1 /****************************************************************************
2 * Copyright (C) 2014-2015 Intel Corporation. All Rights Reserved.
4 * Permission is hereby granted, free of charge, to any person obtaining a
5 * copy of this software and associated documentation files (the "Software"),
6 * to deal in the Software without restriction, including without limitation
7 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
8 * and/or sell copies of the Software, and to permit persons to whom the
9 * Software is furnished to do so, subject to the following conditions:
11 * The above copyright notice and this permission notice (including the next
12 * paragraph) shall be included in all copies or substantial portions of the
15 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
18 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
20 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
25 * @brief Definitions for API state.
27 ******************************************************************************/
30 #include "common/formats.h"
31 #include "common/intrin.h"
32 using gfxptr_t
= unsigned long long;
36 //////////////////////////////////////////////////////////////////////////
37 /// PRIMITIVE_TOPOLOGY.
38 //////////////////////////////////////////////////////////////////////////
39 enum PRIMITIVE_TOPOLOGY
45 TOP_TRIANGLE_LIST
= 0x4,
46 TOP_TRIANGLE_STRIP
= 0x5,
47 TOP_TRIANGLE_FAN
= 0x6,
50 TOP_LINE_LIST_ADJ
= 0x9,
51 TOP_LISTSTRIP_ADJ
= 0xA,
52 TOP_TRI_LIST_ADJ
= 0xB,
53 TOP_TRI_STRIP_ADJ
= 0xC,
54 TOP_TRI_STRIP_REVERSE
= 0xD,
58 TOP_POINT_LIST_BF
= 0x11,
59 TOP_LINE_STRIP_CONT
= 0x12,
60 TOP_LINE_STRIP_BF
= 0x13,
61 TOP_LINE_STRIP_CONT_BF
= 0x14,
62 TOP_TRIANGLE_FAN_NOSTIPPLE
= 0x16,
63 TOP_TRIANGLE_DISC
= 0x17, /// @todo What is this??
65 TOP_PATCHLIST_BASE
= 0x1F, // Invalid topology, used to calculate num verts for a patchlist.
66 TOP_PATCHLIST_1
= 0x20, // List of 1-vertex patches
67 TOP_PATCHLIST_2
= 0x21,
68 TOP_PATCHLIST_3
= 0x22,
69 TOP_PATCHLIST_4
= 0x23,
70 TOP_PATCHLIST_5
= 0x24,
71 TOP_PATCHLIST_6
= 0x25,
72 TOP_PATCHLIST_7
= 0x26,
73 TOP_PATCHLIST_8
= 0x27,
74 TOP_PATCHLIST_9
= 0x28,
75 TOP_PATCHLIST_10
= 0x29,
76 TOP_PATCHLIST_11
= 0x2A,
77 TOP_PATCHLIST_12
= 0x2B,
78 TOP_PATCHLIST_13
= 0x2C,
79 TOP_PATCHLIST_14
= 0x2D,
80 TOP_PATCHLIST_15
= 0x2E,
81 TOP_PATCHLIST_16
= 0x2F,
82 TOP_PATCHLIST_17
= 0x30,
83 TOP_PATCHLIST_18
= 0x31,
84 TOP_PATCHLIST_19
= 0x32,
85 TOP_PATCHLIST_20
= 0x33,
86 TOP_PATCHLIST_21
= 0x34,
87 TOP_PATCHLIST_22
= 0x35,
88 TOP_PATCHLIST_23
= 0x36,
89 TOP_PATCHLIST_24
= 0x37,
90 TOP_PATCHLIST_25
= 0x38,
91 TOP_PATCHLIST_26
= 0x39,
92 TOP_PATCHLIST_27
= 0x3A,
93 TOP_PATCHLIST_28
= 0x3B,
94 TOP_PATCHLIST_29
= 0x3C,
95 TOP_PATCHLIST_30
= 0x3D,
96 TOP_PATCHLIST_31
= 0x3E,
97 TOP_PATCHLIST_32
= 0x3F, // List of 32-vertex patches
100 //////////////////////////////////////////////////////////////////////////
102 //////////////////////////////////////////////////////////////////////////
115 //////////////////////////////////////////////////////////////////////////
116 /// SWR_RENDERTARGET_ATTACHMENT
117 /// @todo Its not clear what an "attachment" means. Its not common term.
118 //////////////////////////////////////////////////////////////////////////
119 enum SWR_RENDERTARGET_ATTACHMENT
121 SWR_ATTACHMENT_COLOR0
,
122 SWR_ATTACHMENT_COLOR1
,
123 SWR_ATTACHMENT_COLOR2
,
124 SWR_ATTACHMENT_COLOR3
,
125 SWR_ATTACHMENT_COLOR4
,
126 SWR_ATTACHMENT_COLOR5
,
127 SWR_ATTACHMENT_COLOR6
,
128 SWR_ATTACHMENT_COLOR7
,
129 SWR_ATTACHMENT_DEPTH
,
130 SWR_ATTACHMENT_STENCIL
,
135 #define SWR_NUM_RENDERTARGETS 8
137 #define SWR_ATTACHMENT_COLOR0_BIT 0x001
138 #define SWR_ATTACHMENT_COLOR1_BIT 0x002
139 #define SWR_ATTACHMENT_COLOR2_BIT 0x004
140 #define SWR_ATTACHMENT_COLOR3_BIT 0x008
141 #define SWR_ATTACHMENT_COLOR4_BIT 0x010
142 #define SWR_ATTACHMENT_COLOR5_BIT 0x020
143 #define SWR_ATTACHMENT_COLOR6_BIT 0x040
144 #define SWR_ATTACHMENT_COLOR7_BIT 0x080
145 #define SWR_ATTACHMENT_DEPTH_BIT 0x100
146 #define SWR_ATTACHMENT_STENCIL_BIT 0x200
147 #define SWR_ATTACHMENT_MASK_ALL 0x3ff
148 #define SWR_ATTACHMENT_MASK_COLOR 0x0ff
151 //////////////////////////////////////////////////////////////////////////
152 /// @brief SWR Inner Tessellation factor ID
153 /// See above GetTessFactorOutputPosition code for documentation
154 enum SWR_INNER_TESSFACTOR_ID
156 SWR_QUAD_U_TRI_INSIDE
,
159 SWR_NUM_INNER_TESS_FACTORS
,
162 //////////////////////////////////////////////////////////////////////////
163 /// @brief SWR Outer Tessellation factor ID
164 /// See above GetTessFactorOutputPosition code for documentation
165 enum SWR_OUTER_TESSFACTOR_ID
167 SWR_QUAD_U_EQ0_TRI_U_LINE_DETAIL
,
168 SWR_QUAD_V_EQ0_TRI_V_LINE_DENSITY
,
169 SWR_QUAD_U_EQ1_TRI_W
,
172 SWR_NUM_OUTER_TESS_FACTORS
,
176 /////////////////////////////////////////////////////////////////////////
178 /// @brief Defines a vertex element that holds all the data for SIMD vertices.
