.0F : 0.0F
521 dst.w = (src0.w == src1.w) ? 1.0F : 0.0F
524 .. opcode:: SFL - Set On False
526 This instruction replicates its result.
534 Considered for removal.
537 .. opcode:: SGT - Set On Greater Than
541 dst.x = (src0.x > src1.x) ? 1.0F : 0.0F
543 dst.y = (src0.y > src1.y) ? 1.0F : 0.0F
545 dst.z = (src0.z > src1.z) ? 1.0F : 0.0F
547 dst.w = (src0.w > src1.w) ? 1.0F : 0.0F
550 .. opcode:: SIN - Sine
552 This instruction replicates its result.
559 .. opcode:: SLE - Set On Less Equal Than
563 dst.x = (src0.x <= src1.x) ? 1.0F : 0.0F
565 dst.y = (src0.y <= src1.y) ? 1.0F : 0.0F
567 dst.z = (src0.z <= src1.z) ? 1.0F : 0.0F
569 dst.w = (src0.w <= src1.w) ? 1.0F : 0.0F
572 .. opcode:: SNE - Set On Not Equal
576 dst.x = (src0.x != src1.x) ? 1.0F : 0.0F
578 dst.y = (src0.y != src1.y) ? 1.0F : 0.0F
580 dst.z = (src0.z != src1.z) ? 1.0F : 0.0F
582 dst.w = (src0.w != src1.w) ? 1.0F : 0.0F
585 .. opcode:: STR - Set On True
587 This instruction replicates its result.
594 .. opcode:: TEX - Texture Lookup
602 dst = texture_sample(unit, coord, bias)
604 for array textures src0.y contains the slice for 1D,
605 and src0.z contain the slice for 2D.
606 for shadow textures with no arrays, src0.z contains
608 for shadow textures with arrays, src0.z contains
609 the reference value for 1D arrays, and src0.w contains
610 the reference value for 2D arrays.
611 There is no way to pass a bias in the .w value for
612 shadow arrays, and GLSL doesn't allow this.
613 GLSL does allow cube shadows maps to take a bias value,
614 and we have to determine how this will look in TGSI.
616 .. opcode:: TXD - Texture Lookup with Derivatives
628 dst = texture_sample_deriv(unit, coord, bias, ddx, ddy)
631 .. opcode:: TXP - Projective Texture Lookup
635 coord.x = src0.x / src.w
637 coord.y = src0.y / src.w
639 coord.z = src0.z / src.w
645 dst = texture_sample(unit, coord, bias)
648 .. opcode:: UP2H - Unpack Two 16-Bit Floats
654 Considered for removal.
656 .. opcode:: UP2US - Unpack Two Unsigned 16-Bit Scalars
662 Considered for removal.
664 .. opcode:: UP4B - Unpack Four Signed 8-Bit Values
670 Considered for removal.
672 .. opcode:: UP4UB - Unpack Four Unsigned 8-Bit Scalars
678 Considered for removal.
680 .. opcode:: X2D - 2D Coordinate Transformation
684 dst.x = src0.x + src1.x \times src2.x + src1.y \times src2.y
686 dst.y = src0.y + src1.x \times src2.z + src1.y \times src2.w
688 dst.z = src0.x + src1.x \times src2.x + src1.y \times src2.y
690 dst.w = src0.y + src1.x \times src2.z + src1.y \times src2.w
694 Considered for removal.
697 .. opcode:: ARA - Address Register Add
703 Considered for removal.
705 .. opcode:: ARR - Address Register Load With Round
718 .. opcode:: SSG - Set Sign
722 dst.x = (src.x > 0) ? 1 : (src.x < 0) ? -1 : 0
724 dst.y = (src.y > 0) ? 1 : (src.y < 0) ? -1 : 0
726 dst.z = (src.z > 0) ? 1 : (src.z < 0) ? -1 : 0
728 dst.w = (src.w > 0) ? 1 : (src.w < 0) ? -1 : 0
731 .. opcode:: CMP - Compare
735 dst.x = (src0.x < 0) ? src1.x : src2.x
737 dst.y = (src0.y < 0) ? src1.y : src2.y
739 dst.z = (src0.z < 0) ? src1.z : src2.z
741 dst.w = (src0.w < 0) ? src1.w : src2.w
744 .. opcode:: KILL_IF - Conditional Discard
746 Conditional discard. Allowed in fragment shaders only.
750 if (src.x < 0 || src.y < 0 || src.z < 0 || src.w < 0)
755 .. opcode:: KILL - Discard
757 Unconditional discard. Allowed in fragment shaders only.
760 .. opcode:: SCS - Sine Cosine
773 .. opcode:: TXB - Texture Lookup With Bias
787 dst = texture_sample(unit, coord, bias)
790 .. opcode:: NRM - 3-component Vector Normalise
794 dst.x = src.x / (src.x \times src.x + src.y \times src.y + src.z \times src.z)
796 dst.y = src.y / (src.x \times src.x + src.y \times src.y + src.z \times src.z)
798 dst.z = src.z / (src.x \times src.x + src.y \times src.y + src.z \times src.z)
803 .. opcode:: DIV - Divide
807 dst.x = \frac{src0.x}{src1.x}
809 dst.y = \frac{src0.y}{src1.y}
811 dst.z = \frac{src0.z}{src1.z}
813 dst.w = \frac{src0.w}{src1.w}
816 .. opcode:: DP2 - 2-component Dot Product
818 This instruction replicates its result.
822 dst = src0.x \times src1.x + src0.y \times src1.y
825 .. opcode:: TXL - Texture Lookup With explicit LOD
839 dst = texture_sample(unit, coord, lod)
842 .. opcode:: PUSHA - Push Address Register On Stack
851 Considered for cleanup.
855 Considered for removal.
857 .. opcode:: POPA - Pop Address Register From Stack
866 Considered for cleanup.
870 Considered for removal.
873 .. opcode:: BRA - Branch
879 Considered for removal.
882 .. opcode:: CALLNZ - Subroutine Call If Not Zero
888 Considered for cleanup.
892 Considered for removal.
896 ^^^^^^^^^^^^^^^^^^^^^^^^
898 These opcodes are primarily provided for special-use computational shaders.
899 Support for these opcodes indicated by a special pipe capability bit (TBD).
901 XXX doesn't look like most of the opcodes really belong here.
903 .. opcode:: CEIL - Ceiling
907 dst.x = \lceil src.x\rceil
909 dst.y = \lceil src.y\rceil
911 dst.z = \lceil src.z\rceil
913 dst.w = \lceil src.w\rceil
916 .. opcode:: TRUNC - Truncate
929 .. opcode:: MOD - Modulus
933 dst.x = src0.x \bmod src1.x
935 dst.y = src0.y \bmod src1.y
937 dst.z = src0.z \bmod src1.z
939 dst.w = src0.w \bmod src1.w
942 .. opcode:: UARL - Integer Address Register Load
944 Moves the contents of the source register, assumed to be an integer, into the
945 destination register, which is assumed to be an address (ADDR) register.
948 .. opcode:: SAD - Sum Of Absolute Differences
952 dst.x = |src0.x - src1.x| + src2.x
954 dst.y = |src0.y - src1.y| + src2.y
956 dst.z = |src0.z - src1.z| + src2.z
958 dst.w = |src0.w - src1.w| + src2.w
961 .. opcode:: TXF - Texel Fetch (as per NV_gpu_shader4), extract a single texel
962 from a specified texture image. The source sampler may
963 not be a CUBE or SHADOW.
964 src 0 is a four-component signed integer vector used to
965 identify the single texel accessed. 3 components + level.
