• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1.. _screen:
2
3Screen
4======
5
6A screen is an object representing the context-independent part of a device.
7
8Flags and enumerations
9----------------------
10
11XXX some of these don't belong in this section.
12
13
14.. _pipe_cap:
15
16PIPE_CAP_*
17^^^^^^^^^^
18
19Capability queries return information about the features and limits of the
20driver/GPU.  For floating-point values, use :ref:`get_paramf`, and for boolean
21or integer values, use :ref:`get_param`.
22
23The integer capabilities:
24
25* ``PIPE_CAP_NPOT_TEXTURES``: Whether :term:`NPOT` textures may have repeat modes,
26  normalized coordinates, and mipmaps.
27* ``PIPE_CAP_TWO_SIDED_STENCIL``: Whether the stencil test can also affect back-facing
28  polygons.
29* ``PIPE_CAP_MAX_DUAL_SOURCE_RENDER_TARGETS``: How many dual-source blend RTs are support.
30  :ref:`Blend` for more information.
31* ``PIPE_CAP_ANISOTROPIC_FILTER``: Whether textures can be filtered anisotropically.
32* ``PIPE_CAP_POINT_SPRITE``: Whether point sprites are available.
33* ``PIPE_CAP_MAX_RENDER_TARGETS``: The maximum number of render targets that may be
34  bound.
35* ``PIPE_CAP_OCCLUSION_QUERY``: Whether occlusion queries are available.
36* ``PIPE_CAP_QUERY_TIME_ELAPSED``: Whether PIPE_QUERY_TIME_ELAPSED queries are available.
37* ``PIPE_CAP_TEXTURE_SHADOW_MAP``: indicates whether the fragment shader hardware
38  can do the depth texture / Z comparison operation in TEX instructions
39  for shadow testing.
40* ``PIPE_CAP_TEXTURE_SWIZZLE``: Whether swizzling through sampler views is
41  supported.
42* ``PIPE_CAP_MAX_TEXTURE_2D_LEVELS``: The maximum number of mipmap levels available
43  for a 2D texture.
44* ``PIPE_CAP_MAX_TEXTURE_3D_LEVELS``: The maximum number of mipmap levels available
45  for a 3D texture.
46* ``PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS``: The maximum number of mipmap levels available
47  for a cubemap.
48* ``PIPE_CAP_TEXTURE_MIRROR_CLAMP``: Whether mirrored texture coordinates with clamp
49  are supported.
50* ``PIPE_CAP_BLEND_EQUATION_SEPARATE``: Whether alpha blend equations may be different
51  from color blend equations, in :ref:`Blend` state.
52* ``PIPE_CAP_SM3``: Whether the vertex shader and fragment shader support equivalent
53  opcodes to the Shader Model 3 specification. XXX oh god this is horrible
54* ``PIPE_CAP_MAX_STREAM_OUTPUT_BUFFERS``: The maximum number of stream buffers.
55* ``PIPE_CAP_PRIMITIVE_RESTART``: Whether primitive restart is supported.
56* ``PIPE_CAP_INDEP_BLEND_ENABLE``: Whether per-rendertarget blend enabling and channel
57  masks are supported. If 0, then the first rendertarget's blend mask is
58  replicated across all MRTs.
59* ``PIPE_CAP_INDEP_BLEND_FUNC``: Whether per-rendertarget blend functions are
60  available. If 0, then the first rendertarget's blend functions affect all
61  MRTs.
62* ``PIPE_CAP_MAX_TEXTURE_ARRAY_LAYERS``: The maximum number of texture array
63  layers supported. If 0, the array textures are not supported at all and
64  the ARRAY texture targets are invalid.
65* ``PIPE_CAP_TGSI_FS_COORD_ORIGIN_UPPER_LEFT``: Whether the TGSI property
66  FS_COORD_ORIGIN with value UPPER_LEFT is supported.
67* ``PIPE_CAP_TGSI_FS_COORD_ORIGIN_LOWER_LEFT``: Whether the TGSI property
68  FS_COORD_ORIGIN with value LOWER_LEFT is supported.
69* ``PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_HALF_INTEGER``: Whether the TGSI
70  property FS_COORD_PIXEL_CENTER with value HALF_INTEGER is supported.
71* ``PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_INTEGER``: Whether the TGSI
72  property FS_COORD_PIXEL_CENTER with value INTEGER is supported.
73* ``PIPE_CAP_DEPTH_CLIP_DISABLE``: Whether the driver is capable of disabling
74  depth clipping (through pipe_rasterizer_state)
75* ``PIPE_CAP_SHADER_STENCIL_EXPORT``: Whether a stencil reference value can be
76  written from a fragment shader.
77* ``PIPE_CAP_TGSI_INSTANCEID``: Whether TGSI_SEMANTIC_INSTANCEID is supported
78  in the vertex shader.
79* ``PIPE_CAP_VERTEX_ELEMENT_INSTANCE_DIVISOR``: Whether the driver supports
80  per-instance vertex attribs.
81* ``PIPE_CAP_FRAGMENT_COLOR_CLAMPED``: Whether fragment color clamping is
82  supported.  That is, is the pipe_rasterizer_state::clamp_fragment_color
83  flag supported by the driver?  If not, the state tracker will insert
84  clamping code into the fragment shaders when needed.