179 /// Contains space for position, SGV, and 32 generic attributes
180 /////////////////////////////////////////////////////////////////////////
184 VERTEX_SGV_RTAI_COMP
= 0,
185 VERTEX_SGV_VAI_COMP
= 1,
186 VERTEX_SGV_POINT_SIZE_COMP
= 2,
187 VERTEX_POSITION_SLOT
= 1,
188 VERTEX_POSITION_END_SLOT
= 1,
189 VERTEX_CLIPCULL_DIST_LO_SLOT
= (1 + VERTEX_POSITION_END_SLOT
), // VS writes lower 4 clip/cull dist
190 VERTEX_CLIPCULL_DIST_HI_SLOT
= (2 + VERTEX_POSITION_END_SLOT
), // VS writes upper 4 clip/cull dist
191 VERTEX_ATTRIB_START_SLOT
= (3 + VERTEX_POSITION_END_SLOT
),
192 VERTEX_ATTRIB_END_SLOT
= (34 + VERTEX_POSITION_END_SLOT
),
193 SWR_VTX_NUM_SLOTS
= (1 + VERTEX_ATTRIB_END_SLOT
)
199 simdvector attrib
[SWR_VTX_NUM_SLOTS
];
202 #if ENABLE_AVX512_SIMD16
205 simd16vector attrib
[SWR_VTX_NUM_SLOTS
];
210 template<typename SIMD_T
>
213 typename
SIMD_T::Vec4 attrib
[SWR_VTX_NUM_SLOTS
];
216 //////////////////////////////////////////////////////////////////////////
218 /// @brief Input to vertex shader
219 /////////////////////////////////////////////////////////////////////////
220 struct SWR_VS_CONTEXT
222 simdvertex
* pVin
; // IN: SIMD input vertex data store
223 simdvertex
* pVout
; // OUT: SIMD output vertex data store
225 uint32_t InstanceID
; // IN: Instance ID, constant across all verts of the SIMD
226 simdscalari VertexID
; // IN: Vertex ID
227 simdscalari mask
; // IN: Active mask for shader
228 #if USE_SIMD16_FRONTEND
229 uint32_t AlternateOffset
; // IN: amount to offset for interleaving even/odd simd8 in simd16vertex output
233 /////////////////////////////////////////////////////////////////////////
235 /// @brief defines a control point element as passed from the output
236 /// of the hull shader to the input of the domain shader
237 /////////////////////////////////////////////////////////////////////////
248 ScalarAttrib attrib
[SWR_VTX_NUM_SLOTS
];
251 //////////////////////////////////////////////////////////////////////////
252 /// SWR_TESSELLATION_FACTORS
253 /// @brief Tessellation factors structure (non-vector)
254 /////////////////////////////////////////////////////////////////////////
255 struct SWR_TESSELLATION_FACTORS
257 float OuterTessFactors
[SWR_NUM_OUTER_TESS_FACTORS
];
258 float InnerTessFactors
[SWR_NUM_INNER_TESS_FACTORS
];
261 #define MAX_NUM_VERTS_PER_PRIM 32 // support up to 32 control point patches
264 SWR_TESSELLATION_FACTORS tessFactors
;
265 ScalarCPoint cp
[MAX_NUM_VERTS_PER_PRIM
];
266 ScalarCPoint patchData
;
269 //////////////////////////////////////////////////////////////////////////
271 /// @brief Input to hull shader
272 /////////////////////////////////////////////////////////////////////////
273 struct SWR_HS_CONTEXT
275 simdvertex vert
[MAX_NUM_VERTS_PER_PRIM
]; // IN: (SIMD) input primitive data
276 simdscalari PrimitiveID
; // IN: (SIMD) primitive ID generated from the draw call
277 simdscalari mask
; // IN: Active mask for shader
278 ScalarPatch
* pCPout
; // OUT: Output control point patch
279 // SIMD-sized-array of SCALAR patches
282 //////////////////////////////////////////////////////////////////////////
284 /// @brief Input to domain shader
285 /////////////////////////////////////////////////////////////////////////
286 struct SWR_DS_CONTEXT
288 uint32_t PrimitiveID
; // IN: (SCALAR) PrimitiveID for the patch associated with the DS invocation
289 uint32_t vectorOffset
; // IN: (SCALAR) vector index offset into SIMD data.