966 src 1 is a 3 component constant signed integer vector,
967 with each component only have a range of
968 -8..+8 (hw only seems to deal with this range, interface
969 allows for up to unsigned int).
970 TXF(uint_vec coord, int_vec offset).
973 .. opcode:: TXQ - Texture Size Query (as per NV_gpu_program4)
974 retrieve the dimensions of the texture
975 depending on the target. For 1D (width), 2D/RECT/CUBE
976 (width, height), 3D (width, height, depth),
977 1D array (width, layers), 2D array (width, height, layers)
983 dst.x = texture_width(unit, lod)
985 dst.y = texture_height(unit, lod)
987 dst.z = texture_depth(unit, lod)
989 .. opcode:: TG4 - Texture Gather (as per ARB_texture_gather)
990 Gathers the four texels to be used in a bi-linear
991 filtering operation and packs them into a single register.
992 Only works with 2D, 2D array, cubemaps, and cubemaps arrays.
993 For 2D textures, only the addressing modes of the sampler and
994 the top level of any mip pyramid are used. Set W to zero.
995 It behaves like the TEX instruction, but a filtered
996 sample is not generated. The four samples that contribute
997 to filtering are placed into xyzw in clockwise order,
998 starting with the (u,v) texture coordinate delta at the
999 following locations (-, +), (+, +), (+, -), (-, -), where
1000 the magnitude of the deltas are half a texel.
1002 PIPE_CAP_TEXTURE_SM5 enhances this instruction to support
1003 shadow per-sample depth compares, single component selection,
1004 and a non-constant offset. It doesn't allow support for the
1005 GL independent offset to get i0,j0. This would require another
1006 CAP is hw can do it natively. For now we lower that before
1015 dst = texture_gather4 (unit, coord, component)
1017 (with SM5 - cube array shadow)
1023 dst = texture_gather (uint, coord, compare)
1027 ^^^^^^^^^^^^^^^^^^^^^^^^
1028 These opcodes are used for integer operations.
1029 Support for these opcodes indicated by PIPE_SHADER_CAP_INTEGERS (all of them?)
1032 .. opcode:: I2F - Signed Integer To Float
1034 Rounding is unspecified (round to nearest even suggested).
1038 dst.x = (float) src.x
1040 dst.y = (float) src.y
1042 dst.z = (float) src.z
1044 dst.w = (float) src.w
1047 .. opcode:: U2F - Unsigned Integer To Float
1049 Rounding is unspecified (round to nearest even suggested).
1053 dst.x = (float) src.x
1055 dst.y = (float) src.y
1057 dst.z = (float) src.z
1059 dst.w = (float) src.w
1062 .. opcode:: F2I - Float to Signed Integer
1064 Rounding is towards zero (truncate).
1065 Values outside signed range (including NaNs) produce undefined results.
1078 .. opcode:: F2U - Float to Unsigned Integer
1080 Rounding is towards zero (truncate).
1081 Values outside unsigned range (including NaNs) produce undefined results.
1085 dst.x = (unsigned) src.x
1087 dst.y = (unsigned) src.y
1089 dst.z = (unsigned) src.z
1091 dst.w = (unsigned) src.w
1094 .. opcode:: UADD - Integer Add
1096 This instruction works the same for signed and unsigned integers.
1097 The low 32bit of the result is returned.
1101 dst.x = src0.x + src1.x
1103 dst.y = src0.y + src1.y
1105 dst.z = src0.z + src1.z
1107 dst.w = src0.w + src1.w
1110 .. opcode:: UMAD - Integer Multiply And Add
1112 This instruction works the same for signed and unsigned integers.
1113 The multiplication returns the low 32bit (as does the result itself).
1117 dst.x = src0.x \times src1.x + src2.x
1119 dst.y = src0.y \times src1.y + src2.y
1121 dst.z = src0.z \times src1.z + src2.z
1123 dst.w = src0.w \times src1.w + src2.w
1126 .. opcode:: UMUL - Integer Multiply
1128 This instruction works the same for signed and unsigned integers.
1129 The low 32bit of the result is returned.
1133 dst.x = src0.x \times src1.x
1135 dst.y = src0.y \times src1.y
1137 dst.z = src0.z \times src1.z
1139 dst.w = src0.w \times src1.w
1142 .. opcode:: IMUL_HI - Signed Integer Multiply High Bits
1144 The high 32bits of the multiplication of 2 signed integers are returned.
1148 dst.x = (src0.x \times src1.x) >> 32
1150 dst.y = (src0.y \times src1.y) >> 32
1152 dst.z = (src0.z \times src1.z) >> 32
1154 dst.w = (src0.w \times src1.w) >> 32
1157 .. opcode:: UMUL_HI - Unsigned Integer Multiply High Bits
1159 The high 32bits of the multiplication of 2 unsigned integers are returned.
1163 dst.x = (src0.x \times src1.x) >> 32
1165 dst.y = (src0.y \times src1.y) >> 32
1167 dst.z = (src0.z \times src1.z) >> 32
1169 dst.w = (src0.w \times src1.w) >> 32
1172 .. opcode:: IDIV - Signed Integer Division
1174 TBD: behavior for division by zero.
1178 dst.x = src0.x \ src1.x
1180 dst.y = src0.y \ src1.y
1182 dst.z = src0.z \ src1.z
1184 dst.w = src0.w \ src1.w
1187 .. opcode:: UDIV - Unsigned Integer Division
1189 For division by zero, 0xffffffff is returned.
1193 dst.x = src0.x \ src1.x
1195 dst.y = src0.y \ src1.y
1197 dst.z = src0.z \ src1.z
1199 dst.w = src0.w \ src1.w
1202 .. opcode:: UMOD - Unsigned Integer Remainder
1204 If second arg is zero, 0xffffffff is returned.
1208 dst.x = src0.x \ src1.x
1210 dst.y = src0.y \ src1.y
1212 dst.z = src0.z \ src1.z
1214 dst.w = src0.w \ src1.w
1217 .. opcode:: NOT - Bitwise Not
1230 .. opcode:: AND - Bitwise And
1234 dst.x = src0.x & src1.x
1236 dst.y = src0.y & src1.y
1238 dst.z = src0.z & src1.z
1240 dst.w = src0.w & src1.w
1243 .. opcode:: OR - Bitwise Or
1247 dst.x = src0.x | src1.x
1249 dst.y = src0.y | src1.y
1251 dst.z = src0.z | src1.z
1253 dst.w = src0.w | src1.w
1256 .. opcode:: XOR - Bitwise Xor
1260 dst.x = src0.x \oplus src1.x
1262 dst.y = src0.y \oplus src1.y
1264 dst.z = src0.z \oplus src1.z
1266 dst.w = src0.w \oplus src1.w
1269 .. opcode:: IMAX - Maximum of Signed Integers
1273 dst.x = max(src0.x, src1.x)
1275 dst.y = max(src0.y, src1.y)
1277 dst.z = max(src0.z, src1.z)
1279 dst.w = max(src0.w, src1.w)
1282 .. opcode:: UMAX - Maximum of Unsigned Integers
1286 dst.x = max(src0.x, src1.x)
1288 dst.y = max(src0.y, src1.y)
1290 dst.z = max(src0.z, src1.z)
1292 dst.w = max(src0.w, src1.w)
1295 .. opcode:: IMIN - Minimum of Signed Integers
1299 dst.x = min(src0.x, src1.x)
1301 dst.y = min(src0.y, src1.y)
1303 dst.z = min(src0.z, src1.z)
1305 dst.w = min(src0.w, src1.w)
1308 .. opcode:: UMIN - Minimum of Unsigned Integers
1312 dst.x = min(src0.x, src1.x)
1314 dst.y = min(src0.y, src1.y)
1316 dst.z = min(src0.z, src1.z)
1318 dst.w = min(src0.w, src1.w)