85
86* ``PIPE_CAP_MIXED_COLORBUFFER_FORMATS``: Whether mixed colorbuffer formats are
87  supported, e.g. RGBA8 and RGBA32F as the first and second colorbuffer, resp.
88* ``PIPE_CAP_VERTEX_COLOR_UNCLAMPED``: Whether the driver is capable of
89  outputting unclamped vertex colors from a vertex shader. If unsupported,
90  the vertex colors are always clamped. This is the default for DX9 hardware.
91* ``PIPE_CAP_VERTEX_COLOR_CLAMPED``: Whether the driver is capable of
92  clamping vertex colors when they come out of a vertex shader, as specified
93  by the pipe_rasterizer_state::clamp_vertex_color flag.  If unsupported,
94  the vertex colors are never clamped. This is the default for DX10 hardware.
95  If both clamped and unclamped CAPs are supported, the clamping can be
96  controlled through pipe_rasterizer_state.  If the driver cannot do vertex
97  color clamping, the state tracker may insert clamping code into the vertex
98  shader.
99* ``PIPE_CAP_GLSL_FEATURE_LEVEL``: Whether the driver supports features
100  equivalent to a specific GLSL version. E.g. for GLSL 1.3, report 130.
101* ``PIPE_CAP_QUADS_FOLLOW_PROVOKING_VERTEX_CONVENTION``: Whether quads adhere to
102  the flatshade_first setting in ``pipe_rasterizer_state``.
103* ``PIPE_CAP_USER_VERTEX_BUFFERS``: Whether the driver supports user vertex
104  buffers.  If not, the state tracker must upload all data which is not in hw
105  resources.  If user-space buffers are supported, the driver must also still
106  accept HW resource buffers.
107* ``PIPE_CAP_VERTEX_BUFFER_OFFSET_4BYTE_ALIGNED_ONLY``: This CAP describes a hw
108  limitation.  If true, pipe_vertex_buffer::buffer_offset must always be aligned
109  to 4.  If false, there are no restrictions on the offset.
110* ``PIPE_CAP_VERTEX_BUFFER_STRIDE_4BYTE_ALIGNED_ONLY``: This CAP describes a hw
111  limitation.  If true, pipe_vertex_buffer::stride must always be aligned to 4.
112  If false, there are no restrictions on the stride.
113* ``PIPE_CAP_VERTEX_ELEMENT_SRC_OFFSET_4BYTE_ALIGNED_ONLY``: This CAP describes
114  a hw limitation.  If true, pipe_vertex_element::src_offset must always be
115  aligned to 4.  If false, there are no restrictions on src_offset.
116* ``PIPE_CAP_COMPUTE``: Whether the implementation supports the
117  compute entry points defined in pipe_context and pipe_screen.
118* ``PIPE_CAP_USER_INDEX_BUFFERS``: Whether user index buffers are supported.
119  If not, the state tracker must upload all indices which are not in hw
120  resources.  If user-space buffers are supported, the driver must also still
121  accept HW resource buffers.
122* ``PIPE_CAP_USER_CONSTANT_BUFFERS``: Whether user-space constant buffers
123  are supported.  If not, the state tracker must put constants into HW
124  resources/buffers.  If user-space constant buffers are supported, the
125  driver must still accept HW constant buffers also.
126* ``PIPE_CAP_CONSTANT_BUFFER_OFFSET_ALIGNMENT``: Describes the required
127  alignment of pipe_constant_buffer::buffer_offset.
128* ``PIPE_CAP_START_INSTANCE``: Whether the driver supports
129  pipe_draw_info::start_instance.
130* ``PIPE_CAP_QUERY_TIMESTAMP``: Whether PIPE_QUERY_TIMESTAMP and
131  the pipe_screen::get_timestamp hook are implemented.
132* ``PIPE_CAP_TEXTURE_MULTISAMPLE``: Whether all MSAA resources supported
133  for rendering are also supported for texturing.
134* ``PIPE_CAP_MIN_MAP_BUFFER_ALIGNMENT``: The minimum alignment that should be
135  expected for a pointer returned by transfer_map if the resource is
136  PIPE_BUFFER. In other words, the pointer returned by transfer_map is
137  always aligned to this value.
138* ``PIPE_CAP_TEXTURE_BUFFER_OFFSET_ALIGNMENT``: Describes the required
139  alignment for pipe_sampler_view::u.buf.offset, in bytes.
140  If a driver does not support offset/size, it should return 0.
141* ``PIPE_CAP_BUFFER_SAMPLER_VIEW_RGBA_ONLY``: Whether the driver only
142  supports R, RG, RGB and RGBA formats for PIPE_BUFFER sampler views.
143  When this is the case it should be assumed that the swizzle parameters
144  in the sampler view have no effect.
145* ``PIPE_CAP_TGSI_TEXCOORD``: This CAP describes a hw limitation.
146  If true, the hardware cannot replace arbitrary shader inputs with sprite
147  coordinates and hence the inputs that are desired to be replaceable must
148  be declared with TGSI_SEMANTIC_TEXCOORD instead of TGSI_SEMANTIC_GENERIC.
149  The rasterizer's sprite_coord_enable state therefore also applies to the
150  TEXCOORD semantic.
151  Also, TGSI_SEMANTIC_PCOORD becomes available, which labels a fragment shader
152  input that will always be replaced with sprite coordinates.