290 uint32_t vectorStride
; // IN: (SCALAR) stride (in vectors) of output data per attribute-component
291 ScalarPatch
* pCpIn
; // IN: (SCALAR) Control patch
292 simdscalar
* pDomainU
; // IN: (SIMD) Domain Point U coords
293 simdscalar
* pDomainV
; // IN: (SIMD) Domain Point V coords
294 simdscalari mask
; // IN: Active mask for shader
295 simdscalar
* pOutputData
; // OUT: (SIMD) Vertex Attributes (2D array of vectors, one row per attribute-component)
298 //////////////////////////////////////////////////////////////////////////
300 /// @brief Input to geometry shader.
301 /////////////////////////////////////////////////////////////////////////
302 struct SWR_GS_CONTEXT
304 simdvector
* pVerts
; // IN: input primitive data for SIMD prims
305 uint32_t inputVertStride
; // IN: input vertex stride, in attributes
306 simdscalari PrimitiveID
; // IN: input primitive ID generated from the draw call
307 uint32_t InstanceID
; // IN: input instance ID
308 simdscalari mask
; // IN: Active mask for shader
309 uint8_t* pStreams
[KNOB_SIMD_WIDTH
]; // OUT: output stream (contains vertices for all output streams)
312 struct PixelPositions
320 #define SWR_MAX_NUM_MULTISAMPLES 16
322 //////////////////////////////////////////////////////////////////////////
324 /// @brief Input to pixel shader.
325 /////////////////////////////////////////////////////////////////////////
326 struct SWR_PS_CONTEXT
328 PixelPositions vX
; // IN: x location(s) of pixels
329 PixelPositions vY
; // IN: x location(s) of pixels
330 simdscalar vZ
; // INOUT: z location of pixels
331 simdscalari activeMask
; // OUT: mask for kill
332 simdscalar inputMask
; // IN: input coverage mask for all samples
333 simdscalari oMask
; // OUT: mask for output coverage
335 PixelPositions vI
; // barycentric coords evaluated at pixel center, sample position, centroid
337 PixelPositions vOneOverW
; // IN: 1/w
339 const float* pAttribs
; // IN: pointer to attribute barycentric coefficients
340 const float* pPerspAttribs
; // IN: pointer to attribute/w barycentric coefficients
341 const float* pRecipW
; // IN: pointer to 1/w coord for each vertex
342 const float *I
; // IN: Barycentric A, B, and C coefs used to compute I
343 const float *J
; // IN: Barycentric A, B, and C coefs used to compute J
344 float recipDet
; // IN: 1/Det, used when barycentric interpolating attributes
345 const float* pSamplePosX
; // IN: array of sample positions
346 const float* pSamplePosY
; // IN: array of sample positions
347 simdvector shaded
[SWR_NUM_RENDERTARGETS
];
348 // OUT: result color per rendertarget
350 uint32_t frontFace
; // IN: front- 1, back- 0
351 uint32_t sampleIndex
; // IN: sampleIndex
352 uint32_t renderTargetArrayIndex
; // IN: render target array index from GS
353 uint32_t rasterizerSampleCount
; // IN: sample count used by the rasterizer
355 uint8_t* pColorBuffer
[SWR_NUM_RENDERTARGETS
]; // IN: Pointers to render target hottiles
358 //////////////////////////////////////////////////////////////////////////
360 /// @brief Input to compute shader.
361 /////////////////////////////////////////////////////////////////////////
362 struct SWR_CS_CONTEXT
364 // The ThreadGroupId is the current thread group index relative
365 // to all thread groups in the Dispatch call. The ThreadId, ThreadIdInGroup,
366 // and ThreadIdInGroupFlattened can be derived from ThreadGroupId in the shader.
368 // Compute shader accepts the following system values.
369 // o ThreadId - Current thread id relative to all other threads in dispatch.
370 // o ThreadGroupId - Current thread group id relative to all other groups in dispatch.
371 // o ThreadIdInGroup - Current thread relative to all threads in the current thread group.
372 // o ThreadIdInGroupFlattened - Flattened linear id derived from ThreadIdInGroup.
374 // All of these system values can be computed in the shader. They will be
375 // derived from the current tile counter. The tile counter is an atomic counter that
376 // resides in the draw context and is initialized to the product of the dispatch dims.
378 // tileCounter = dispatchDims.x * dispatchDims.y * dispatchDims.z
380 // Each CPU worker thread will atomically decrement this counter and passes the current
381 // count into the shader. When the count reaches 0 then all thread groups in the
382 // dispatch call have been completed.
384 uint32_t tileCounter
; // The tile counter value for this thread group.
386 // Dispatch dimensions used by shader to compute system values from the tile counter.
387 uint32_t dispatchDims
[3];
389 uint8_t* pTGSM
; // Thread Group Shared Memory pointer.
391 uint8_t* pSpillFillBuffer
; // Spill/fill buffer for barrier support
393 uint8_t* pScratchSpace
; // Pointer to scratch space buffer used by the shader, shader is responsible
394 // for subdividing scratch space per instance/simd
396 uint32_t scratchSpacePerSimd
; // Scratch space per work item x SIMD_WIDTH
402 SWR_TILE_NONE
= 0x0, // Linear mode (no tiling)
403 SWR_TILE_MODE_WMAJOR
, // W major tiling
404 SWR_TILE_MODE_XMAJOR
, // X major tiling
405 SWR_TILE_MODE_YMAJOR
, // Y major tiling
406 SWR_TILE_SWRZ
, // SWR-Z tiling
411 enum SWR_SURFACE_TYPE
418 SURFACE_STRUCTURED_BUFFER
= 5,
447 enum SWR_BLEND_FACTOR
450 BLENDFACTOR_SRC_COLOR
,
451 BLENDFACTOR_SRC_ALPHA
,
452 BLENDFACTOR_DST_ALPHA
,
453 BLENDFACTOR_DST_COLOR
,
454 BLENDFACTOR_SRC_ALPHA_SATURATE
,
455 BLENDFACTOR_CONST_COLOR
,
456 BLENDFACTOR_CONST_ALPHA
,
457 BLENDFACTOR_SRC1_COLOR
,
458 BLENDFACTOR_SRC1_ALPHA
,
460 BLENDFACTOR_INV_SRC_COLOR
,
461 BLENDFACTOR_INV_SRC_ALPHA
,
462 BLENDFACTOR_INV_DST_ALPHA
,
463 BLENDFACTOR_INV_DST_COLOR
,
464 BLENDFACTOR_INV_CONST_COLOR
,
465 BLENDFACTOR_INV_CONST_ALPHA
,
466 BLENDFACTOR_INV_SRC1_COLOR
,
467 BLENDFACTOR_INV_SRC1_ALPHA
483 LOGICOP_AND_INVERTED
,
484 LOGICOP_COPY_INVERTED
,
499 //////////////////////////////////////////////////////////////////////////
501 /// @brief Specifies how the auxiliary buffer is used by the driver.
