1 /****************************************************************************
2 * Copyright (C) 2014-2018 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 Structure passed to shader for stats collection.
219 /////////////////////////////////////////////////////////////////////////
220 struct SWR_SHADER_STATS
222 uint32_t numInstExecuted
; // This is roughly the API instructions executed and not x86.
225 //////////////////////////////////////////////////////////////////////////
227 /// @brief Input to vertex shader
228 /////////////////////////////////////////////////////////////////////////
229 struct SWR_VS_CONTEXT
231 simdvertex
* pVin
; // IN: SIMD input vertex data store
232 simdvertex
* pVout
; // OUT: SIMD output vertex data store
234 uint32_t InstanceID
; // IN: Instance ID, constant across all verts of the SIMD
235 simdscalari VertexID
; // IN: Vertex ID
236 simdscalari mask
; // IN: Active mask for shader
238 // SIMD16 Frontend fields.
239 uint32_t AlternateOffset
; // IN: amount to offset for interleaving even/odd simd8 in simd16vertex output
240 simd16scalari mask16
; // IN: Active mask for shader (16-wide)
241 simd16scalari VertexID16
; // IN: Vertex ID (16-wide)
243 SWR_SHADER_STATS stats
; // OUT: shader statistics used for archrast.
246 /////////////////////////////////////////////////////////////////////////
248 /// @brief defines a control point element as passed from the output
249 /// of the hull shader to the input of the domain shader
250 /////////////////////////////////////////////////////////////////////////
261 ScalarAttrib attrib
[SWR_VTX_NUM_SLOTS
];
264 //////////////////////////////////////////////////////////////////////////
265 /// SWR_TESSELLATION_FACTORS
266 /// @brief Tessellation factors structure (non-vector)
267 /////////////////////////////////////////////////////////////////////////
268 struct SWR_TESSELLATION_FACTORS
270 float OuterTessFactors
[SWR_NUM_OUTER_TESS_FACTORS
];
271 float InnerTessFactors
[SWR_NUM_INNER_TESS_FACTORS
];
274 #define MAX_NUM_VERTS_PER_PRIM 32 // support up to 32 control point patches
277 SWR_TESSELLATION_FACTORS tessFactors
;
278 ScalarCPoint cp
[MAX_NUM_VERTS_PER_PRIM
];
279 ScalarCPoint patchData
;
282 //////////////////////////////////////////////////////////////////////////
284 /// @brief Input to hull shader
285 /////////////////////////////////////////////////////////////////////////
286 struct SWR_HS_CONTEXT
288 simdvertex vert
[MAX_NUM_VERTS_PER_PRIM
]; // IN: (SIMD) input primitive data
289 simdscalari PrimitiveID
; // IN: (SIMD) primitive ID generated from the draw call
290 simdscalari mask
; // IN: Active mask for shader
291 ScalarPatch
* pCPout
; // OUT: Output control point patch
292 // SIMD-sized-array of SCALAR patches
293 SWR_SHADER_STATS stats
; // OUT: shader statistics used for archrast.
296 //////////////////////////////////////////////////////////////////////////
298 /// @brief Input to domain shader
299 /////////////////////////////////////////////////////////////////////////
300 struct SWR_DS_CONTEXT
302 uint32_t PrimitiveID
; // IN: (SCALAR) PrimitiveID for the patch associated with the DS invocation
303 uint32_t vectorOffset
; // IN: (SCALAR) vector index offset into SIMD data.
304 uint32_t vectorStride
; // IN: (SCALAR) stride (in vectors) of output data per attribute-component
305 uint32_t outVertexAttribOffset
; // IN: (SCALAR) Offset to the attributes as processed by the next shader stage.
306 ScalarPatch
* pCpIn
; // IN: (SCALAR) Control patch
307 simdscalar
* pDomainU
; // IN: (SIMD) Domain Point U coords
308 simdscalar
* pDomainV
; // IN: (SIMD) Domain Point V coords
309 simdscalari mask
; // IN: Active mask for shader
310 simdscalar
* pOutputData
; // OUT: (SIMD) Vertex Attributes (2D array of vectors, one row per attribute-component)
311 SWR_SHADER_STATS stats
; // OUT: shader statistics used for archrast.
314 //////////////////////////////////////////////////////////////////////////
316 /// @brief Input to geometry shader.
317 /////////////////////////////////////////////////////////////////////////
318 struct SWR_GS_CONTEXT
320 simdvector
* pVerts
; // IN: input primitive data for SIMD prims
321 uint32_t inputVertStride
; // IN: input vertex stride, in attributes
322 simdscalari PrimitiveID
; // IN: input primitive ID generated from the draw call
323 uint32_t InstanceID
; // IN: input instance ID
324 simdscalari mask
; // IN: Active mask for shader
325 uint8_t* pStreams
[KNOB_SIMD_WIDTH
]; // OUT: output stream (contains vertices for all output streams)
326 SWR_SHADER_STATS stats
; // OUT: shader statistics used for archrast.