153* ``PIPE_CAP_PREFER_BLIT_BASED_TEXTURE_TRANSFER``: Whether it is preferable
154  to use a blit to implement a texture transfer which needs format conversions
155  and swizzling in state trackers. Generally, all hardware drivers with
156  dedicated memory should return 1 and all software rasterizers should return 0.
157* ``PIPE_CAP_QUERY_PIPELINE_STATISTICS``: Whether PIPE_QUERY_PIPELINE_STATISTICS
158  is supported.
159* ``PIPE_CAP_TEXTURE_BORDER_COLOR_QUIRK``: Bitmask indicating whether special
160  considerations have to be given to the interaction between the border color
161  in the sampler object and the sampler view used with it.
162  If PIPE_QUIRK_TEXTURE_BORDER_COLOR_SWIZZLE_R600 is set, the border color
163  may be affected in undefined ways for any kind of permutational swizzle
164  (any swizzle XYZW where X/Y/Z/W are not ZERO, ONE, or R/G/B/A respectively)
165  in the sampler view.
166  If PIPE_QUIRK_TEXTURE_BORDER_COLOR_SWIZZLE_NV50 is set, the border color
167  state should be swizzled manually according to the swizzle in the sampler
168  view it is intended to be used with, or herein undefined results may occur
169  for permutational swizzles.
170* ``PIPE_CAP_MAX_TEXTURE_BUFFER_SIZE``: The maximum accessible size with
171  a buffer sampler view, in texels.
172* ``PIPE_CAP_MAX_VIEWPORTS``: The maximum number of viewports (and scissors
173  since they are linked) a driver can support. Returning 0 is equivalent
174  to returning 1 because every driver has to support at least a single
175  viewport/scissor combination.
176* ``PIPE_CAP_ENDIANNESS``:: The endianness of the device.  Either
177  PIPE_ENDIAN_BIG or PIPE_ENDIAN_LITTLE.
178* ``PIPE_CAP_MIXED_FRAMEBUFFER_SIZES``: Whether it is allowed to have
179  different sizes for fb color/zs attachments. This controls whether
180  ARB_framebuffer_object is provided.
181* ``PIPE_CAP_TGSI_VS_LAYER_VIEWPORT``: Whether ``TGSI_SEMANTIC_LAYER`` and
182  ``TGSI_SEMANTIC_VIEWPORT_INDEX`` are supported as vertex shader
183  outputs. Note that the viewport will only be used if multiple viewports are
184  exposed.
185* ``PIPE_CAP_MAX_GEOMETRY_OUTPUT_VERTICES``: The maximum number of vertices
186  output by a single invocation of a geometry shader.
187* ``PIPE_CAP_MAX_GEOMETRY_TOTAL_OUTPUT_COMPONENTS``: The maximum number of
188  vertex components output by a single invocation of a geometry shader.
189  This is the product of the number of attribute components per vertex and
190  the number of output vertices.
191* ``PIPE_CAP_MAX_TEXTURE_GATHER_COMPONENTS``: Max number of components
192  in format that texture gather can operate on. 1 == RED, ALPHA etc,
193  4 == All formats.
194* ``PIPE_CAP_TEXTURE_GATHER_SM5``: Whether the texture gather
195  hardware implements the SM5 features, component selection,
196  shadow comparison, and run-time offsets.
197* ``PIPE_CAP_BUFFER_MAP_PERSISTENT_COHERENT``: Whether
198  PIPE_TRANSFER_PERSISTENT and PIPE_TRANSFER_COHERENT are supported
199  for buffers.
200* ``PIPE_CAP_TEXTURE_QUERY_LOD``: Whether the ``LODQ`` instruction is
201  supported.
202* ``PIPE_CAP_MIN_TEXTURE_GATHER_OFFSET``: The minimum offset that can be used
203  in conjunction with a texture gather opcode.
204* ``PIPE_CAP_MAX_TEXTURE_GATHER_OFFSET``: The maximum offset that can be used
205  in conjunction with a texture gather opcode.
206* ``PIPE_CAP_SAMPLE_SHADING``: Whether there is support for per-sample
207  shading. The context->set_min_samples function will be expected to be
208  implemented.
209* ``PIPE_CAP_TEXTURE_GATHER_OFFSETS``: Whether the ``TG4`` instruction can
210  accept 4 offsets.
211* ``PIPE_CAP_TGSI_VS_WINDOW_SPACE_POSITION``: Whether
212  TGSI_PROPERTY_VS_WINDOW_SPACE_POSITION is supported, which disables clipping
213  and viewport transformation.
214* ``PIPE_CAP_MAX_VERTEX_STREAMS``: The maximum number of vertex streams
215  supported by the geometry shader. If stream-out is supported, this should be
216  at least 1. If stream-out is not supported, this should be 0.
217* ``PIPE_CAP_DRAW_INDIRECT``: Whether the driver supports taking draw arguments
218  { count, instance_count, start, index_bias } from a PIPE_BUFFER resource.
219  See pipe_draw_info.
220* ``PIPE_CAP_MULTI_DRAW_INDIRECT``: Whether the driver supports
221  pipe_draw_info::indirect_stride and ::indirect_count
222* ``PIPE_CAP_MULTI_DRAW_INDIRECT_PARAMS``: Whether the driver supports
223  taking the number of indirect draws from a separate parameter
224  buffer, see pipe_draw_info::indirect_params.