502 //////////////////////////////////////////////////////////////////////////
511 //////////////////////////////////////////////////////////////////////////
512 /// SWR_SURFACE_STATE
513 //////////////////////////////////////////////////////////////////////////
514 struct SWR_SURFACE_STATE
516 gfxptr_t xpBaseAddress
;
517 SWR_SURFACE_TYPE type
; // @llvm_enum
518 SWR_FORMAT format
; // @llvm_enum
523 uint32_t samplePattern
;
526 uint32_t minLod
; // for sampled surfaces, the most detailed LOD that can be accessed by sampler
527 uint32_t maxLod
; // for sampled surfaces, the max LOD that can be accessed
528 float resourceMinLod
; // for sampled surfaces, the most detailed fractional mip that can be accessed by sampler
529 uint32_t lod
; // for render targets, the lod being rendered to
530 uint32_t arrayIndex
; // for render targets, the array index being rendered to for arrayed surfaces
531 SWR_TILE_MODE tileMode
; // @llvm_enum
537 uint32_t lodOffsets
[2][15]; // lod offsets for sampled surfaces
539 gfxptr_t xpAuxBaseAddress
; // Used for compression, append/consume counter, etc.
540 SWR_AUX_MODE auxMode
; // @llvm_enum
543 bool bInterleavedSamples
; // are MSAA samples stored interleaved or planar
546 // vertex fetch state
547 // WARNING- any changes to this struct need to be reflected
548 // in the fetch shader jit
549 struct SWR_VERTEX_BUFFER_STATE
553 const uint8_t *pData
;
556 uint32_t minVertex
; // min vertex (for bounds checking)
557 uint32_t maxVertex
; // size / pitch. precalculated value used by fetch shader for OOB checks
558 uint32_t partialInboundsSize
; // size % pitch. precalculated value used by fetch shader for partially OOB vertices
561 struct SWR_INDEX_BUFFER_STATE
563 // Format type for indices (e.g. UINT16, UINT32, etc.)
564 SWR_FORMAT format
; // @llvm_enum
565 const void *pIndices
;
570 //////////////////////////////////////////////////////////////////////////
571 /// SWR_FETCH_CONTEXT
572 /// @brief Input to fetch shader.
573 /// @note WARNING - Changes to this struct need to be reflected in the
574 /// fetch shader jit.
575 /////////////////////////////////////////////////////////////////////////
576 struct SWR_FETCH_CONTEXT
578 const SWR_VERTEX_BUFFER_STATE
* pStreams
; // IN: array of bound vertex buffers
579 const int32_t* pIndices
; // IN: pointer to index buffer for indexed draws
580 const int32_t* pLastIndex
; // IN: pointer to end of index buffer, used for bounds checking
581 uint32_t CurInstance
; // IN: current instance
582 uint32_t BaseVertex
; // IN: base vertex
583 uint32_t StartVertex
; // IN: start vertex
584 uint32_t StartInstance
; // IN: start instance
585 simdscalari VertexID
; // OUT: vector of vertex IDs
586 simdscalari CutMask
; // OUT: vector mask of indices which have the cut index value
587 #if USE_SIMD16_SHADERS
588 // simd16scalari VertexID; // OUT: vector of vertex IDs
589 // simd16scalari CutMask; // OUT: vector mask of indices which have the cut index value
590 simdscalari VertexID2
; // OUT: vector of vertex IDs
591 simdscalari CutMask2
; // OUT: vector mask of indices which have the cut index value
595 //////////////////////////////////////////////////////////////////////////
598 /// @brief All statistics generated by SWR go here. These are public
600 /////////////////////////////////////////////////////////////////////////
601 OSALIGNLINE(struct) SWR_STATS
604 uint64_t DepthPassCount
; // Number of passing depth tests. Not exact.
607 uint64_t PsInvocations
; // Number of Pixel Shader invocations
608 uint64_t CsInvocations
; // Number of Compute Shader invocations
612 //////////////////////////////////////////////////////////////////////////
615 /// @brief All statistics generated by FE.
616 /////////////////////////////////////////////////////////////////////////
617 OSALIGNLINE(struct) SWR_STATS_FE
619 uint64_t IaVertices
; // Number of Fetch Shader vertices
620 uint64_t IaPrimitives
; // Number of PA primitives.
621 uint64_t VsInvocations
; // Number of Vertex Shader invocations
622 uint64_t HsInvocations
; // Number of Hull Shader invocations
623 uint64_t DsInvocations
; // Number of Domain Shader invocations
624 uint64_t GsInvocations
; // Number of Geometry Shader invocations
625 uint64_t GsPrimitives
; // Number of prims GS outputs.
626 uint64_t CInvocations
; // Number of clipper invocations
627 uint64_t CPrimitives
; // Number of clipper primitives.