1321 .. opcode:: SHL - Shift Left
1323 The shift count is masked with 0x1f before the shift is applied.
1327 dst.x = src0.x << (0x1f & src1.x)
1329 dst.y = src0.y << (0x1f & src1.y)
1331 dst.z = src0.z << (0x1f & src1.z)
1333 dst.w = src0.w << (0x1f & src1.w)
1336 .. opcode:: ISHR - Arithmetic Shift Right (of Signed Integer)
1338 The shift count is masked with 0x1f before the shift is applied.
1342 dst.x = src0.x >> (0x1f & src1.x)
1344 dst.y = src0.y >> (0x1f & src1.y)
1346 dst.z = src0.z >> (0x1f & src1.z)
1348 dst.w = src0.w >> (0x1f & src1.w)
1351 .. opcode:: USHR - Logical Shift Right
1353 The shift count is masked with 0x1f before the shift is applied.
1357 dst.x = src0.x >> (unsigned) (0x1f & src1.x)
1359 dst.y = src0.y >> (unsigned) (0x1f & src1.y)
1361 dst.z = src0.z >> (unsigned) (0x1f & src1.z)
1363 dst.w = src0.w >> (unsigned) (0x1f & src1.w)
1366 .. opcode:: UCMP - Integer Conditional Move
1370 dst.x = src0.x ? src1.x : src2.x
1372 dst.y = src0.y ? src1.y : src2.y
1374 dst.z = src0.z ? src1.z : src2.z
1376 dst.w = src0.w ? src1.w : src2.w
1380 .. opcode:: ISSG - Integer Set Sign
1384 dst.x = (src0.x < 0) ? -1 : (src0.x > 0) ? 1 : 0
1386 dst.y = (src0.y < 0) ? -1 : (src0.y > 0) ? 1 : 0
1388 dst.z = (src0.z < 0) ? -1 : (src0.z > 0) ? 1 : 0
1390 dst.w = (src0.w < 0) ? -1 : (src0.w > 0) ? 1 : 0
1394 .. opcode:: FSLT - Float Set On Less Than (ordered)
1396 Same comparison as SLT but returns integer instead of 1.0/0.0 float
1400 dst.x = (src0.x < src1.x) ? ~0 : 0
1402 dst.y = (src0.y < src1.y) ? ~0 : 0
1404 dst.z = (src0.z < src1.z) ? ~0 : 0
1406 dst.w = (src0.w < src1.w) ? ~0 : 0
1409 .. opcode:: ISLT - Signed Integer Set On Less Than
1413 dst.x = (src0.x < src1.x) ? ~0 : 0
1415 dst.y = (src0.y < src1.y) ? ~0 : 0
1417 dst.z = (src0.z < src1.z) ? ~0 : 0
1419 dst.w = (src0.w < src1.w) ? ~0 : 0
1422 .. opcode:: USLT - Unsigned Integer Set On Less Than
1426 dst.x = (src0.x < src1.x) ? ~0 : 0
1428 dst.y = (src0.y < src1.y) ? ~0 : 0
1430 dst.z = (src0.z < src1.z) ? ~0 : 0
1432 dst.w = (src0.w < src1.w) ? ~0 : 0
1435 .. opcode:: FSGE - Float Set On Greater Equal Than (ordered)
1437 Same comparison as SGE but returns integer instead of 1.0/0.0 float
1441 dst.x = (src0.x >= src1.x) ? ~0 : 0
1443 dst.y = (src0.y >= src1.y) ? ~0 : 0
1445 dst.z = (src0.z >= src1.z) ? ~0 : 0
1447 dst.w = (src0.w >= src1.w) ? ~0 : 0
1450 .. opcode:: ISGE - Signed Integer Set On Greater Equal Than
1454 dst.x = (src0.x >= src1.x) ? ~0 : 0
1456 dst.y = (src0.y >= src1.y) ? ~0 : 0
1458 dst.z = (src0.z >= src1.z) ? ~0 : 0
1460 dst.w = (src0.w >= src1.w) ? ~0 : 0
1463 .. opcode:: USGE - Unsigned Integer Set On Greater Equal Than
1467 dst.x = (src0.x >= src1.x) ? ~0 : 0
1469 dst.y = (src0.y >= src1.y) ? ~0 : 0
1471 dst.z = (src0.z >= src1.z) ? ~0 : 0
1473 dst.w = (src0.w >= src1.w) ? ~0 : 0
1476 .. opcode:: FSEQ - Float Set On Equal (ordered)
1478 Same comparison as SEQ but returns integer instead of 1.0/0.0 float
1482 dst.x = (src0.x == src1.x) ? ~0 : 0
1484 dst.y = (src0.y == src1.y) ? ~0 : 0
1486 dst.z = (src0.z == src1.z) ? ~0 : 0
1488 dst.w = (src0.w == src1.w) ? ~0 : 0
1491 .. opcode:: USEQ - Integer Set On Equal
1495 dst.x = (src0.x == src1.x) ? ~0 : 0
1497 dst.y = (src0.y == src1.y) ? ~0 : 0
1499 dst.z = (src0.z == src1.z) ? ~0 : 0
1501 dst.w = (src0.w == src1.w) ? ~0 : 0
1504 .. opcode:: FSNE - Float Set On Not Equal (unordered)
1506 Same comparison as SNE but returns integer instead of 1.0/0.0 float
1510 dst.x = (src0.x != src1.x) ? ~0 : 0
1512 dst.y = (src0.y != src1.y) ? ~0 : 0
1514 dst.z = (src0.z != src1.z) ? ~0 : 0
1516 dst.w = (src0.w != src1.w) ? ~0 : 0
1519 .. opcode:: USNE - Integer Set On Not Equal
1523 dst.x = (src0.x != src1.x) ? ~0 : 0
1525 dst.y = (src0.y != src1.y) ? ~0 : 0
1527 dst.z = (src0.z != src1.z) ? ~0 : 0
1529 dst.w = (src0.w != src1.w) ? ~0 : 0
1532 .. opcode:: INEG - Integer Negate
1547 .. opcode:: IABS - Integer Absolute Value
1561 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1563 These opcodes are only supported in geometry shaders; they have no meaning
1564 in any other type of shader.
1566 .. opcode:: EMIT - Emit
1568 Generate a new vertex for the current primitive using the values in the
1572 .. opcode:: ENDPRIM - End Primitive
1574 Complete the current primitive (consisting of the emitted vertices),
1575 and start a new one.
1581 These opcodes are part of :term:`GLSL`'s opcode set. Support for these
1582 opcodes is determined by a special capability bit, ``GLSL``.
1583 Some require glsl version 1.30 (UIF/BREAKC/SWITCH/CASE/DEFAULT/ENDSWITCH).
1585 .. opcode:: CAL - Subroutine Call
1591 .. opcode:: RET - Subroutine Call Return
1596 .. opcode:: CONT - Continue
1598 Unconditionally moves the point of execution to the instruction after the
1599 last bgnloop. The instruction must appear within a bgnloop/endloop.
1603 Support for CONT is determined by a special capability bit,
1604 ``TGSI_CONT_SUPPORTED``. See :ref:`Screen` for more information.