225* ``PIPE_CAP_TGSI_FS_FINE_DERIVATIVE``: Whether the fragment shader supports
226  the FINE versions of DDX/DDY.
227* ``PIPE_CAP_VENDOR_ID``: The vendor ID of the underlying hardware. If it's
228  not available one should return 0xFFFFFFFF.
229* ``PIPE_CAP_DEVICE_ID``: The device ID (PCI ID) of the underlying hardware.
230  0xFFFFFFFF if not available.
231* ``PIPE_CAP_ACCELERATED``: Whether the renderer is hardware accelerated.
232* ``PIPE_CAP_VIDEO_MEMORY``: The amount of video memory in megabytes.
233* ``PIPE_CAP_UMA``: If the device has a unified memory architecture or on-card
234  memory and GART.
235* ``PIPE_CAP_CONDITIONAL_RENDER_INVERTED``: Whether the driver supports inverted
236  condition for conditional rendering.
237* ``PIPE_CAP_MAX_VERTEX_ATTRIB_STRIDE``: The maximum supported vertex stride.
238* ``PIPE_CAP_SAMPLER_VIEW_TARGET``: Whether the sampler view's target can be
239  different than the underlying resource's, as permitted by
240  ARB_texture_view. For example a 2d array texture may be reinterpreted as a
241  cube (array) texture and vice-versa.
242* ``PIPE_CAP_CLIP_HALFZ``: Whether the driver supports the
243  pipe_rasterizer_state::clip_halfz being set to true. This is required
244  for enabling ARB_clip_control.
245* ``PIPE_CAP_VERTEXID_NOBASE``: If true, the driver only supports
246  TGSI_SEMANTIC_VERTEXID_NOBASE (and not TGSI_SEMANTIC_VERTEXID). This means
247  state trackers for APIs whose vertexIDs are offset by basevertex (such as GL)
248  will need to lower TGSI_SEMANTIC_VERTEXID to TGSI_SEMANTIC_VERTEXID_NOBASE
249  and TGSI_SEMANTIC_BASEVERTEX, so drivers setting this must handle both these
250  semantics. Only relevant if geometry shaders are supported.
251  (BASEVERTEX could be exposed separately too via ``PIPE_CAP_DRAW_PARAMETERS``).
252* ``PIPE_CAP_POLYGON_OFFSET_CLAMP``: If true, the driver implements support
253  for ``pipe_rasterizer_state::offset_clamp``.
254* ``PIPE_CAP_MULTISAMPLE_Z_RESOLVE``: Whether the driver supports blitting
255  a multisampled depth buffer into a single-sampled texture (or depth buffer).
256  Only the first sampled should be copied.
257* ``PIPE_CAP_RESOURCE_FROM_USER_MEMORY``: Whether the driver can create
258  a pipe_resource where an already-existing piece of (malloc'd) user memory
259  is used as its backing storage. In other words, whether the driver can map
260  existing user memory into the device address space for direct device access.
261  The create function is pipe_screen::resource_from_user_memory. The address
262  and size must be page-aligned.
263* ``PIPE_CAP_DEVICE_RESET_STATUS_QUERY``:
264  Whether pipe_context::get_device_reset_status is implemented.
265* ``PIPE_CAP_MAX_SHADER_PATCH_VARYINGS``:
266  How many per-patch outputs and inputs are supported between tessellation
267  control and tessellation evaluation shaders, not counting in TESSINNER and
268  TESSOUTER. The minimum allowed value for OpenGL is 30.
269* ``PIPE_CAP_TEXTURE_FLOAT_LINEAR``: Whether the linear minification and
270  magnification filters are supported with single-precision floating-point
271  textures.
272* ``PIPE_CAP_TEXTURE_HALF_FLOAT_LINEAR``: Whether the linear minification and
273  magnification filters are supported with half-precision floating-point
274  textures.
275* ``PIPE_CAP_DEPTH_BOUNDS_TEST``: Whether bounds_test, bounds_min, and
276  bounds_max states of pipe_depth_stencil_alpha_state behave according
277  to the GL_EXT_depth_bounds_test specification.
278* ``PIPE_CAP_TGSI_TXQS``: Whether the `TXQS` opcode is supported
279* ``PIPE_CAP_FORCE_PERSAMPLE_INTERP``: If the driver can force per-sample
280  interpolation for all fragment shader inputs if
281  pipe_rasterizer_state::force_persample_interp is set. This is only used
282  by GL3-level sample shading (ARB_sample_shading). GL4-level sample shading
283  (ARB_gpu_shader5) doesn't use this. While GL3 hardware has a state for it,
284  GL4 hardware will likely need to emulate it with a shader variant, or by
285  selecting the interpolation weights with a conditional assignment
286  in the shader.
287* ``PIPE_CAP_SHAREABLE_SHADERS``: Whether shader CSOs can be used by any
288  pipe_context.
289* ``PIPE_CAP_COPY_BETWEEN_COMPRESSED_AND_PLAIN_FORMATS``:
290  Whether copying between compressed and plain formats is supported where
291  a compressed block is copied to/from a plain pixel of the same size.
292* ``PIPE_CAP_CLEAR_TEXTURE``: Whether `clear_texture` will be
293  available in contexts.
294* ``PIPE_CAP_DRAW_PARAMETERS``: Whether ``TGSI_SEMANTIC_BASEVERTEX``,
295  ``TGSI_SEMANTIC_BASEINSTANCE``, and ``TGSI_SEMANTIC_DRAWID`` are
296  supported in vertex shaders.