630 uint64_t SoPrimStorageNeeded
[4];
631 uint64_t SoNumPrimsWritten
[4];
634 //////////////////////////////////////////////////////////////////////////
635 /// STREAMOUT_BUFFERS
636 /////////////////////////////////////////////////////////////////////////
638 #define MAX_SO_STREAMS 4
639 #define MAX_SO_BUFFERS 4
640 #define MAX_ATTRIBUTES 32
642 struct SWR_STREAMOUT_BUFFER
647 // Pointers to streamout buffers.
650 // Size of buffer in dwords.
653 // Vertex pitch of buffer in dwords.
656 // Offset into buffer in dwords. SOS will increment this offset.
657 uint32_t streamOffset
;
659 // Offset to the SO write offset. If not null then we update offset here.
660 uint32_t* pWriteOffset
;
664 //////////////////////////////////////////////////////////////////////////
666 /////////////////////////////////////////////////////////////////////////
667 struct SWR_STREAMOUT_STATE
669 // This disables stream output.
672 // which streams are enabled for streamout
673 bool streamEnable
[MAX_SO_STREAMS
];
675 // If set then do not send any streams to the rasterizer.
676 bool rasterizerDisable
;
678 // Specifies which stream to send to the rasterizer.
679 uint32_t streamToRasterizer
;
681 // The stream masks specify which attributes are sent to which streams.
682 // These masks help the FE to setup the pPrimData buffer that is passed
683 // the Stream Output Shader (SOS) function.
684 uint32_t streamMasks
[MAX_SO_STREAMS
];
686 // Number of attributes, including position, per vertex that are streamed out.
687 // This should match number of bits in stream mask.
688 uint32_t streamNumEntries
[MAX_SO_STREAMS
];
690 // Offset to the start of the attributes of the input vertices, in simdvector units
691 uint32_t vertexAttribOffset
[MAX_SO_STREAMS
];
694 //////////////////////////////////////////////////////////////////////////
695 /// STREAMOUT_CONTEXT - Passed to SOS
696 /////////////////////////////////////////////////////////////////////////
697 struct SWR_STREAMOUT_CONTEXT
700 SWR_STREAMOUT_BUFFER
* pBuffer
[MAX_SO_STREAMS
];
702 // Num prims written for this stream
703 uint32_t numPrimsWritten
;
705 // Num prims that should have been written if there were no overflow.
706 uint32_t numPrimStorageNeeded
;
709 //////////////////////////////////////////////////////////////////////////
710 /// SWR_GS_STATE - Geometry shader state
711 /////////////////////////////////////////////////////////////////////////
716 // Number of input attributes per vertex. Used by the frontend to
717 // optimize assembling primitives for GS
718 uint32_t numInputAttribs
;
720 // Stride of incoming verts in attributes
721 uint32_t inputVertStride
;
723 // Output topology - can be point, tristrip, or linestrip
724 PRIMITIVE_TOPOLOGY outputTopology
; // @llvm_enum
726 // Maximum number of verts that can be emitted by a single instance of the GS
727 uint32_t maxNumVerts
;
730 uint32_t instanceCount
;
732 // If true, geometry shader emits a single stream, with separate cut buffer.
733 // If false, geometry shader emits vertices for multiple streams to the stream buffer, with a separate StreamID buffer
734 // to map vertices to streams
737 // When single stream is enabled, singleStreamID dictates which stream is being output.
738 // field ignored if isSingleStream is false
739 uint32_t singleStreamID
;
741 // Total amount of memory to allocate for one instance of the shader output in bytes
742 uint32_t allocationSize
;
744 // Offset to the start of the attributes of the input vertices, in simdvector units, as read by the GS
745 uint32_t vertexAttribOffset
;
747 // Offset to the attributes as stored by the preceding shader stage.
748 uint32_t srcVertexAttribOffset
;
750 // Size of the control data section which contains cut or streamID data, in simdscalar units. Should be sized to handle
751 // the maximum number of verts output by the GS. Can be 0 if there are no cuts or streamID bits.
752 uint32_t controlDataSize
;
754 // Offset to the control data section, in bytes
755 uint32_t controlDataOffset
;
757 // Total size of an output vertex, in simdvector units
758 uint32_t outputVertexSize
;
760 // Offset to the start of the vertex section, in bytes
761 uint32_t outputVertexOffset
;
763 // Set this to non-zero to indicate that the shader outputs a static number of verts. If zero, shader is
764 // expected to store the final vertex count in the first dword of the gs output stream.