1607 .. opcode:: BGNLOOP - Begin a Loop
1609 Start a loop. Must have a matching endloop.
1612 .. opcode:: BGNSUB - Begin Subroutine
1614 Starts definition of a subroutine. Must have a matching endsub.
1617 .. opcode:: ENDLOOP - End a Loop
1619 End a loop started with bgnloop.
1622 .. opcode:: ENDSUB - End Subroutine
1624 Ends definition of a subroutine.
1627 .. opcode:: NOP - No Operation
1632 .. opcode:: BRK - Break
1634 Unconditionally moves the point of execution to the instruction after the
1635 next endloop or endswitch. The instruction must appear within a loop/endloop
1636 or switch/endswitch.
1639 .. opcode:: BREAKC - Break Conditional
1641 Conditionally moves the point of execution to the instruction after the
1642 next endloop or endswitch. The instruction must appear within a loop/endloop
1643 or switch/endswitch.
1644 Condition evaluates to true if src0.x != 0 where src0.x is interpreted
1645 as an integer register.
1649 Considered for removal as it's quite inconsistent wrt other opcodes
1650 (could emulate with UIF/BRK/ENDIF).
1653 .. opcode:: IF - Float If
1655 Start an IF ... ELSE .. ENDIF block. Condition evaluates to true if
1659 where src0.x is interpreted as a floating point register.
1662 .. opcode:: UIF - Bitwise If
1664 Start an UIF ... ELSE .. ENDIF block. Condition evaluates to true if
1668 where src0.x is interpreted as an integer register.
1671 .. opcode:: ELSE - Else
1673 Starts an else block, after an IF or UIF statement.
1676 .. opcode:: ENDIF - End If
1678 Ends an IF or UIF block.
1681 .. opcode:: SWITCH - Switch
1683 Starts a C-style switch expression. The switch consists of one or multiple
1684 CASE statements, and at most one DEFAULT statement. Execution of a statement
1685 ends when a BRK is hit, but just like in C falling through to other cases
1686 without a break is allowed. Similarly, DEFAULT label is allowed anywhere not
1687 just as last statement, and fallthrough is allowed into/from it.
1688 CASE src arguments are evaluated at bit level against the SWITCH src argument.
1693 (some instructions here)
1696 (some instructions here)
1699 (some instructions here)
1704 .. opcode:: CASE - Switch case
1706 This represents a switch case label. The src arg must be an integer immediate.
1709 .. opcode:: DEFAULT - Switch default
1711 This represents the default case in the switch, which is taken if no other
1715 .. opcode:: ENDSWITCH - End of switch
1717 Ends a switch expression.
1720 .. opcode:: NRM4 - 4-component Vector Normalise
1722 This instruction replicates its result.
1726 dst = \frac{src.x}{src.x \times src.x + src.y \times src.y + src.z \times src.z + src.w \times src.w}
1734 The double-precision opcodes reinterpret four-component vectors into
1735 two-component vectors with doubled precision in each component.
1737 Support for these opcodes is XXX undecided. :T
1739 .. opcode:: DADD - Add
1743 dst.xy = src0.xy + src1.xy
1745 dst.zw = src0.zw + src1.zw
1748 .. opcode:: DDIV - Divide
1752 dst.xy = src0.xy / src1.xy
1754 dst.zw = src0.zw / src1.zw
1756 .. opcode:: DSEQ - Set on Equal
1760 dst.xy = src0.xy == src1.xy ? 1.0F : 0.0F
1762 dst.zw = src0.zw == src1.zw ? 1.0F : 0.0F
1764 .. opcode:: DSLT - Set on Less than
1768 dst.xy = src0.xy < src1.xy ? 1.0F : 0.0F
1770 dst.zw = src0.zw < src1.zw ? 1.0F : 0.0F
1772 .. opcode:: DFRAC - Fraction
1776 dst.xy = src.xy - \lfloor src.xy\rfloor
1778 dst.zw = src.zw - \lfloor src.zw\rfloor
1781 .. opcode:: DFRACEXP - Convert Number to Fractional and Integral Components
1783 Like the ``frexp()`` routine in many math libraries, this opcode stores the
1784 exponent of its source to ``dst0``, and the significand to ``dst1``, such that
1785 :math:`dst1 \times 2^{dst0} = src` .
1789 dst0.xy = exp(src.xy)
1791 dst1.xy = frac(src.xy)
1793 dst0.zw = exp(src.zw)
1795 dst1.zw = frac(src.zw)
1797 .. opcode:: DLDEXP - Multiply Number by Integral Power of 2
1799 This opcode is the inverse of :opcode:`DFRACEXP`.
1803 dst.xy = src0.xy \times 2^{src1.xy}
1805 dst.zw = src0.zw \times 2^{src1.zw}
1807 .. opcode:: DMIN - Minimum
1811 dst.xy = min(src0.xy, src1.xy)
1813 dst.zw = min(src0.zw, src1.zw)
1815 .. opcode:: DMAX - Maximum
1819 dst.xy = max(src0.xy, src1.xy)
1821 dst.zw = max(src0.zw, src1.zw)
1823 .. opcode:: DMUL - Multiply
1827 dst.xy = src0.xy \times src1.xy
1829 dst.zw = src0.zw \times src1.zw
1832 .. opcode:: DMAD - Multiply And Add
1836 dst.xy = src0.xy \times src1.xy + src2.xy
1838 dst.zw = src0.zw \times src1.zw + src2.zw
1841 .. opcode:: DRCP - Reciprocal
1845 dst.xy = \frac{1}{src.xy}
1847 dst.zw = \frac{1}{src.zw}
1849 .. opcode:: DSQRT - Square Root
1853 dst.xy = \sqrt{src.xy}
1855 dst.zw = \sqrt{src.zw}
1858 .. _samplingopcodes:
1860 Resource Sampling Opcodes
1861 ^^^^^^^^^^^^^^^^^^^^^^^^^
1863 Those opcodes follow very closely semantics of the respective Direct3D
1864 instructions. If in doubt double check Direct3D documentation.
1865 Note that the swizzle on SVIEW (src1) determines texel swizzling
1868 .. opcode:: SAMPLE - Using provided address, sample data from the
1869 specified texture using the filtering mode identified
1870 by the gven sampler. The source data may come from
1871 any resource type other than buffers.
1872 SAMPLE dst, address, sampler_view, sampler
1874 SAMPLE TEMP[0], TEMP[1], SVIEW[0], SAMP[0]
1876 .. opcode:: SAMPLE_I - Simplified alternative to the SAMPLE instruction.
1877 Using the provided integer address, SAMPLE_I fetches data
1878 from the specified sampler view without any filtering.
1879 The source data may come from any resource type other
1881 SAMPLE_I dst, address, sampler_view
1883 SAMPLE_I TEMP[0], TEMP[1], SVIEW[0]
1884 The 'address' is specified as unsigned integers. If the
1885 'address' is out of range [0...(# texels - 1)] the
1886 result of the fetch is always 0 in all components.
1887 As such the instruction doesn't honor address wrap
1888 modes, in cases where that behavior is desirable
1889 'SAMPLE' instruction should be used.
1890 address.w always provides an unsigned integer mipmap
1891 level. If the value is out of the range then the
1892 instruction always returns 0 in all components.
1893 address.yz are ignored for buffers and 1d textures.
1894 address.z is ignored for 1d texture arrays and 2d
1896 For 1D texture arrays address.y provides the array
1897 index (also as unsigned integer). If the value is
1898 out of the range of available array indices
1899 [0... (array size - 1)] then the opcode always returns
1900 0 in all components.
1901 For 2D texture arrays address.z provides the array
1902 index, otherwise it exhibits the same behavior as in
1903 the case for 1D texture arrays.