297* ``PIPE_CAP_TGSI_PACK_HALF_FLOAT``: Whether the ``UP2H`` and ``PK2H``
298  TGSI opcodes are supported.
299* ``PIPE_CAP_TGSI_FS_POSITION_IS_SYSVAL``: If state trackers should use
300  a system value for the POSITION fragment shader input.
301* ``PIPE_CAP_TGSI_FS_FACE_IS_INTEGER_SYSVAL``: If state trackers should use
302  a system value for the FACE fragment shader input.
303  Also, the FACE system value is integer, not float.
304* ``PIPE_CAP_SHADER_BUFFER_OFFSET_ALIGNMENT``: Describes the required
305  alignment for pipe_shader_buffer::buffer_offset, in bytes. Maximum
306  value allowed is 256 (for GL conformance). 0 is only allowed if
307  shader buffers are not supported.
308* ``PIPE_CAP_INVALIDATE_BUFFER``: Whether the use of ``invalidate_resource``
309  for buffers is supported.
310* ``PIPE_CAP_GENERATE_MIPMAP``: Indicates whether pipe_context::generate_mipmap
311  is supported.
312* ``PIPE_CAP_STRING_MARKER``: Whether pipe->emit_string_marker() is supported.
313* ``PIPE_CAP_SURFACE_REINTERPRET_BLOCKS``: Indicates whether
314  pipe_context::create_surface supports reinterpreting a texture as a surface
315  of a format with different block width/height (but same block size in bits).
316  For example, a compressed texture image can be interpreted as a
317  non-compressed surface whose texels are the same number of bits as the
318  compressed blocks, and vice versa. The width and height of the surface is
319  adjusted appropriately.
320* ``PIPE_CAP_QUERY_BUFFER_OBJECT``: Driver supports
321  context::get_query_result_resource callback.
322* ``PIPE_CAP_PCI_GROUP``: Return the PCI segment group number.
323* ``PIPE_CAP_PCI_BUS``: Return the PCI bus number.
324* ``PIPE_CAP_PCI_DEVICE``: Return the PCI device number.
325* ``PIPE_CAP_PCI_FUNCTION``: Return the PCI function number.
326* ``PIPE_CAP_FRAMEBUFFER_NO_ATTACHMENT``:
327  If non-zero, rendering to framebuffers with no surface attachments
328  is supported. The context->is_format_supported function will be expected
329  to be implemented with PIPE_FORMAT_NONE yeilding the MSAA modes the hardware
330  supports. N.B., The maximum number of layers supported for rasterizing a
331  primitive on a layer is obtained from ``PIPE_CAP_MAX_TEXTURE_ARRAY_LAYERS``
332  even though it can be larger than the number of layers supported by either
333  rendering or textures.
334* ``PIPE_CAP_ROBUST_BUFFER_ACCESS_BEHAVIOR``: Implementation uses bounds
335  checking on resource accesses by shader if the context is created with
336  PIPE_CONTEXT_ROBUST_BUFFER_ACCESS. See the ARB_robust_buffer_access_behavior
337  extension for information on the required behavior for out of bounds accesses
338  and accesses to unbound resources.
339* ``PIPE_CAP_CULL_DISTANCE``: Whether the driver supports the arb_cull_distance
340  extension and thus implements proper support for culling planes.
341* ``PIPE_CAP_PRIMITIVE_RESTART_FOR_PATCHES``: Whether primitive restart is
342  supported for patch primitives.
343* ``PIPE_CAP_TGSI_VOTE``: Whether the ``VOTE_*`` ops can be used in shaders.
344* ``PIPE_CAP_MAX_WINDOW_RECTANGLES``: The maxium number of window rectangles
345  supported in ``set_window_rectangles``.
346* ``PIPE_CAP_POLYGON_OFFSET_UNITS_UNSCALED``: If true, the driver implements support
347  for ``pipe_rasterizer_state::offset_units_unscaled``.
348* ``PIPE_CAP_VIEWPORT_SUBPIXEL_BITS``: Number of bits of subpixel precision for
349  floating point viewport bounds.
350* ``PIPE_CAP_MIXED_COLOR_DEPTH_BITS``: Whether there is non-fallback
351  support for color/depth format combinations that use a different
352  number of bits. For the purpose of this cap, Z24 is treated as
353  32-bit. If set to off, that means that a B5G6R5 + Z24 or RGBA8 + Z16
354  combination will require a driver fallback, and should not be
355  advertised in the GLX/EGL config list.
356* ``PIPE_CAP_TGSI_ARRAY_COMPONENTS``: If true, the driver interprets the
357  UsageMask of input and output declarations and allows declaring arrays
358  in overlapping ranges. The components must be a contiguous range, e.g. a
359  UsageMask of  xy or yzw is allowed, but xz or yw isn't. Declarations with
360  overlapping locations must have matching semantic names and indices, and
361  equal interpolation qualifiers.
362  Components may overlap, notably when the gaps in an array of dvec3 are
363  filled in.
364* ``PIPE_CAP_STREAM_OUTPUT_INTERLEAVE_BUFFERS``: Whether interleaved stream
365  output mode is able to interleave across buffers. This is required for
366  ARB_transform_feedback3.
367* ``PIPE_CAP_TGSI_CAN_READ_OUTPUTS``: Whether every TGSI shader stage can read
368  from the output file.