329 struct PixelPositions
337 #define SWR_MAX_NUM_MULTISAMPLES 16
339 //////////////////////////////////////////////////////////////////////////
341 /// @brief Input to pixel shader.
342 /////////////////////////////////////////////////////////////////////////
343 struct SWR_PS_CONTEXT
345 PixelPositions vX
; // IN: x location(s) of pixels
346 PixelPositions vY
; // IN: x location(s) of pixels
347 simdscalar vZ
; // INOUT: z location of pixels
348 simdscalari activeMask
; // OUT: mask for kill
349 simdscalar inputMask
; // IN: input coverage mask for all samples
350 simdscalari oMask
; // OUT: mask for output coverage
352 PixelPositions vI
; // barycentric coords evaluated at pixel center, sample position, centroid
354 PixelPositions vOneOverW
; // IN: 1/w
356 const float* pAttribs
; // IN: pointer to attribute barycentric coefficients
357 const float* pPerspAttribs
; // IN: pointer to attribute/w barycentric coefficients
358 const float* pRecipW
; // IN: pointer to 1/w coord for each vertex
359 const float *I
; // IN: Barycentric A, B, and C coefs used to compute I
360 const float *J
; // IN: Barycentric A, B, and C coefs used to compute J
361 float recipDet
; // IN: 1/Det, used when barycentric interpolating attributes
362 const float* pSamplePosX
; // IN: array of sample positions
363 const float* pSamplePosY
; // IN: array of sample positions
364 simdvector shaded
[SWR_NUM_RENDERTARGETS
];
365 // OUT: result color per rendertarget
367 uint32_t frontFace
; // IN: front- 1, back- 0
368 uint32_t sampleIndex
; // IN: sampleIndex
369 uint32_t renderTargetArrayIndex
; // IN: render target array index from GS
370 uint32_t rasterizerSampleCount
; // IN: sample count used by the rasterizer
372 uint8_t* pColorBuffer
[SWR_NUM_RENDERTARGETS
]; // IN: Pointers to render target hottiles
374 SWR_SHADER_STATS stats
; // OUT: shader statistics used for archrast.
377 //////////////////////////////////////////////////////////////////////////
379 /// @brief Input to compute shader.
380 /////////////////////////////////////////////////////////////////////////
381 struct SWR_CS_CONTEXT
383 // The ThreadGroupId is the current thread group index relative
384 // to all thread groups in the Dispatch call. The ThreadId, ThreadIdInGroup,
385 // and ThreadIdInGroupFlattened can be derived from ThreadGroupId in the shader.
387 // Compute shader accepts the following system values.
388 // o ThreadId - Current thread id relative to all other threads in dispatch.
389 // o ThreadGroupId - Current thread group id relative to all other groups in dispatch.
390 // o ThreadIdInGroup - Current thread relative to all threads in the current thread group.
391 // o ThreadIdInGroupFlattened - Flattened linear id derived from ThreadIdInGroup.
393 // All of these system values can be computed in the shader. They will be
394 // derived from the current tile counter. The tile counter is an atomic counter that
395 // resides in the draw context and is initialized to the product of the dispatch dims.
397 // tileCounter = dispatchDims.x * dispatchDims.y * dispatchDims.z
399 // Each CPU worker thread will atomically decrement this counter and passes the current
400 // count into the shader. When the count reaches 0 then all thread groups in the
401 // dispatch call have been completed.
403 uint32_t tileCounter
; // The tile counter value for this thread group.
405 // Dispatch dimensions used by shader to compute system values from the tile counter.
406 uint32_t dispatchDims
[3];
408 uint8_t* pTGSM
; // Thread Group Shared Memory pointer.
409 uint8_t* pSpillFillBuffer
; // Spill/fill buffer for barrier support
410 uint8_t* pScratchSpace
; // Pointer to scratch space buffer used by the shader, shader is responsible
411 // for subdividing scratch space per instance/simd
412 uint32_t scratchSpacePerSimd
; // Scratch space per work item x SIMD_WIDTH
414 SWR_SHADER_STATS stats
; // OUT: shader statistics used for archrast.