1904 The exact semantics of the source address are presented
1906 resource type X Y Z W
1907 ------------- ------------------------
1908 PIPE_BUFFER x ignored
1909 PIPE_TEXTURE_1D x mpl
1910 PIPE_TEXTURE_2D x y mpl
1911 PIPE_TEXTURE_3D x y z mpl
1912 PIPE_TEXTURE_RECT x y mpl
1913 PIPE_TEXTURE_CUBE not allowed as source
1914 PIPE_TEXTURE_1D_ARRAY x idx mpl
1915 PIPE_TEXTURE_2D_ARRAY x y idx mpl
1917 Where 'mpl' is a mipmap level and 'idx' is the
1920 .. opcode:: SAMPLE_I_MS - Just like SAMPLE_I but allows fetch data from
1921 multi-sampled surfaces.
1922 SAMPLE_I_MS dst, address, sampler_view, sample
1924 .. opcode:: SAMPLE_B - Just like the SAMPLE instruction with the
1925 exception that an additional bias is applied to the
1926 level of detail computed as part of the instruction
1928 SAMPLE_B dst, address, sampler_view, sampler, lod_bias
1930 SAMPLE_B TEMP[0], TEMP[1], SVIEW[0], SAMP[0], TEMP[2].x
1932 .. opcode:: SAMPLE_C - Similar to the SAMPLE instruction but it
1933 performs a comparison filter. The operands to SAMPLE_C
1934 are identical to SAMPLE, except that there is an additional
1935 float32 operand, reference value, which must be a register
1936 with single-component, or a scalar literal.
1937 SAMPLE_C makes the hardware use the current samplers
1938 compare_func (in pipe_sampler_state) to compare
1939 reference value against the red component value for the
1940 surce resource at each texel that the currently configured
1941 texture filter covers based on the provided coordinates.
1942 SAMPLE_C dst, address, sampler_view.r, sampler, ref_value
1944 SAMPLE_C TEMP[0], TEMP[1], SVIEW[0].r, SAMP[0], TEMP[2].x
1946 .. opcode:: SAMPLE_C_LZ - Same as SAMPLE_C, but LOD is 0 and derivatives
1947 are ignored. The LZ stands for level-zero.
1948 SAMPLE_C_LZ dst, address, sampler_view.r, sampler, ref_value
1950 SAMPLE_C_LZ TEMP[0], TEMP[1], SVIEW[0].r, SAMP[0], TEMP[2].x
1953 .. opcode:: SAMPLE_D - SAMPLE_D is identical to the SAMPLE opcode except
1954 that the derivatives for the source address in the x
1955 direction and the y direction are provided by extra
1957 SAMPLE_D dst, address, sampler_view, sampler, der_x, der_y
1959 SAMPLE_D TEMP[0], TEMP[1], SVIEW[0], SAMP[0], TEMP[2], TEMP[3]
1961 .. opcode:: SAMPLE_L - SAMPLE_L is identical to the SAMPLE opcode except
1962 that the LOD is provided directly as a scalar value,
1963 representing no anisotropy.
1964 SAMPLE_L dst, address, sampler_view, sampler, explicit_lod
1966 SAMPLE_L TEMP[0], TEMP[1], SVIEW[0], SAMP[0], TEMP[2].x
1968 .. opcode:: GATHER4 - Gathers the four texels to be used in a bi-linear
1969 filtering operation and packs them into a single register.
1970 Only works with 2D, 2D array, cubemaps, and cubemaps arrays.
1971 For 2D textures, only the addressing modes of the sampler and
1972 the top level of any mip pyramid are used. Set W to zero.
1973 It behaves like the SAMPLE instruction, but a filtered
1974 sample is not generated. The four samples that contribute
1975 to filtering are placed into xyzw in counter-clockwise order,
1976 starting with the (u,v) texture coordinate delta at the
1977 following locations (-, +), (+, +), (+, -), (-, -), where
1978 the magnitude of the deltas are half a texel.
1981 .. opcode:: SVIEWINFO - query the dimensions of a given sampler view.
1982 dst receives width, height, depth or array size and
1983 number of mipmap levels as int4. The dst can have a writemask
1984 which will specify what info is the caller interested
1986 SVIEWINFO dst, src_mip_level, sampler_view
1988 SVIEWINFO TEMP[0], TEMP[1].x, SVIEW[0]
1989 src_mip_level is an unsigned integer scalar. If it's
1990 out of range then returns 0 for width, height and
1991 depth/array size but the total number of mipmap is
1992 still returned correctly for the given sampler view.
1993 The returned width, height and depth values are for
1994 the mipmap level selected by the src_mip_level and
1995 are in the number of texels.
1996 For 1d texture array width is in dst.x, array size
1997 is in dst.y and dst.z is 0. The number of mipmaps
1999 In contrast to d3d10 resinfo, there's no way in the
2000 tgsi instruction encoding to specify the return type
2001 (float/rcpfloat/uint), hence always using uint. Also,
2002 unlike the SAMPLE instructions, the swizzle on src1
2003 resinfo allowing swizzling dst values is ignored (due
2004 to the interaction with rcpfloat modifier which requires
2005 some swizzle handling in the state tracker anyway).
2007 .. opcode:: SAMPLE_POS - query the position of a given sample.
2008 dst receives float4 (x, y, 0, 0) indicated where the
2009 sample is located. If the resource is not a multi-sample
2010 resource and not a render target, the result is 0.
2012 .. opcode:: SAMPLE_INFO - dst receives number of samples in x.
2013 If the resource is not a multi-sample resource and
2014 not a render target, the result is 0.
2017 .. _resourceopcodes:
2019 Resource Access Opcodes
2020 ^^^^^^^^^^^^^^^^^^^^^^^
2022 .. opcode:: LOAD - Fetch data from a shader resource
2024 Syntax: ``LOAD dst, resource, address``
2026 Example: ``LOAD TEMP[0], RES[0], TEMP[1]``
2028 Using the provided integer address, LOAD fetches data
2029 from the specified buffer or texture without any
2032 The 'address' is specified as a vector of unsigned
2033 integers. If the 'address' is out of range the result
2036 Only the first mipmap level of a resource can be read
2037 from using this instruction.
2039 For 1D or 2D texture arrays, the array index is
2040 provided as an unsigned integer in address.y or
2041 address.z, respectively. address.yz are ignored for
2042 buffers and 1D textures. address.z is ignored for 1D
2043 texture arrays and 2D textures. address.w is always
2046 .. opcode:: STORE - Write data to a shader resource
2048 Syntax: ``STORE resource, address, src``
2050 Example: ``STORE RES[0], TEMP[0], TEMP[1]``
2052 Using the provided integer address, STORE writes data
2053 to the specified buffer or texture.
2055 The 'address' is specified as a vector of unsigned
2056 integers. If the 'address' is out of range the result
2059 Only the first mipmap level of a resource can be
2060 written to using this instruction.
2062 For 1D or 2D texture arrays, the array index is
2063 provided as an unsigned integer in address.y or
2064 address.z, respectively. address.yz are ignored for
2065 buffers and 1D textures. address.z is ignored for 1D
2066 texture arrays and 2D textures. address.w is always
2070 .. _threadsyncopcodes:
2072 Inter-thread synchronization opcodes
2073 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2075 These opcodes are intended for communication between threads running
2076 within the same compute grid. For now they're only valid in compute
2079 .. opcode:: MFENCE - Memory fence
2081 Syntax: ``MFENCE resource``
2083 Example: ``MFENCE RES[0]``
2085 This opcode forces strong ordering between any memory access
2086 operations that affect the specified resource. This means that
2087 previous loads and stores (and only those) will be performed and
2088 visible to other threads before the program execution continues.