369* ``PIPE_CAP_GLSL_OPTIMIZE_CONSERVATIVELY``: Tell the GLSL compiler to use
370  the minimum amount of optimizations just to be able to do all the linking
371  and lowering.
372* ``PIPE_CAP_TGSI_FS_FBFETCH``: Whether a fragment shader can use the FBFETCH
373  opcode to retrieve the current value in the framebuffer.
374
375
376.. _pipe_capf:
377
378PIPE_CAPF_*
379^^^^^^^^^^^^^^^^
380
381The floating-point capabilities are:
382
383* ``PIPE_CAPF_MAX_LINE_WIDTH``: The maximum width of a regular line.
384* ``PIPE_CAPF_MAX_LINE_WIDTH_AA``: The maximum width of a smoothed line.
385* ``PIPE_CAPF_MAX_POINT_WIDTH``: The maximum width and height of a point.
386* ``PIPE_CAPF_MAX_POINT_WIDTH_AA``: The maximum width and height of a smoothed point.
387* ``PIPE_CAPF_MAX_TEXTURE_ANISOTROPY``: The maximum level of anisotropy that can be
388  applied to anisotropically filtered textures.
389* ``PIPE_CAPF_MAX_TEXTURE_LOD_BIAS``: The maximum :term:`LOD` bias that may be applied
390  to filtered textures.
391* ``PIPE_CAPF_GUARD_BAND_LEFT``,
392  ``PIPE_CAPF_GUARD_BAND_TOP``,
393  ``PIPE_CAPF_GUARD_BAND_RIGHT``,
394  ``PIPE_CAPF_GUARD_BAND_BOTTOM``: TODO
395
396
397.. _pipe_shader_cap:
398
399PIPE_SHADER_CAP_*
400^^^^^^^^^^^^^^^^^
401
402These are per-shader-stage capabitity queries. Different shader stages may
403support different features.
404
405* ``PIPE_SHADER_CAP_MAX_INSTRUCTIONS``: The maximum number of instructions.
406* ``PIPE_SHADER_CAP_MAX_ALU_INSTRUCTIONS``: The maximum number of arithmetic instructions.
407* ``PIPE_SHADER_CAP_MAX_TEX_INSTRUCTIONS``: The maximum number of texture instructions.
408* ``PIPE_SHADER_CAP_MAX_TEX_INDIRECTIONS``: The maximum number of texture indirections.
409* ``PIPE_SHADER_CAP_MAX_CONTROL_FLOW_DEPTH``: The maximum nested control flow depth.
410* ``PIPE_SHADER_CAP_MAX_INPUTS``: The maximum number of input registers.
411* ``PIPE_SHADER_CAP_MAX_OUTPUTS``: The maximum number of output registers.
412  This is valid for all shaders except the fragment shader.
413* ``PIPE_SHADER_CAP_MAX_CONST_BUFFER_SIZE``: The maximum size per constant buffer in bytes.
414* ``PIPE_SHADER_CAP_MAX_CONST_BUFFERS``: Maximum number of constant buffers that can be bound
415  to any shader stage using ``set_constant_buffer``. If 0 or 1, the pipe will
416  only permit binding one constant buffer per shader, and the shaders will
417  not permit two-dimensional access to constants.
418
419If a value greater than 0 is returned, the driver can have multiple
420constant buffers bound to shader stages. The CONST register file can
421be accessed with two-dimensional indices, like in the example below.
422
423DCL CONST[0][0..7]       # declare first 8 vectors of constbuf 0
424DCL CONST[3][0]          # declare first vector of constbuf 3
425MOV OUT[0], CONST[0][3]  # copy vector 3 of constbuf 0
426
427For backwards compatibility, one-dimensional access to CONST register
428file is still supported. In that case, the constbuf index is assumed
429to be 0.
430
431* ``PIPE_SHADER_CAP_MAX_TEMPS``: The maximum number of temporary registers.
432* ``PIPE_SHADER_CAP_MAX_PREDS``: The maximum number of predicate registers.
433* ``PIPE_SHADER_CAP_TGSI_CONT_SUPPORTED``: Whether the continue opcode is supported.
434* ``PIPE_SHADER_CAP_INDIRECT_INPUT_ADDR``: Whether indirect addressing
435  of the input file is supported.
436* ``PIPE_SHADER_CAP_INDIRECT_OUTPUT_ADDR``: Whether indirect addressing
437  of the output file is supported.
438* ``PIPE_SHADER_CAP_INDIRECT_TEMP_ADDR``: Whether indirect addressing
439  of the temporary file is supported.
440* ``PIPE_SHADER_CAP_INDIRECT_CONST_ADDR``: Whether indirect addressing
441  of the constant file is supported.
442* ``PIPE_SHADER_CAP_SUBROUTINES``: Whether subroutines are supported, i.e.
443  BGNSUB, ENDSUB, CAL, and RET, including RET in the main block.
444* ``PIPE_SHADER_CAP_INTEGERS``: Whether integer opcodes are supported.
445  If unsupported, only float opcodes are supported.
446* ``PIPE_SHADER_CAP_MAX_TEXTURE_SAMPLERS``: The maximum number of texture
447  samplers.