765 uint32_t staticVertexCount
;
769 //////////////////////////////////////////////////////////////////////////
770 /// SWR_TS_OUTPUT_TOPOLOGY - Defines data output by the tessellator / DS
771 /////////////////////////////////////////////////////////////////////////
772 enum SWR_TS_OUTPUT_TOPOLOGY
776 SWR_TS_OUTPUT_TRI_CW
,
777 SWR_TS_OUTPUT_TRI_CCW
,
779 SWR_TS_OUTPUT_TOPOLOGY_COUNT
782 //////////////////////////////////////////////////////////////////////////
783 /// SWR_TS_PARTITIONING - Defines tessellation algorithm
784 /////////////////////////////////////////////////////////////////////////
785 enum SWR_TS_PARTITIONING
788 SWR_TS_ODD_FRACTIONAL
,
789 SWR_TS_EVEN_FRACTIONAL
,
791 SWR_TS_PARTITIONING_COUNT
794 //////////////////////////////////////////////////////////////////////////
795 /// SWR_TS_DOMAIN - Defines Tessellation Domain
796 /////////////////////////////////////////////////////////////////////////
806 //////////////////////////////////////////////////////////////////////////
807 /// SWR_TS_STATE - Tessellation state
808 /////////////////////////////////////////////////////////////////////////
812 SWR_TS_OUTPUT_TOPOLOGY tsOutputTopology
; // @llvm_enum
813 SWR_TS_PARTITIONING partitioning
; // @llvm_enum
814 SWR_TS_DOMAIN domain
; // @llvm_enum
816 PRIMITIVE_TOPOLOGY postDSTopology
; // @llvm_enum
818 uint32_t numHsInputAttribs
;
819 uint32_t numHsOutputAttribs
;
820 uint32_t numDsOutputAttribs
;
822 // Offset to the start of the attributes of the input vertices, in simdvector units
823 uint32_t vertexAttribOffset
;
826 // output merger state
827 struct SWR_RENDER_TARGET_BLEND_STATE
829 uint8_t writeDisableRed
: 1;
830 uint8_t writeDisableGreen
: 1;
831 uint8_t writeDisableBlue
: 1;
832 uint8_t writeDisableAlpha
: 1;
834 static_assert(sizeof(SWR_RENDER_TARGET_BLEND_STATE
) == 1, "Invalid SWR_RENDER_TARGET_BLEND_STATE size");
836 enum SWR_MULTISAMPLE_COUNT
838 SWR_MULTISAMPLE_1X
= 0,
843 SWR_MULTISAMPLE_TYPE_COUNT
846 INLINE
uint32_t GetNumSamples(SWR_MULTISAMPLE_COUNT sampleCount
) // @llvm_func_start
848 static const uint32_t sampleCountLUT
[SWR_MULTISAMPLE_TYPE_COUNT
] {1, 2, 4, 8, 16};
849 assert(sampleCount
< SWR_MULTISAMPLE_TYPE_COUNT
);
850 return sampleCountLUT
[sampleCount
];
853 struct SWR_BLEND_STATE
855 // constant blend factor color in RGBA float
856 float constantColor
[4];
858 // alpha test reference value in unorm8 or float32
859 uint32_t alphaTestReference
;
861 // all RT's have the same sample count
862 ///@todo move this to Output Merger state when we refactor
863 SWR_MULTISAMPLE_COUNT sampleCount
; // @llvm_enum
865 SWR_RENDER_TARGET_BLEND_STATE renderTarget
[SWR_NUM_RENDERTARGETS
];
867 static_assert(sizeof(SWR_BLEND_STATE
) == 36, "Invalid SWR_BLEND_STATE size");
869 //////////////////////////////////////////////////////////////////////////
870 /// FUNCTION POINTERS FOR SHADERS
872 #if USE_SIMD16_SHADERS
873 typedef void(__cdecl
*PFN_FETCH_FUNC
)(SWR_FETCH_CONTEXT
& fetchInfo
, simd16vertex
& out
);
875 typedef void(__cdecl
*PFN_FETCH_FUNC
)(SWR_FETCH_CONTEXT
& fetchInfo
, simdvertex
& out
);
877 typedef void(__cdecl
*PFN_VERTEX_FUNC
)(HANDLE hPrivateData
, SWR_VS_CONTEXT
* pVsContext
);
878 typedef void(__cdecl
*PFN_HS_FUNC
)(HANDLE hPrivateData
, SWR_HS_CONTEXT
* pHsContext
);
879 typedef void(__cdecl
*PFN_DS_FUNC
)(HANDLE hPrivateData
, SWR_DS_CONTEXT
* pDsContext
);
880 typedef void(__cdecl
*PFN_GS_FUNC
)(HANDLE hPrivateData
, SWR_GS_CONTEXT
* pGsContext
);
881 typedef void(__cdecl
*PFN_CS_FUNC
)(HANDLE hPrivateData
, SWR_CS_CONTEXT
* pCsContext
);
882 typedef void(__cdecl
*PFN_SO_FUNC
)(SWR_STREAMOUT_CONTEXT
& soContext
);
883 typedef void(__cdecl
*PFN_PIXEL_KERNEL
)(HANDLE hPrivateData
, SWR_PS_CONTEXT
*pContext
);
884 typedef void(__cdecl
*PFN_CPIXEL_KERNEL
)(HANDLE hPrivateData
, SWR_PS_CONTEXT
*pContext
);
885 typedef void(__cdecl
*PFN_BLEND_JIT_FUNC
)(const SWR_BLEND_STATE
*,
886 simdvector
& vSrc
, simdvector
& vSrc1
, simdscalar
& vSrc0Alpha
, uint32_t sample
,
887 uint8_t* pDst
, simdvector
& vResult
, simdscalari
* vOMask
, simdscalari
* vCoverageMask
);
888 typedef simdscalar(*PFN_QUANTIZE_DEPTH
)(simdscalar
const &);
892 //////////////////////////////////////////////////////////////////////////
894 /////////////////////////////////////////////////////////////////////////
895 struct SWR_FRONTEND_STATE
897 // skip clip test, perspective divide, and viewport transform
898 // intended for verts in screen space
899 bool vpTransformDisable
;
900 bool bEnableCutIndex
;
906 uint32_t lineStripList
: 1;
907 uint32_t triStripList
: 2;
911 uint32_t topologyProvokingVertex
; // provoking vertex for the draw topology
913 // Size of a vertex in simdvector units. Should be sized to the
914 // maximum of the input/output of the vertex shader.