2091 .. opcode:: LFENCE - Load memory fence
2093 Syntax: ``LFENCE resource``
2095 Example: ``LFENCE RES[0]``
2097 Similar to MFENCE, but it only affects the ordering of memory loads.
2100 .. opcode:: SFENCE - Store memory fence
2102 Syntax: ``SFENCE resource``
2104 Example: ``SFENCE RES[0]``
2106 Similar to MFENCE, but it only affects the ordering of memory stores.
2109 .. opcode:: BARRIER - Thread group barrier
2113 This opcode suspends the execution of the current thread until all
2114 the remaining threads in the working group reach the same point of
2115 the program. Results are unspecified if any of the remaining
2116 threads terminates or never reaches an executed BARRIER instruction.
2124 These opcodes provide atomic variants of some common arithmetic and
2125 logical operations. In this context atomicity means that another
2126 concurrent memory access operation that affects the same memory
2127 location is guaranteed to be performed strictly before or after the
2128 entire execution of the atomic operation.
2130 For the moment they're only valid in compute programs.
2132 .. opcode:: ATOMUADD - Atomic integer addition
2134 Syntax: ``ATOMUADD dst, resource, offset, src``
2136 Example: ``ATOMUADD TEMP[0], RES[0], TEMP[1], TEMP[2]``
2138 The following operation is performed atomically on each component:
2142 dst_i = resource[offset]_i
2144 resource[offset]_i = dst_i + src_i
2147 .. opcode:: ATOMXCHG - Atomic exchange
2149 Syntax: ``ATOMXCHG dst, resource, offset, src``
2151 Example: ``ATOMXCHG TEMP[0], RES[0], TEMP[1], TEMP[2]``
2153 The following operation is performed atomically on each component:
2157 dst_i = resource[offset]_i
2159 resource[offset]_i = src_i
2162 .. opcode:: ATOMCAS - Atomic compare-and-exchange
2164 Syntax: ``ATOMCAS dst, resource, offset, cmp, src``
2166 Example: ``ATOMCAS TEMP[0], RES[0], TEMP[1], TEMP[2], TEMP[3]``
2168 The following operation is performed atomically on each component:
2172 dst_i = resource[offset]_i
2174 resource[offset]_i = (dst_i == cmp_i ? src_i : dst_i)
2177 .. opcode:: ATOMAND - Atomic bitwise And
2179 Syntax: ``ATOMAND dst, resource, offset, src``
2181 Example: ``ATOMAND TEMP[0], RES[0], TEMP[1], TEMP[2]``
2183 The following operation is performed atomically on each component:
2187 dst_i = resource[offset]_i
2189 resource[offset]_i = dst_i \& src_i
2192 .. opcode:: ATOMOR - Atomic bitwise Or
2194 Syntax: ``ATOMOR dst, resource, offset, src``
2196 Example: ``ATOMOR TEMP[0], RES[0], TEMP[1], TEMP[2]``
2198 The following operation is performed atomically on each component:
2202 dst_i = resource[offset]_i
2204 resource[offset]_i = dst_i | src_i
2207 .. opcode:: ATOMXOR - Atomic bitwise Xor
2209 Syntax: ``ATOMXOR dst, resource, offset, src``
2211 Example: ``ATOMXOR TEMP[0], RES[0], TEMP[1], TEMP[2]``
2213 The following operation is performed atomically on each component:
2217 dst_i = resource[offset]_i
2219 resource[offset]_i = dst_i \oplus src_i
2222 .. opcode:: ATOMUMIN - Atomic unsigned minimum
2224 Syntax: ``ATOMUMIN dst, resource, offset, src``
2226 Example: ``ATOMUMIN TEMP[0], RES[0], TEMP[1], TEMP[2]``
2228 The following operation is performed atomically on each component:
2232 dst_i = resource[offset]_i
2234 resource[offset]_i = (dst_i < src_i ? dst_i : src_i)
2237 .. opcode:: ATOMUMAX - Atomic unsigned maximum
2239 Syntax: ``ATOMUMAX dst, resource, offset, src``
2241 Example: ``ATOMUMAX TEMP[0], RES[0], TEMP[1], TEMP[2]``
2243 The following operation is performed atomically on each component:
2247 dst_i = resource[offset]_i
2249 resource[offset]_i = (dst_i > src_i ? dst_i : src_i)
2252 .. opcode:: ATOMIMIN - Atomic signed minimum
2254 Syntax: ``ATOMIMIN dst, resource, offset, src``
2256 Example: ``ATOMIMIN TEMP[0], RES[0], TEMP[1], TEMP[2]``
2258 The following operation is performed atomically on each component:
2262 dst_i = resource[offset]_i
2264 resource[offset]_i = (dst_i < src_i ? dst_i : src_i)
2267 .. opcode:: ATOMIMAX - Atomic signed maximum
2269 Syntax: ``ATOMIMAX dst, resource, offset, src``
2271 Example: ``ATOMIMAX TEMP[0], RES[0], TEMP[1], TEMP[2]``
2273 The following operation is performed atomically on each component:
2277 dst_i = resource[offset]_i
2279 resource[offset]_i = (dst_i > src_i ? dst_i : src_i)
2283 Explanation of symbols used
2284 ------------------------------
2291 :math:`|x|` Absolute value of `x`.
2293 :math:`\lceil x \rceil` Ceiling of `x`.
2295 clamp(x,y,z) Clamp x between y and z.
2296 (x < y) ? y : (x > z) ? z : x
2298 :math:`\lfloor x\rfloor` Floor of `x`.
2300 :math:`\log_2{x}` Logarithm of `x`, base 2.
2302 max(x,y) Maximum of x and y.
2305 min(x,y) Minimum of x and y.
2308 partialx(x) Derivative of x relative to fragment's X.
2310 partialy(x) Derivative of x relative to fragment's Y.
2312 pop() Pop from stack.
2314 :math:`x^y` `x` to the power `y`.
2316 push(x) Push x on stack.
2320 trunc(x) Truncate x, i.e. drop the fraction bits.
2327 discard Discard fragment.
2331 target Label of target instruction.
2342 Declares a register that is will be referenced as an operand in Instruction
2345 File field contains register file that is being declared and is one
2348 UsageMask field specifies which of the register components can be accessed
2349 and is one of TGSI_WRITEMASK.
2351 The Local flag specifies that a given value isn't intended for
2352 subroutine parameter passing and, as a result, the implementation
2353 isn't required to give any guarantees of it being preserved across
2354 subroutine boundaries. As it's merely a compiler hint, the
2355 implementation is free to ignore it.
2357 If Dimension flag is set to 1, a Declaration Dimension token follows.
2359 If Semantic flag is set to 1, a Declaration Semantic token follows.
2361 If Interpolate flag is set to 1, a Declaration Interpolate token follows.
2363 If file is TGSI_FILE_RESOURCE, a Declaration Resource token follows.
2365 If Array flag is set to 1, a Declaration Array token follows.
2368 ^^^^^^^^^^^^^^^^^^^^^^^^
2370 Declarations can optional have an ArrayID attribute which can be referred by
2371 indirect addressing operands. An ArrayID of zero is reserved and treaded as
2372 if no ArrayID is specified.
2374 If an indirect addressing operand refers to a specific declaration by using
2375 an ArrayID only the registers in this declaration are guaranteed to be
2376 accessed, accessing any register outside this declaration results in undefined
2377 behavior. Note that for compatibility the effective index is zero-based and
2378 not relative to the specified declaration
2380 If no ArrayID is specified with an indirect addressing operand the whole
2381 register file might be accessed by this operand. This is strongly discouraged
2382 and will prevent packing of scalar/vec2 arrays and effective alias analysis.