448* ``PIPE_SHADER_CAP_PREFERRED_IR``: Preferred representation of the
449  program.  It should be one of the ``pipe_shader_ir`` enum values.
450* ``PIPE_SHADER_CAP_MAX_SAMPLER_VIEWS``: The maximum number of texture
451  sampler views. Must not be lower than PIPE_SHADER_CAP_MAX_TEXTURE_SAMPLERS.
452* ``PIPE_SHADER_CAP_DOUBLES``: Whether double precision floating-point
453  operations are supported.
454* ``PIPE_SHADER_CAP_TGSI_DROUND_SUPPORTED``: Whether double precision rounding
455  is supported. If it is, DTRUNC/DCEIL/DFLR/DROUND opcodes may be used.
456* ``PIPE_SHADER_CAP_TGSI_DFRACEXP_DLDEXP_SUPPORTED``: Whether DFRACEXP and
457  DLDEXP are supported.
458* ``PIPE_SHADER_CAP_TGSI_FMA_SUPPORTED``: Whether FMA and DFMA (doubles only)
459  are supported.
460* ``PIPE_SHADER_CAP_TGSI_ANY_INOUT_DECL_RANGE``: Whether the driver doesn't
461  ignore tgsi_declaration_range::Last for shader inputs and outputs.
462* ``PIPE_SHADER_CAP_MAX_UNROLL_ITERATIONS_HINT``: This is the maximum number
463  of iterations that loops are allowed to have to be unrolled. It is only
464  a hint to state trackers. Whether any loops will be unrolled is not
465  guaranteed.
466* ``PIPE_SHADER_CAP_MAX_SHADER_BUFFERS``: Maximum number of memory buffers
467  (also used to implement atomic counters). Having this be non-0 also
468  implies support for the ``LOAD``, ``STORE``, and ``ATOM*`` TGSI
469  opcodes.
470* ``PIPE_SHADER_CAP_SUPPORTED_IRS``: Supported representations of the
471  program.  It should be a mask of ``pipe_shader_ir`` bits.
472* ``PIPE_SHADER_CAP_MAX_SHADER_IMAGES``: Maximum number of image units.
473* ``PIPE_SHADER_CAP_LOWER_IF_THRESHOLD``: IF and ELSE branches with a lower
474  cost than this value should be lowered by the state tracker for better
475  performance. This is a tunable for the GLSL compiler and the behavior is
476  specific to the compiler.
477
478
479.. _pipe_compute_cap:
480
481PIPE_COMPUTE_CAP_*
482^^^^^^^^^^^^^^^^^^
483
484Compute-specific capabilities. They can be queried using
485pipe_screen::get_compute_param.
486
487* ``PIPE_COMPUTE_CAP_IR_TARGET``: A description of the target of the form
488  ``processor-arch-manufacturer-os`` that will be passed on to the compiler.
489  This CAP is only relevant for drivers that specify PIPE_SHADER_IR_LLVM
490  or PIPE_SHADER_IR_NATIVE for their preferred IR.
491  Value type: null-terminated string. Shader IR type dependent.
492* ``PIPE_COMPUTE_CAP_GRID_DIMENSION``: Number of supported dimensions
493  for grid and block coordinates.  Value type: ``uint64_t``. Shader IR type dependent.
494* ``PIPE_COMPUTE_CAP_MAX_GRID_SIZE``: Maximum grid size in block
495  units.  Value type: ``uint64_t []``.  Shader IR type dependent.
496* ``PIPE_COMPUTE_CAP_MAX_BLOCK_SIZE``: Maximum block size in thread
497  units.  Value type: ``uint64_t []``. Shader IR type dependent.
498* ``PIPE_COMPUTE_CAP_MAX_THREADS_PER_BLOCK``: Maximum number of threads that
499  a single block can contain.  Value type: ``uint64_t``. Shader IR type dependent.
500  This may be less than the product of the components of MAX_BLOCK_SIZE and is
501  usually limited by the number of threads that can be resident simultaneously
502  on a compute unit.
503* ``PIPE_COMPUTE_CAP_MAX_GLOBAL_SIZE``: Maximum size of the GLOBAL
504  resource.  Value type: ``uint64_t``. Shader IR type dependent.
505* ``PIPE_COMPUTE_CAP_MAX_LOCAL_SIZE``: Maximum size of the LOCAL
506  resource.  Value type: ``uint64_t``. Shader IR type dependent.
507* ``PIPE_COMPUTE_CAP_MAX_PRIVATE_SIZE``: Maximum size of the PRIVATE
508  resource.  Value type: ``uint64_t``. Shader IR type dependent.
509* ``PIPE_COMPUTE_CAP_MAX_INPUT_SIZE``: Maximum size of the INPUT
510  resource.  Value type: ``uint64_t``. Shader IR type dependent.
511* ``PIPE_COMPUTE_CAP_MAX_MEM_ALLOC_SIZE``: Maximum size of a memory object
512  allocation in bytes.  Value type: ``uint64_t``.
513* ``PIPE_COMPUTE_CAP_MAX_CLOCK_FREQUENCY``: Maximum frequency of the GPU
514  clock in MHz. Value type: ``uint32_t``
515* ``PIPE_COMPUTE_CAP_MAX_COMPUTE_UNITS``: Maximum number of compute units
516  Value type: ``uint32_t``
517* ``PIPE_COMPUTE_CAP_IMAGES_SUPPORTED``: Whether images are supported
518  non-zero means yes, zero means no. Value type: ``uint32_t``
519* ``PIPE_COMPUTE_CAP_SUBGROUP_SIZE``: The size of a basic execution unit in
520  threads. Also known as wavefront size, warp size or SIMD width.