420 SWR_TILE_NONE
= 0x0, // Linear mode (no tiling)
421 SWR_TILE_MODE_WMAJOR
, // W major tiling
422 SWR_TILE_MODE_XMAJOR
, // X major tiling
423 SWR_TILE_MODE_YMAJOR
, // Y major tiling
424 SWR_TILE_SWRZ
, // SWR-Z tiling
429 enum SWR_SURFACE_TYPE
436 SURFACE_STRUCTURED_BUFFER
= 5,
465 enum SWR_BLEND_FACTOR
468 BLENDFACTOR_SRC_COLOR
,
469 BLENDFACTOR_SRC_ALPHA
,
470 BLENDFACTOR_DST_ALPHA
,
471 BLENDFACTOR_DST_COLOR
,
472 BLENDFACTOR_SRC_ALPHA_SATURATE
,
473 BLENDFACTOR_CONST_COLOR
,
474 BLENDFACTOR_CONST_ALPHA
,
475 BLENDFACTOR_SRC1_COLOR
,
476 BLENDFACTOR_SRC1_ALPHA
,
478 BLENDFACTOR_INV_SRC_COLOR
,
479 BLENDFACTOR_INV_SRC_ALPHA
,
480 BLENDFACTOR_INV_DST_ALPHA
,
481 BLENDFACTOR_INV_DST_COLOR
,
482 BLENDFACTOR_INV_CONST_COLOR
,
483 BLENDFACTOR_INV_CONST_ALPHA
,
484 BLENDFACTOR_INV_SRC1_COLOR
,
485 BLENDFACTOR_INV_SRC1_ALPHA
501 LOGICOP_AND_INVERTED
,
502 LOGICOP_COPY_INVERTED
,
517 //////////////////////////////////////////////////////////////////////////
519 /// @brief Specifies how the auxiliary buffer is used by the driver.
520 //////////////////////////////////////////////////////////////////////////
529 struct SWR_LOD_OFFSETS
531 uint32_t offsets
[2][15];
534 //////////////////////////////////////////////////////////////////////////
535 /// SWR_SURFACE_STATE
536 //////////////////////////////////////////////////////////////////////////
537 struct SWR_SURFACE_STATE
539 gfxptr_t xpBaseAddress
;
540 SWR_SURFACE_TYPE type
; // @llvm_enum
541 SWR_FORMAT format
; // @llvm_enum
546 uint32_t samplePattern
;
549 uint32_t minLod
; // for sampled surfaces, the most detailed LOD that can be accessed by sampler
550 uint32_t maxLod
; // for sampled surfaces, the max LOD that can be accessed
551 float resourceMinLod
; // for sampled surfaces, the most detailed fractional mip that can be accessed by sampler
552 uint32_t lod
; // for render targets, the lod being rendered to
553 uint32_t arrayIndex
; // for render targets, the array index being rendered to for arrayed surfaces
554 SWR_TILE_MODE tileMode
; // @llvm_enum
560 uint32_t lodOffsets
[2][15]; // lod offsets for sampled surfaces
562 gfxptr_t xpAuxBaseAddress
; // Used for compression, append/consume counter, etc.
563 SWR_AUX_MODE auxMode
; // @llvm_enum
566 bool bInterleavedSamples
; // are MSAA samples stored interleaved or planar
569 // vertex fetch state
570 // WARNING- any changes to this struct need to be reflected
571 // in the fetch shader jit
572 struct SWR_VERTEX_BUFFER_STATE
578 uint32_t minVertex
; // min vertex (for bounds checking)
579 uint32_t maxVertex
; // size / pitch. precalculated value used by fetch shader for OOB checks
580 uint32_t partialInboundsSize
; // size % pitch. precalculated value used by fetch shader for partially OOB vertices
583 struct SWR_INDEX_BUFFER_STATE
586 // Format type for indices (e.g. UINT16, UINT32, etc.)
587 SWR_FORMAT format
; // @llvm_enum
592 //////////////////////////////////////////////////////////////////////////
593 /// SWR_FETCH_CONTEXT
594 /// @brief Input to fetch shader.
595 /// @note WARNING - Changes to this struct need to be reflected in the
596 /// fetch shader jit.
597 /////////////////////////////////////////////////////////////////////////
598 struct SWR_FETCH_CONTEXT
600 const SWR_VERTEX_BUFFER_STATE
* pStreams
; // IN: array of bound vertex buffers
601 gfxptr_t xpIndices
; // IN: pointer to int32 index buffer for indexed draws
602 gfxptr_t xpLastIndex
; // IN: pointer to end of index buffer, used for bounds checking
603 uint32_t CurInstance
; // IN: current instance
604 uint32_t BaseVertex
; // IN: base vertex
605 uint32_t StartVertex
; // IN: start vertex
606 uint32_t StartInstance
; // IN: start instance
607 simdscalari VertexID
; // OUT: vector of vertex IDs
608 simdscalari CutMask
; // OUT: vector mask of indices which have the cut index value
609 #if USE_SIMD16_SHADERS
610 // simd16scalari VertexID; // OUT: vector of vertex IDs
611 // simd16scalari CutMask; // OUT: vector mask of indices which have the cut index value
612 simdscalari VertexID2
; // OUT: vector of vertex IDs
613 simdscalari CutMask2
; // OUT: vector mask of indices which have the cut index value
617 //////////////////////////////////////////////////////////////////////////
620 /// @brief All statistics generated by SWR go here. These are public
622 /////////////////////////////////////////////////////////////////////////
623 OSALIGNLINE(struct) SWR_STATS
626 uint64_t DepthPassCount
; // Number of passing depth tests. Not exact.