2384 Declaration Semantic
2385 ^^^^^^^^^^^^^^^^^^^^^^^^
2387 Vertex and fragment shader input and output registers may be labeled
2388 with semantic information consisting of a name and index.
2390 Follows Declaration token if Semantic bit is set.
2392 Since its purpose is to link a shader with other stages of the pipeline,
2393 it is valid to follow only those Declaration tokens that declare a register
2394 either in INPUT or OUTPUT file.
2396 SemanticName field contains the semantic name of the register being declared.
2397 There is no default value.
2399 SemanticIndex is an optional subscript that can be used to distinguish
2400 different register declarations with the same semantic name. The default value
2403 The meanings of the individual semantic names are explained in the following
2406 TGSI_SEMANTIC_POSITION
2407 """"""""""""""""""""""
2409 For vertex shaders, TGSI_SEMANTIC_POSITION indicates the vertex shader
2410 output register which contains the homogeneous vertex position in the clip
2411 space coordinate system. After clipping, the X, Y and Z components of the
2412 vertex will be divided by the W value to get normalized device coordinates.
2414 For fragment shaders, TGSI_SEMANTIC_POSITION is used to indicate that
2415 fragment shader input contains the fragment's window position. The X
2416 component starts at zero and always increases from left to right.
2417 The Y component starts at zero and always increases but Y=0 may either
2418 indicate the top of the window or the bottom depending on the fragment
2419 coordinate origin convention (see TGSI_PROPERTY_FS_COORD_ORIGIN).
2420 The Z coordinate ranges from 0 to 1 to represent depth from the front
2421 to the back of the Z buffer. The W component contains the reciprocol
2422 of the interpolated vertex position W component.
2424 Fragment shaders may also declare an output register with
2425 TGSI_SEMANTIC_POSITION. Only the Z component is writable. This allows
2426 the fragment shader to change the fragment's Z position.
2433 For vertex shader outputs or fragment shader inputs/outputs, this
2434 label indicates that the resister contains an R,G,B,A color.
2436 Several shader inputs/outputs may contain colors so the semantic index
2437 is used to distinguish them. For example, color[0] may be the diffuse
2438 color while color[1] may be the specular color.
2440 This label is needed so that the flat/smooth shading can be applied
2441 to the right interpolants during rasterization.
2445 TGSI_SEMANTIC_BCOLOR
2446 """"""""""""""""""""
2448 Back-facing colors are only used for back-facing polygons, and are only valid
2449 in vertex shader outputs. After rasterization, all polygons are front-facing
2450 and COLOR and BCOLOR end up occupying the same slots in the fragment shader,
2451 so all BCOLORs effectively become regular COLORs in the fragment shader.
2457 Vertex shader inputs and outputs and fragment shader inputs may be
2458 labeled with TGSI_SEMANTIC_FOG to indicate that the register contains
2459 a fog coordinate. Typically, the fragment shader will use the fog coordinate
2460 to compute a fog blend factor which is used to blend the normal fragment color
2461 with a constant fog color. But fog coord really is just an ordinary vec4
2462 register like regular semantics.
2468 Vertex shader input and output registers may be labeled with
2469 TGIS_SEMANTIC_PSIZE to indicate that the register contains a point size
2470 in the form (S, 0, 0, 1). The point size controls the width or diameter
2471 of points for rasterization. This label cannot be used in fragment
2474 When using this semantic, be sure to set the appropriate state in the
2475 :ref:`rasterizer` first.
2478 TGSI_SEMANTIC_TEXCOORD
2479 """"""""""""""""""""""
2481 Only available if PIPE_CAP_TGSI_TEXCOORD is exposed !
2483 Vertex shader outputs and fragment shader inputs may be labeled with
2484 this semantic to make them replaceable by sprite coordinates via the
2485 sprite_coord_enable state in the :ref:`rasterizer`.
2486 The semantic index permitted with this semantic is limited to <= 7.
2488 If the driver does not support TEXCOORD, sprite coordinate replacement
2489 applies to inputs with the GENERIC semantic instead.
2491 The intended use case for this semantic is gl_TexCoord.
2494 TGSI_SEMANTIC_PCOORD
2495 """"""""""""""""""""
2497 Only available if PIPE_CAP_TGSI_TEXCOORD is exposed !
2499 Fragment shader inputs may be labeled with TGSI_SEMANTIC_PCOORD to indicate
2500 that the register contains sprite coordinates in the form (x, y, 0, 1), if
2501 the current primitive is a point and point sprites are enabled. Otherwise,
2502 the contents of the register are undefined.
2504 The intended use case for this semantic is gl_PointCoord.
2507 TGSI_SEMANTIC_GENERIC
2508 """""""""""""""""""""
2510 All vertex/fragment shader inputs/outputs not labeled with any other
2511 semantic label can be considered to be generic attributes. Typical
2512 uses of generic inputs/outputs are texcoords and user-defined values.
2515 TGSI_SEMANTIC_NORMAL
2516 """"""""""""""""""""
2518 Indicates that a vertex shader input is a normal vector. This is
2519 typically only used for legacy graphics APIs.
2525 This label applies to fragment shader inputs only and indicates that
2526 the register contains front/back-face information of the form (F, 0,
2527 0, 1). The first component will be positive when the fragment belongs
2528 to a front-facing polygon, and negative when the fragment belongs to a
2529 back-facing polygon.
2532 TGSI_SEMANTIC_EDGEFLAG
2533 """"""""""""""""""""""
2535 For vertex shaders, this sematic label indicates that an input or
2536 output is a boolean edge flag. The register layout is [F, x, x, x]
2537 where F is 0.0 or 1.0 and x = don't care. Normally, the vertex shader
2538 simply copies the edge flag input to the edgeflag output.
2540 Edge flags are used to control which lines or points are actually
2541 drawn when the polygon mode converts triangles/quads/polygons into
2545 TGSI_SEMANTIC_STENCIL
2546 """""""""""""""""""""
2548 For fragment shaders, this semantic label indicates that an output
2549 is a writable stencil reference value. Only the Y component is writable.
2550 This allows the fragment shader to change the fragments stencilref value.
2553 TGSI_SEMANTIC_VIEWPORT_INDEX
2554 """"""""""""""""""""""""""""
2556 For geometry shaders, this semantic label indicates that an output
2557 contains the index of the viewport (and scissor) to use.
2558 Only the X value is used.
2564 For geometry shaders, this semantic label indicates that an output
2565 contains the layer value to use for the color and depth/stencil surfaces.
2566 Only the X value is used. (Also known as rendertarget array index.)
2569 TGSI_SEMANTIC_CULLDIST
2570 """"""""""""""""""""""
2572 Used as distance to plane for performing application-defined culling
2573 of individual primitives against a plane. When components of vertex
2574 elements are given this label, these values are assumed to be a
2575 float32 signed distance to a plane. Primitives will be completely
2576 discarded if the plane distance for all of the vertices in the
2577 primitive are < 0. If a vertex has a cull distance of NaN, that
2578 vertex counts as "out" (as if its < 0);
2579 The limits on both clip and cull distances are bound
2580 by the PIPE_MAX_CLIP_OR_CULL_DISTANCE_COUNT define which defines
2581 the maximum number of components that can be used to hold the
2582 distances and by the PIPE_MAX_CLIP_OR_CULL_DISTANCE_ELEMENT_COUNT
2583 which specifies the maximum number of registers which can be
2584 annotated with those semantics.
2587 TGSI_SEMANTIC_CLIPDIST
2588 """"""""""""""""""""""
2590 When components of vertex elements are identified this way, these
2591 values are each assumed to be a float32 signed distance to a plane.