915 uint32_t vsVertexSize
;
918 //////////////////////////////////////////////////////////////////////////
920 /////////////////////////////////////////////////////////////////////////
921 struct SWR_VIEWPORT_MATRIX
931 //////////////////////////////////////////////////////////////////////////
932 /// VIEWPORT_MATRIXES
933 /////////////////////////////////////////////////////////////////////////
934 struct SWR_VIEWPORT_MATRICES
936 float m00
[KNOB_NUM_VIEWPORTS_SCISSORS
];
937 float m11
[KNOB_NUM_VIEWPORTS_SCISSORS
];
938 float m22
[KNOB_NUM_VIEWPORTS_SCISSORS
];
939 float m30
[KNOB_NUM_VIEWPORTS_SCISSORS
];
940 float m31
[KNOB_NUM_VIEWPORTS_SCISSORS
];
941 float m32
[KNOB_NUM_VIEWPORTS_SCISSORS
];
944 //////////////////////////////////////////////////////////////////////////
946 /////////////////////////////////////////////////////////////////////////
957 //////////////////////////////////////////////////////////////////////////
959 //////////////////////////////////////////////////////////////////////////
971 SWR_FILLMODE_WIREFRAME
,
975 enum SWR_FRONTWINDING
982 enum SWR_PIXEL_LOCATION
984 SWR_PIXEL_LOCATION_CENTER
,
985 SWR_PIXEL_LOCATION_UL
,
988 // fixed point screen space sample locations within a pixel
989 struct SWR_MULTISAMPLE_POS
992 INLINE
void SetXi(uint32_t sampleNum
, uint32_t val
) { _xi
[sampleNum
] = val
; }; // @llvm_func
993 INLINE
void SetYi(uint32_t sampleNum
, uint32_t val
) { _yi
[sampleNum
] = val
; }; // @llvm_func
994 INLINE
uint32_t Xi(uint32_t sampleNum
) const { return _xi
[sampleNum
]; }; // @llvm_func
995 INLINE
uint32_t Yi(uint32_t sampleNum
) const { return _yi
[sampleNum
]; }; // @llvm_func
996 INLINE
void SetX(uint32_t sampleNum
, float val
) { _x
[sampleNum
] = val
; }; // @llvm_func
997 INLINE
void SetY(uint32_t sampleNum
, float val
) { _y
[sampleNum
] = val
; }; // @llvm_func
998 INLINE
float X(uint32_t sampleNum
) const { return _x
[sampleNum
]; }; // @llvm_func
999 INLINE
float Y(uint32_t sampleNum
) const { return _y
[sampleNum
]; }; // @llvm_func
1000 typedef const float(&sampleArrayT
)[SWR_MAX_NUM_MULTISAMPLES
]; //@llvm_typedef
1001 INLINE sampleArrayT
X() const { return _x
; }; // @llvm_func
1002 INLINE sampleArrayT
Y() const { return _y
; }; // @llvm_func
1003 INLINE
const __m128i
& vXi(uint32_t sampleNum
) const { return _vXi
[sampleNum
]; }; // @llvm_func
1004 INLINE
const __m128i
& vYi(uint32_t sampleNum
) const { return _vYi
[sampleNum
]; }; // @llvm_func
1005 INLINE
const simdscalar
& vX(uint32_t sampleNum
) const { return _vX
[sampleNum
]; }; // @llvm_func
1006 INLINE
const simdscalar
& vY(uint32_t sampleNum
) const { return _vY
[sampleNum
]; }; // @llvm_func
1007 INLINE
const __m128i
& TileSampleOffsetsX() const { return tileSampleOffsetsX
; }; // @llvm_func
1008 INLINE
const __m128i
& TileSampleOffsetsY() const { return tileSampleOffsetsY
; }; // @llvm_func
1010 INLINE
void PrecalcSampleData(int numSamples
); //@llvm_func
1013 template <typename MaskT
>
1014 INLINE __m128i
expandThenBlend4(uint32_t* min
, uint32_t* max
); // @llvm_func
1015 INLINE
void CalcTileSampleOffsets(int numSamples
); // @llvm_func
1017 // scalar sample values
1018 uint32_t _xi
[SWR_MAX_NUM_MULTISAMPLES
];
1019 uint32_t _yi
[SWR_MAX_NUM_MULTISAMPLES
];
1020 float _x
[SWR_MAX_NUM_MULTISAMPLES
];
1021 float _y
[SWR_MAX_NUM_MULTISAMPLES
];
1023 // precalc'd / vectorized samples
1024 __m128i _vXi
[SWR_MAX_NUM_MULTISAMPLES
];
1025 __m128i _vYi
[SWR_MAX_NUM_MULTISAMPLES
];
1026 simdscalar _vX
[SWR_MAX_NUM_MULTISAMPLES
];
1027 simdscalar _vY
[SWR_MAX_NUM_MULTISAMPLES
];
1028 __m128i tileSampleOffsetsX
;
1029 __m128i tileSampleOffsetsY
;
1032 //////////////////////////////////////////////////////////////////////////
1034 //////////////////////////////////////////////////////////////////////////
1035 struct SWR_RASTSTATE
1037 uint32_t cullMode
: 2;
1038 uint32_t fillMode
: 2;
1039 uint32_t frontWinding
: 1;
1040 uint32_t scissorEnable
: 1;
1041 uint32_t depthClipEnable
: 1;
1042 uint32_t clipHalfZ
: 1;
1043 uint32_t pointParam
: 1;
1044 uint32_t pointSpriteEnable
: 1;
1045 uint32_t pointSpriteTopOrigin
: 1;
1046 uint32_t forcedSampleCount
: 1;
1047 uint32_t pixelOffset
: 1;
1048 uint32_t depthBiasPreAdjusted
: 1; ///< depth bias constant is in float units, not per-format Z units
1049 uint32_t conservativeRast
: 1;
1055 float slopeScaledDepthBias
;
1056 float depthBiasClamp
;
1057 SWR_FORMAT depthFormat