629 uint64_t PsInvocations
; // Number of Pixel Shader invocations
630 uint64_t CsInvocations
; // Number of Compute Shader invocations
634 //////////////////////////////////////////////////////////////////////////
637 /// @brief All statistics generated by FE.
638 /////////////////////////////////////////////////////////////////////////
639 OSALIGNLINE(struct) SWR_STATS_FE
641 uint64_t IaVertices
; // Number of Fetch Shader vertices
642 uint64_t IaPrimitives
; // Number of PA primitives.
643 uint64_t VsInvocations
; // Number of Vertex Shader invocations
644 uint64_t HsInvocations
; // Number of Hull Shader invocations
645 uint64_t DsInvocations
; // Number of Domain Shader invocations
646 uint64_t GsInvocations
; // Number of Geometry Shader invocations
647 uint64_t GsPrimitives
; // Number of prims GS outputs.
648 uint64_t CInvocations
; // Number of clipper invocations
649 uint64_t CPrimitives
; // Number of clipper primitives.
652 uint64_t SoPrimStorageNeeded
[4];
653 uint64_t SoNumPrimsWritten
[4];
656 //////////////////////////////////////////////////////////////////////////
657 /// STREAMOUT_BUFFERS
658 /////////////////////////////////////////////////////////////////////////
660 #define MAX_SO_STREAMS 4
661 #define MAX_SO_BUFFERS 4
662 #define MAX_ATTRIBUTES 32
664 struct SWR_STREAMOUT_BUFFER
666 // Pointers to streamout buffers.
669 // Offset to the SO write offset. If not null then we update offset here.
670 uint32_t* pWriteOffset
;
675 // Size of buffer in dwords.
678 // Vertex pitch of buffer in dwords.
681 // Offset into buffer in dwords. SOS will increment this offset.
682 uint32_t streamOffset
;
685 //////////////////////////////////////////////////////////////////////////
687 /////////////////////////////////////////////////////////////////////////
688 struct SWR_STREAMOUT_STATE
690 // This disables stream output.
693 // which streams are enabled for streamout
694 bool streamEnable
[MAX_SO_STREAMS
];
696 // If set then do not send any streams to the rasterizer.
697 bool rasterizerDisable
;
699 // Specifies which stream to send to the rasterizer.
700 uint32_t streamToRasterizer
;
702 // The stream masks specify which attributes are sent to which streams.
703 // These masks help the FE to setup the pPrimData buffer that is passed
704 // the Stream Output Shader (SOS) function.
705 uint32_t streamMasks
[MAX_SO_STREAMS
];
707 // Number of attributes, including position, per vertex that are streamed out.
708 // This should match number of bits in stream mask.
709 uint32_t streamNumEntries
[MAX_SO_STREAMS
];
711 // Offset to the start of the attributes of the input vertices, in simdvector units
712 uint32_t vertexAttribOffset
[MAX_SO_STREAMS
];
715 //////////////////////////////////////////////////////////////////////////
716 /// STREAMOUT_CONTEXT - Passed to SOS
717 /////////////////////////////////////////////////////////////////////////
718 struct SWR_STREAMOUT_CONTEXT
721 SWR_STREAMOUT_BUFFER
* pBuffer
[MAX_SO_STREAMS
];
723 // Num prims written for this stream
724 uint32_t numPrimsWritten
;
726 // Num prims that should have been written if there were no overflow.
727 uint32_t numPrimStorageNeeded
;
730 //////////////////////////////////////////////////////////////////////////
731 /// SWR_GS_STATE - Geometry shader state
732 /////////////////////////////////////////////////////////////////////////
737 // If true, geometry shader emits a single stream, with separate cut buffer.
738 // If false, geometry shader emits vertices for multiple streams to the stream buffer, with a separate StreamID buffer
739 // to map vertices to streams
742 // Number of input attributes per vertex. Used by the frontend to
743 // optimize assembling primitives for GS
744 uint32_t numInputAttribs
;
746 // Stride of incoming verts in attributes
747 uint32_t inputVertStride
;
749 // Output topology - can be point, tristrip, or linestrip
750 PRIMITIVE_TOPOLOGY outputTopology
; // @llvm_enum
752 // Maximum number of verts that can be emitted by a single instance of the GS
753 uint32_t maxNumVerts
;
756 uint32_t instanceCount
;
758 // When single stream is enabled, singleStreamID dictates which stream is being output.
759 // field ignored if isSingleStream is false
760 uint32_t singleStreamID
;
762 // Total amount of memory to allocate for one instance of the shader output in bytes
763 uint32_t allocationSize
;
765 // Offset to the start of the attributes of the input vertices, in simdvector units, as read by the GS
766 uint32_t vertexAttribOffset
;
768 // Offset to the attributes as stored by the preceding shader stage.