2592 Primitive setup only invokes rasterization on pixels for which
2593 the interpolated plane distances are >= 0. Multiple clip planes
2594 can be implemented simultaneously, by annotating multiple
2595 components of one or more vertex elements with the above specified
2596 semantic. The limits on both clip and cull distances are bound
2597 by the PIPE_MAX_CLIP_OR_CULL_DISTANCE_COUNT define which defines
2598 the maximum number of components that can be used to hold the
2599 distances and by the PIPE_MAX_CLIP_OR_CULL_DISTANCE_ELEMENT_COUNT
2600 which specifies the maximum number of registers which can be
2601 annotated with those semantics.
2604 Declaration Interpolate
2605 ^^^^^^^^^^^^^^^^^^^^^^^
2607 This token is only valid for fragment shader INPUT declarations.
2609 The Interpolate field specifes the way input is being interpolated by
2610 the rasteriser and is one of TGSI_INTERPOLATE_*.
2612 The CylindricalWrap bitfield specifies which register components
2613 should be subject to cylindrical wrapping when interpolating by the
2614 rasteriser. If TGSI_CYLINDRICAL_WRAP_X is set to 1, the X component
2615 should be interpolated according to cylindrical wrapping rules.
2618 Declaration Sampler View
2619 ^^^^^^^^^^^^^^^^^^^^^^^^
2621 Follows Declaration token if file is TGSI_FILE_SAMPLER_VIEW.
2623 DCL SVIEW[#], resource, type(s)
2625 Declares a shader input sampler view and assigns it to a SVIEW[#]
2628 resource can be one of BUFFER, 1D, 2D, 3D, 1DArray and 2DArray.
2630 type must be 1 or 4 entries (if specifying on a per-component
2631 level) out of UNORM, SNORM, SINT, UINT and FLOAT.
2634 Declaration Resource
2635 ^^^^^^^^^^^^^^^^^^^^
2637 Follows Declaration token if file is TGSI_FILE_RESOURCE.
2639 DCL RES[#], resource [, WR] [, RAW]
2641 Declares a shader input resource and assigns it to a RES[#]
2644 resource can be one of BUFFER, 1D, 2D, 3D, CUBE, 1DArray and
2647 If the RAW keyword is not specified, the texture data will be
2648 subject to conversion, swizzling and scaling as required to yield
2649 the specified data type from the physical data format of the bound
2652 If the RAW keyword is specified, no channel conversion will be
2653 performed: the values read for each of the channels (X,Y,Z,W) will
2654 correspond to consecutive words in the same order and format
2655 they're found in memory. No element-to-address conversion will be
2656 performed either: the value of the provided X coordinate will be
2657 interpreted in byte units instead of texel units. The result of
2658 accessing a misaligned address is undefined.
2660 Usage of the STORE opcode is only allowed if the WR (writable) flag
2665 ^^^^^^^^^^^^^^^^^^^^^^^^
2668 Properties are general directives that apply to the whole TGSI program.
2673 Specifies the fragment shader TGSI_SEMANTIC_POSITION coordinate origin.
2674 The default value is UPPER_LEFT.
2676 If UPPER_LEFT, the position will be (0,0) at the upper left corner and
2677 increase downward and rightward.
2678 If LOWER_LEFT, the position will be (0,0) at the lower left corner and
2679 increase upward and rightward.
2681 OpenGL defaults to LOWER_LEFT, and is configurable with the
2682 GL_ARB_fragment_coord_conventions extension.
2684 DirectX 9/10 use UPPER_LEFT.
2686 FS_COORD_PIXEL_CENTER
2687 """""""""""""""""""""
2689 Specifies the fragment shader TGSI_SEMANTIC_POSITION pixel center convention.
2690 The default value is HALF_INTEGER.
2692 If HALF_INTEGER, the fractionary part of the position will be 0.5
2693 If INTEGER, the fractionary part of the position will be 0.0
2695 Note that this does not affect the set of fragments generated by
2696 rasterization, which is instead controlled by half_pixel_center in the
2699 OpenGL defaults to HALF_INTEGER, and is configurable with the
2700 GL_ARB_fragment_coord_conventions extension.
2702 DirectX 9 uses INTEGER.
2703 DirectX 10 uses HALF_INTEGER.
2705 FS_COLOR0_WRITES_ALL_CBUFS
2706 """"""""""""""""""""""""""
2707 Specifies that writes to the fragment shader color 0 are replicated to all
2708 bound cbufs. This facilitates OpenGL's fragColor output vs fragData[0] where
2709 fragData is directed to a single color buffer, but fragColor is broadcast.
2712 """"""""""""""""""""""""""
2713 If this property is set on the program bound to the shader stage before the
2714 fragment shader, user clip planes should have no effect (be disabled) even if
2715 that shader does not write to any clip distance outputs and the rasterizer's
2716 clip_plane_enable is non-zero.
2717 This property is only supported by drivers that also support shader clip
2719 This is useful for APIs that don't have UCPs and where clip distances written
2720 by a shader cannot be disabled.
2723 Texture Sampling and Texture Formats
2724 ------------------------------------
2726 This table shows how texture image components are returned as (x,y,z,w) tuples
2727 by TGSI texture instructions, such as :opcode:`TEX`, :opcode:`TXD`, and
2728 :opcode:`TXP`. For reference, OpenGL and Direct3D conventions are shown as
2731 +--------------------+--------------+--------------------+--------------+
2732 | Texture Components | Gallium | OpenGL | Direct3D 9 |
2733 +====================+==============+====================+==============+
2734 | R | (r, 0, 0, 1) | (r, 0, 0, 1) | (r, 1, 1, 1) |
2735 +--------------------+--------------+--------------------+--------------+
2736 | RG | (r, g, 0, 1) | (r, g, 0, 1) | (r, g, 1, 1) |
2737 +--------------------+--------------+--------------------+--------------+
2738 | RGB | (r, g, b, 1) | (r, g, b, 1) | (r, g, b, 1) |
2739 +--------------------+--------------+--------------------+--------------+
2740 | RGBA | (r, g, b, a) | (r, g, b, a) | (r, g, b, a) |
2741 +--------------------+--------------+--------------------+--------------+
2742 | A | (0, 0, 0, a) | (0, 0, 0, a) | (0, 0, 0, a) |
2743 +--------------------+--------------+--------------------+--------------+
2744 | L | (l, l, l, 1) | (l, l, l, 1) | (l, l, l, 1) |
2745 +--------------------+--------------+--------------------+--------------+
2746 | LA | (l, l, l, a) | (l, l, l, a) | (l, l, l, a) |
2747 +--------------------+--------------+--------------------+--------------+
2748 | I | (i, i, i, i) | (i, i, i, i) | N/A |
2749 +--------------------+--------------+--------------------+--------------+
2750 | UV | XXX TBD | (0, 0, 0, 1) | (u, v, 1, 1) |
2751 | | | [#envmap-bumpmap]_ | |
2752 +--------------------+--------------+--------------------+--------------+
2753 | Z | XXX TBD | (z, z, z, 1) | (0, z, 0, 1) |
2754 | | | [#depth-tex-mode]_ | |
2755 +--------------------+--------------+--------------------+--------------+
2756 | S | (s, s, s, s) | unknown | unknown |
2757 +--------------------+--------------+--------------------+--------------+
2759 .. [#envmap-bumpmap] http://www.opengl.org/registry/specs/ATI/envmap_bumpmap.txt
2760 .. [#depth-tex-mode] the default is (z, z, z, 1) but may also be (0, 0, 0, z)
2761 or (z, z, z, z) depending on the value of GL_DEPTH_TEXTURE_MODE.