; // @llvm_enum
1059 // sample count the rasterizer is running at
1060 SWR_MULTISAMPLE_COUNT sampleCount
; // @llvm_enum
1061 uint32_t pixelLocation
; // UL or Center
1062 SWR_MULTISAMPLE_POS samplePositions
; // @llvm_struct
1063 bool bIsCenterPattern
; // @llvm_enum
1067 enum SWR_CONSTANT_SOURCE
1069 SWR_CONSTANT_SOURCE_CONST_0000
,
1070 SWR_CONSTANT_SOURCE_CONST_0001_FLOAT
,
1071 SWR_CONSTANT_SOURCE_CONST_1111_FLOAT
,
1072 SWR_CONSTANT_SOURCE_PRIM_ID
1075 struct SWR_ATTRIB_SWIZZLE
1077 uint16_t sourceAttrib
: 5; // source attribute
1078 uint16_t constantSource
: 2; // constant source to apply
1079 uint16_t componentOverrideMask
: 4; // override component with constant source
1083 struct SWR_BACKEND_STATE
1085 uint32_t constantInterpolationMask
; // bitmask indicating which attributes have constant interpolation
1086 uint32_t pointSpriteTexCoordMask
; // bitmask indicating the attribute(s) which should be interpreted as tex coordinates
1088 uint8_t numAttributes
; // total number of attributes to send to backend (up to 32)
1089 uint8_t numComponents
[32]; // number of components to setup per attribute, this reduces some calculations for unneeded components
1091 bool swizzleEnable
; // when enabled, core will parse the swizzle map when
1092 // setting up attributes for the backend, otherwise
1093 // all attributes up to numAttributes will be sent
1094 SWR_ATTRIB_SWIZZLE swizzleMap
[32];
1096 bool readRenderTargetArrayIndex
; // Forward render target array index from last FE stage to the backend
1097 bool readViewportArrayIndex
; // Read viewport array index from last FE stage during binning
1099 // Offset to the start of the attributes of the input vertices, in simdvector units
1100 uint32_t vertexAttribOffset
;
1102 // User clip/cull distance enables
1103 uint8_t cullDistanceMask
;
1104 uint8_t clipDistanceMask
;
1106 // Offset to clip/cull attrib section of the vertex, in simdvector units
1107 uint32_t vertexClipCullOffset
;
1111 union SWR_DEPTH_STENCIL_STATE
1116 uint32_t depthWriteEnable
: 1;
1117 uint32_t depthTestEnable
: 1;
1118 uint32_t stencilWriteEnable
: 1;
1119 uint32_t stencilTestEnable
: 1;
1120 uint32_t doubleSidedStencilTestEnable
: 1;
1122 uint32_t depthTestFunc
: 3;
1123 uint32_t stencilTestFunc
: 3;
1125 uint32_t backfaceStencilPassDepthPassOp
: 3;
1126 uint32_t backfaceStencilPassDepthFailOp
: 3;
1127 uint32_t backfaceStencilFailOp
: 3;
1128 uint32_t backfaceStencilTestFunc
: 3;
1129 uint32_t stencilPassDepthPassOp
: 3;
1130 uint32_t stencilPassDepthFailOp
: 3;
1131 uint32_t stencilFailOp
: 3;
1134 uint8_t backfaceStencilWriteMask
;
1135 uint8_t backfaceStencilTestMask
;
1136 uint8_t stencilWriteMask
;
1137 uint8_t stencilTestMask
;
1140 uint8_t backfaceStencilRefValue
;
1141 uint8_t stencilRefValue
;
1146 enum SWR_SHADING_RATE
1148 SWR_SHADING_RATE_PIXEL
,
1149 SWR_SHADING_RATE_SAMPLE
,
1150 SWR_SHADING_RATE_COUNT
,
1153 enum SWR_INPUT_COVERAGE
1155 SWR_INPUT_COVERAGE_NONE
,
1156 SWR_INPUT_COVERAGE_NORMAL
,
1157 SWR_INPUT_COVERAGE_INNER_CONSERVATIVE
,
1158 SWR_INPUT_COVERAGE_COUNT
,
1161 enum SWR_PS_POSITION_OFFSET
1163 SWR_PS_POSITION_SAMPLE_NONE
,
1164 SWR_PS_POSITION_SAMPLE_OFFSET
,
1165 SWR_PS_POSITION_CENTROID_OFFSET
,
1166 SWR_PS_POSITION_OFFSET_COUNT
,
1169 enum SWR_BARYCENTRICS_MASK
1171 SWR_BARYCENTRIC_PER_PIXEL_MASK
= 0x1,
1172 SWR_BARYCENTRIC_CENTROID_MASK
= 0x2,
1173 SWR_BARYCENTRIC_PER_SAMPLE_MASK
= 0x4,
1176 // pixel shader state
1180 PFN_PIXEL_KERNEL pfnPixelShader
; // @llvm_pfn
1183 uint32_t killsPixel
: 1; // pixel shader can kill pixels
1184 uint32_t inputCoverage
: 2; // ps uses input coverage
1185 uint32_t writesODepth
: 1; // pixel shader writes to depth
1186 uint32_t usesSourceDepth
: 1; // pixel shader reads depth
1187 uint32_t shadingRate
: 2; // shading per pixel / sample / coarse pixel
1188 uint32_t posOffset
: 2; // type of offset (none, sample, centroid) to add to pixel position
1189 uint32_t barycentricsMask
: 3; // which type(s) of barycentric coords does the PS interpolate attributes with
1190 uint32_t usesUAV
: 1; // pixel shader accesses UAV
1191 uint32_t forceEarlyZ
: 1; // force execution of early depth/stencil test
1193 uint8_t renderTargetMask
; // Mask of render targets written
1196 // depth bounds state
1197 struct SWR_DEPTH_BOUNDS_STATE
1199 bool depthBoundsTestEnable
;
1200 float depthBoundsTestMinValue
;
1201 float depthBoundsTestMaxValue
;