769 uint32_t srcVertexAttribOffset
;
771 // Size of the control data section which contains cut or streamID data, in simdscalar units. Should be sized to handle
772 // the maximum number of verts output by the GS. Can be 0 if there are no cuts or streamID bits.
773 uint32_t controlDataSize
;
775 // Offset to the control data section, in bytes
776 uint32_t controlDataOffset
;
778 // Total size of an output vertex, in simdvector units
779 uint32_t outputVertexSize
;
781 // Offset to the start of the vertex section, in bytes
782 uint32_t outputVertexOffset
;
784 // Set this to non-zero to indicate that the shader outputs a static number of verts. If zero, shader is
785 // expected to store the final vertex count in the first dword of the gs output stream.
786 uint32_t staticVertexCount
;
790 static_assert(sizeof(SWR_GS_STATE
) == 64,
791 "Adjust padding to keep size (or remove this assert)");
794 //////////////////////////////////////////////////////////////////////////
795 /// SWR_TS_OUTPUT_TOPOLOGY - Defines data output by the tessellator / DS
796 /////////////////////////////////////////////////////////////////////////
797 enum SWR_TS_OUTPUT_TOPOLOGY
801 SWR_TS_OUTPUT_TRI_CW
,
802 SWR_TS_OUTPUT_TRI_CCW
,
804 SWR_TS_OUTPUT_TOPOLOGY_COUNT
807 //////////////////////////////////////////////////////////////////////////
808 /// SWR_TS_PARTITIONING - Defines tessellation algorithm
809 /////////////////////////////////////////////////////////////////////////
810 enum SWR_TS_PARTITIONING
813 SWR_TS_ODD_FRACTIONAL
,
814 SWR_TS_EVEN_FRACTIONAL
,
816 SWR_TS_PARTITIONING_COUNT
819 //////////////////////////////////////////////////////////////////////////
820 /// SWR_TS_DOMAIN - Defines Tessellation Domain
821 /////////////////////////////////////////////////////////////////////////
831 //////////////////////////////////////////////////////////////////////////
832 /// SWR_TS_STATE - Tessellation state
833 /////////////////////////////////////////////////////////////////////////
838 SWR_TS_OUTPUT_TOPOLOGY tsOutputTopology
; // @llvm_enum
839 SWR_TS_PARTITIONING partitioning
; // @llvm_enum
840 SWR_TS_DOMAIN domain
; // @llvm_enum
842 PRIMITIVE_TOPOLOGY postDSTopology
; // @llvm_enum
844 uint32_t numHsInputAttribs
;
845 uint32_t numHsOutputAttribs
;
846 uint32_t numDsOutputAttribs
;
847 uint32_t dsAllocationSize
;
848 uint32_t dsOutVtxAttribOffset
;
850 // Offset to the start of the attributes of the input vertices, in simdvector units
851 uint32_t vertexAttribOffset
;
854 // output merger state
855 struct SWR_RENDER_TARGET_BLEND_STATE
857 uint8_t writeDisableRed
: 1;
858 uint8_t writeDisableGreen
: 1;
859 uint8_t writeDisableBlue
: 1;
860 uint8_t writeDisableAlpha
: 1;
862 static_assert(sizeof(SWR_RENDER_TARGET_BLEND_STATE
) == 1, "Invalid SWR_RENDER_TARGET_BLEND_STATE size");
864 enum SWR_MULTISAMPLE_COUNT
866 SWR_MULTISAMPLE_1X
= 0,
871 SWR_MULTISAMPLE_TYPE_COUNT
874 static INLINE
uint32_t GetNumSamples(/* SWR_SAMPLE_COUNT */ int sampleCountEnum
) // @llvm_func_start
876 return uint32_t(1) << sampleCountEnum
;
879 struct SWR_BLEND_STATE
881 // constant blend factor color in RGBA float
882 float constantColor
[4];
884 // alpha test reference value in unorm8 or float32
885 uint32_t alphaTestReference
;
887 // all RT's have the same sample count
888 ///@todo move this to Output Merger state when we refactor
889 SWR_MULTISAMPLE_COUNT sampleCount
; // @llvm_enum
891 SWR_RENDER_TARGET_BLEND_STATE renderTarget
[SWR_NUM_RENDERTARGETS
];
893 static_assert(sizeof(SWR_BLEND_STATE
) == 36, "Invalid SWR_BLEND_STATE size");
895 struct SWR_BLEND_CONTEXT
897 const SWR_BLEND_STATE
* pBlendState
;
900 simdvector
* src0alpha
;
906 uint32_t isAlphaTested
;
907 uint32_t isAlphaBlended
;
910 //////////////////////////////////////////////////////////////////////////
911 /// FUNCTION POINTERS FOR SHADERS
913 #if USE_SIMD16_SHADERS
914 typedef void(__cdecl
*PFN_FETCH_FUNC
)(HANDLE hPrivateData
, SWR_FETCH_CONTEXT
& fetchInfo
, simd16vertex
& out
);
916 typedef void(__cdecl
*PFN_FETCH_FUNC
)(HANDLE hPrivateData
, SWR_FETCH_CONTEXT
& fetchInfo
, simdvertex
& out
);
918 typedef void(__cdecl
*PFN_VERTEX_FUNC
)(HANDLE hPrivateData
, SWR_VS_CONTEXT
* pVsContext
);
919 typedef void(__cdecl
*PFN_HS_FUNC
)(HANDLE hPrivateData
, SWR_HS_CONTEXT
* pHsContext
);
920 typedef void(__cdecl
*PFN_DS_FUNC
)(HANDLE hPrivateData
, SWR_DS_CONTEXT
* pDsContext
);
921 typedef void(__cdecl
*PFN_GS_FUNC
)(HANDLE hPrivateData
, SWR_GS_CONTEXT
* pGsContext
);
922 typedef void(__cdecl
*PFN_CS_FUNC
)(HANDLE hPrivateData
, SWR_CS_CONTEXT
* pCsContext
);
923 typedef void(__cdecl
*PFN_SO_FUNC
)(SWR_STREAMOUT_CONTEXT
& soContext
);
924 typedef void(__cdecl
*PFN_PIXEL_KERNEL
)(HANDLE hPrivateData
, SWR_PS_CONTEXT
*pContext
);
925 typedef void(__cdecl
*PFN_CPIXEL_KERNEL
)(HANDLE hPrivateData
, SWR_PS_CONTEXT
*pContext
);
926 typedef void(__cdecl
*PFN_BLEND_JIT_FUNC
)(SWR_BLEND_CONTEXT
*);
927 typedef simdscalar(*PFN_QUANTIZE_DEPTH
)(simdscalar
const &);
931 //////////////////////////////////////////////////////////////////////////
933 /////////////////////////////////////////////////////////////////////////
934 struct SWR_FRONTEND_STATE
936 // skip clip test, perspective divide, and viewport transform
937 // intended for verts in screen space
938 bool vpTransformDisable
;
939 bool bEnableCutIndex
;
945 uint32_t lineStripList
: 1;
946 uint32_t triStripList
: 2;
950 uint32_t topologyProvokingVertex
; // provoking vertex for the draw topology
952 // Size of a vertex in simdvector units. Should be sized to the
953 // maximum of the input/output of the vertex shader.
954 uint32_t vsVertexSize
;
957 //////////////////////////////////////////////////////////////////////////
959 /////////////////////////////////////////////////////////////////////////
960 struct SWR_VIEWPORT_MATRIX
970 //////////////////////////////////////////////////////////////////////////
971 /// VIEWPORT_MATRIXES
972 /////////////////////////////////////////////////////////////////////////
973 struct SWR_VIEWPORT_MATRICES
975 float m00
[KNOB_NUM_VIEWPORTS_SCISSORS
];
976 float m11
[KNOB_NUM_VIEWPORTS_SCISSORS
];
977 float m22
[KNOB_NUM_VIEWPORTS_SCISSORS
];
978 float m30
[KNOB_NUM_VIEWPORTS_SCISSORS
];
979 float m31
[KNOB_NUM_VIEWPORTS_SCISSORS
];
980 float m32
[KNOB_NUM_VIEWPORTS_SCISSORS
];
983 //////////////////////////////////////////////////////////////////////////
985 /////////////////////////////////////////////////////////////////////////
996 //////////////////////////////////////////////////////////////////////////
998 //////////////////////////////////////////////////////////////////////////
1010 SWR_FILLMODE_WIREFRAME
,
1014 enum SWR_FRONTWINDING
1016 SWR_FRONTWINDING_CW
,
1017 SWR_FRONTWINDING_CCW
1021 enum SWR_PIXEL_LOCATION
1023 SWR_PIXEL_LOCATION_CENTER
,
1024 SWR_PIXEL_LOCATION_UL
,
1027 // fixed point screen space sample locations within a pixel
1028 struct SWR_MULTISAMPLE_POS
1031 INLINE
void SetXi(uint32_t sampleNum
, uint32_t val
) { _xi
[sampleNum
] = val
; }; // @llvm_func
1032 INLINE
void SetYi(uint32_t sampleNum
, uint32_t val
) { _yi
[sampleNum
] = val
; }; // @llvm_func
1033 INLINE
uint32_t Xi(uint32_t sampleNum
) const { return _xi
[sampleNum
]; }; // @llvm_func
1034 INLINE
uint32_t Yi(uint32_t sampleNum
) const { return _yi
[sampleNum
]; }; // @llvm_func
1035 INLINE
void SetX(uint32_t sampleNum
, float val
) { _x
[sampleNum
] = val
; }; // @llvm_func
1036 INLINE
void SetY(uint32_t sampleNum
, float val
) { _y
[sampleNum
] = val
; }; // @llvm_func
1037 INLINE
float X(uint32_t sampleNum
) const { return _x
[sampleNum
]; }; // @llvm_func
1038 INLINE
float Y(uint32_t sampleNum
) const { return _y
[sampleNum
]; }; // @llvm_func
1039 typedef const float(&sampleArrayT
)[SWR_MAX_NUM_MULTISAMPLES
]; //@llvm_typedef
1040 INLINE sampleArrayT
X() const { return _x
; }; // @llvm_func
1041 INLINE sampleArrayT
Y() const { return _y
; }; // @llvm_func
1042 INLINE
const __m128i
& vXi(uint32_t sampleNum
) const { return _vXi
[sampleNum
]; }; // @llvm_func
1043 INLINE
const __m128i
& vYi(uint32_t sampleNum
) const { return _vYi
[sampleNum
]; }; // @llvm_func
1044 INLINE
const simdscalar
& vX(uint32_t sampleNum
) const { return _vX
[sampleNum
]; }; // @llvm_func
1045 INLINE
const simdscalar
& vY(uint32_t sampleNum
) const { return _vY
[sampleNum
]; }; // @llvm_func
1046 INLINE
const __m128i
& TileSampleOffsetsX() const { return tileSampleOffsetsX
; }; // @llvm_func
1047 INLINE
const __m128i
& TileSampleOffsetsY() const { return tileSampleOffsetsY
; }; // @llvm_func
1049 INLINE
void PrecalcSampleData(int numSamples
); //@llvm_func
1052 template <typename MaskT
>
1053 INLINE __m128i
expandThenBlend4(uint32_t* min
, uint32_t* max
); // @llvm_func
1054 INLINE
void CalcTileSampleOffsets(int numSamples
); // @llvm_func
1056 // scalar sample values
1057 uint32_t _xi
[SWR_MAX_NUM_MULTISAMPLES
];
1058 uint32_t _yi
[SWR_MAX_NUM_MULTISAMPLES
];
1059 float _x
[SWR_MAX_NUM_MULTISAMPLES
];
1060 float _y
[SWR_MAX_NUM_MULTISAMPLES
];
1062 // precalc'd / vectorized samples
1063 __m128i _vXi
[SWR_MAX_NUM_MULTISAMPLES
];
1064 __m128i _vYi
[SWR_MAX_NUM_MULTISAMPLES
];
1065 simdscalar _vX
[SWR_MAX_NUM_MULTISAMPLES
];
1066 simdscalar _vY
[SWR_MAX_NUM_MULTISAMPLES
];
1067 __m128i tileSampleOffsetsX
;
1068 __m128i tileSampleOffsetsY
;
1071 //////////////////////////////////////////////////////////////////////////
1073 //////////////////////////////////////////////////////////////////////////
1074 struct SWR_RASTSTATE
1076 uint32_t cullMode
: 2;
1077 uint32_t fillMode
: 2;
1078 uint32_t frontWinding
: 1;
1079 uint32_t scissorEnable
: 1;
1080 uint32_t depthClipEnable
: 1;
1081 uint32_t clipHalfZ
: 1;
1082 uint32_t pointParam
: 1;
1083 uint32_t pointSpriteEnable
: 1;
1084 uint32_t pointSpriteTopOrigin
: 1;
1085 uint32_t forcedSampleCount
: 1;
1086 uint32_t pixelOffset
: 1;
1087 uint32_t depthBiasPreAdjusted
: 1; ///< depth bias constant is in float units, not per-format Z units
1088 uint32_t conservativeRast
: 1;
1094 float slopeScaledDepthBias
;
1095 float depthBiasClamp
;
1096 SWR_FORMAT depthFormat
; // @llvm_enum
1098 // sample count the rasterizer is running at
1099 SWR_MULTISAMPLE_COUNT sampleCount
; // @llvm_enum
1100 uint32_t pixelLocation
; // UL or Center
1101 SWR_MULTISAMPLE_POS samplePositions
; // @llvm_struct
1102 bool bIsCenterPattern
; // @llvm_enum
1106 enum SWR_CONSTANT_SOURCE
1108 SWR_CONSTANT_SOURCE_CONST_0000
,
1109 SWR_CONSTANT_SOURCE_CONST_0001_FLOAT
,
1110 SWR_CONSTANT_SOURCE_CONST_1111_FLOAT
,
1111 SWR_CONSTANT_SOURCE_PRIM_ID
1114 struct SWR_ATTRIB_SWIZZLE
1116 uint16_t sourceAttrib
: 5; // source attribute
1117 uint16_t constantSource
: 2; // constant source to apply
1118 uint16_t componentOverrideMask
: 4; // override component with constant source
1122 struct SWR_BACKEND_STATE
1124 uint32_t constantInterpolationMask
; // bitmask indicating which attributes have constant interpolation
1125 uint32_t pointSpriteTexCoordMask
; // bitmask indicating the attribute(s) which should be interpreted as tex coordinates
1127 bool swizzleEnable
; // when enabled, core will parse the swizzle map when
1128 // setting up attributes for the backend, otherwise
1129 // all attributes up to numAttributes will be sent
1130 uint8_t numAttributes
; // total number of attributes to send to backend (up to 32)
1131 uint8_t numComponents
[32]; // number of components to setup per attribute, this reduces some calculations for unneeded components
1133 bool readRenderTargetArrayIndex
; // Forward render target array index from last FE stage to the backend
1134 bool readViewportArrayIndex
; // Read viewport array index from last FE stage during binning
1136 // User clip/cull distance enables
1137 uint8_t cullDistanceMask
;
1138 uint8_t clipDistanceMask
;
1140 // padding to ensure swizzleMap starts 64B offset from start of the struct
1141 // and that the next fields are dword aligned.
1144 // Offset to the start of the attributes of the input vertices, in simdvector units
1145 uint32_t vertexAttribOffset
;
1147 // Offset to clip/cull attrib section of the vertex, in simdvector units
1148 uint32_t vertexClipCullOffset
;
1150 SWR_ATTRIB_SWIZZLE swizzleMap
[32];
1152 static_assert(sizeof(SWR_BACKEND_STATE
) == 128,
1153 "Adjust padding to keep size (or remove this assert)");
1156 union SWR_DEPTH_STENCIL_STATE
1161 uint32_t depthWriteEnable
: 1;
1162 uint32_t depthTestEnable
: 1;
1163 uint32_t stencilWriteEnable
: 1;
1164 uint32_t stencilTestEnable
: 1;
1165 uint32_t doubleSidedStencilTestEnable
: 1;
1167 uint32_t depthTestFunc
: 3;
1168 uint32_t stencilTestFunc
: 3;
1170 uint32_t backfaceStencilPassDepthPassOp
: 3;
1171 uint32_t backfaceStencilPassDepthFailOp
: 3;
1172 uint32_t backfaceStencilFailOp
: 3;
1173 uint32_t backfaceStencilTestFunc
: 3;
1174 uint32_t stencilPassDepthPassOp
: 3;
1175 uint32_t stencilPassDepthFailOp
: 3;
1176 uint32_t stencilFailOp
: 3;
1179 uint8_t backfaceStencilWriteMask
;
1180 uint8_t backfaceStencilTestMask
;
1181 uint8_t stencilWriteMask
;
1182 uint8_t stencilTestMask
;
1185 uint8_t backfaceStencilRefValue
;
1186 uint8_t stencilRefValue
;
1191 enum SWR_SHADING_RATE
1193 SWR_SHADING_RATE_PIXEL
,
1194 SWR_SHADING_RATE_SAMPLE
,
1195 SWR_SHADING_RATE_COUNT
,
1198 enum SWR_INPUT_COVERAGE
1200 SWR_INPUT_COVERAGE_NONE
,
1201 SWR_INPUT_COVERAGE_NORMAL
,
1202 SWR_INPUT_COVERAGE_INNER_CONSERVATIVE
,
1203 SWR_INPUT_COVERAGE_COUNT
,
1206 enum SWR_PS_POSITION_OFFSET
1208 SWR_PS_POSITION_SAMPLE_NONE
,
1209 SWR_PS_POSITION_SAMPLE_OFFSET
,
1210 SWR_PS_POSITION_CENTROID_OFFSET
,
1211 SWR_PS_POSITION_OFFSET_COUNT
,
1214 enum SWR_BARYCENTRICS_MASK
1216 SWR_BARYCENTRIC_PER_PIXEL_MASK
= 0x1,
1217 SWR_BARYCENTRIC_CENTROID_MASK
= 0x2,
1218 SWR_BARYCENTRIC_PER_SAMPLE_MASK
= 0x4,
1221 // pixel shader state
1225 PFN_PIXEL_KERNEL pfnPixelShader
; // @llvm_pfn
1228 uint32_t killsPixel
: 1; // pixel shader can kill pixels
1229 uint32_t inputCoverage
: 2; // ps uses input coverage
1230 uint32_t writesODepth
: 1; // pixel shader writes to depth
1231 uint32_t usesSourceDepth
: 1; // pixel shader reads depth
1232 uint32_t shadingRate
: 2; // shading per pixel / sample / coarse pixel
1233 uint32_t posOffset
: 2; // type of offset (none, sample, centroid) to add to pixel position
1234 uint32_t barycentricsMask
: 3; // which type(s) of barycentric coords does the PS interpolate attributes with
1235 uint32_t usesUAV
: 1; // pixel shader accesses UAV
1236 uint32_t forceEarlyZ
: 1; // force execution of early depth/stencil test
1238 uint8_t renderTargetMask
; // Mask of render targets written
1241 // depth bounds state
1242 struct SWR_DEPTH_BOUNDS_STATE
1244 bool depthBoundsTestEnable
;
1245 float depthBoundsTestMinValue
;
1246 float depthBoundsTestMaxValue
;