521* ``PIPE_COMPUTE_CAP_ADDRESS_BITS``: The default compute device address space
522  size specified as an unsigned integer value in bits.
523* ``PIPE_COMPUTE_CAP_MAX_VARIABLE_THREADS_PER_BLOCK``: Maximum variable number
524  of threads that a single block can contain. This is similar to
525  PIPE_COMPUTE_CAP_MAX_THREADS_PER_BLOCK, except that the variable size is not
526  known a compile-time but at dispatch-time.
527
528.. _pipe_bind:
529
530PIPE_BIND_*
531^^^^^^^^^^^
532
533These flags indicate how a resource will be used and are specified at resource
534creation time. Resources may be used in different roles
535during their lifecycle. Bind flags are cumulative and may be combined to create
536a resource which can be used for multiple things.
537Depending on the pipe driver's memory management and these bind flags,
538resources might be created and handled quite differently.
539
540* ``PIPE_BIND_RENDER_TARGET``: A color buffer or pixel buffer which will be
541  rendered to.  Any surface/resource attached to pipe_framebuffer_state::cbufs
542  must have this flag set.
543* ``PIPE_BIND_DEPTH_STENCIL``: A depth (Z) buffer and/or stencil buffer. Any
544  depth/stencil surface/resource attached to pipe_framebuffer_state::zsbuf must
545  have this flag set.
546* ``PIPE_BIND_BLENDABLE``: Used in conjunction with PIPE_BIND_RENDER_TARGET to
547  query whether a device supports blending for a given format.
548  If this flag is set, surface creation may fail if blending is not supported
549  for the specified format. If it is not set, a driver may choose to ignore
550  blending on surfaces with formats that would require emulation.
551* ``PIPE_BIND_DISPLAY_TARGET``: A surface that can be presented to screen. Arguments to
552  pipe_screen::flush_front_buffer must have this flag set.
553* ``PIPE_BIND_SAMPLER_VIEW``: A texture that may be sampled from in a fragment
554  or vertex shader.
555* ``PIPE_BIND_VERTEX_BUFFER``: A vertex buffer.
556* ``PIPE_BIND_INDEX_BUFFER``: An vertex index/element buffer.
557* ``PIPE_BIND_CONSTANT_BUFFER``: A buffer of shader constants.
558* ``PIPE_BIND_STREAM_OUTPUT``: A stream output buffer.
559* ``PIPE_BIND_CUSTOM``:
560* ``PIPE_BIND_SCANOUT``: A front color buffer or scanout buffer.
561* ``PIPE_BIND_SHARED``: A sharable buffer that can be given to another
562  process.
563* ``PIPE_BIND_GLOBAL``: A buffer that can be mapped into the global
564  address space of a compute program.
565* ``PIPE_BIND_SHADER_BUFFER``: A buffer without a format that can be bound
566  to a shader and can be used with load, store, and atomic instructions.
567* ``PIPE_BIND_SHADER_IMAGE``: A buffer or texture with a format that can be
568  bound to a shader and can be used with load, store, and atomic instructions.
569* ``PIPE_BIND_COMPUTE_RESOURCE``: A buffer or texture that can be
570  bound to the compute program as a shader resource.
571* ``PIPE_BIND_COMMAND_ARGS_BUFFER``: A buffer that may be sourced by the
572  GPU command processor. It can contain, for example, the arguments to
573  indirect draw calls.
574
575.. _pipe_usage:
576
577PIPE_USAGE_*
578^^^^^^^^^^^^
579
580The PIPE_USAGE enums are hints about the expected usage pattern of a resource.
581Note that drivers must always support read and write CPU access at any time
582no matter which hint they got.
583
584* ``PIPE_USAGE_DEFAULT``: Optimized for fast GPU access.
585* ``PIPE_USAGE_IMMUTABLE``: Optimized for fast GPU access and the resource is
586  not expected to be mapped or changed (even by the GPU) after the first upload.
587* ``PIPE_USAGE_DYNAMIC``: Expect frequent write-only CPU access. What is
588  uploaded is expected to be used at least several times by the GPU.
589* ``PIPE_USAGE_STREAM``: Expect frequent write-only CPU access. What is
590  uploaded is expected to be used only once by the GPU.
591* ``PIPE_USAGE_STAGING``: Optimized for fast CPU access.
592
593
594Methods
595-------
596
597XXX to-do
598
599get_name
600^^^^^^^^
601
602Returns an identifying name for the screen.
603
604get_vendor
605^^^^^^^^^^
606
607Returns the screen vendor.
608
609get_device_vendor
610^^^^^^^^^^^^^^^^^
611
612Returns the actual vendor of the device driving the screen
613(as opposed to the driver vendor).
614
615.. _get_param:
616
617get_param
618^^^^^^^^^
619
620Get an integer/boolean screen parameter.
621
622**param** is one of the :ref:`PIPE_CAP` names.
623
624.. _get_paramf:
625
626get_paramf
627^^^^^^^^^^
628
629Get a floating-point screen parameter.
630
631**param** is one of the :ref:`PIPE_CAP` names.
632
633context_create
634^^^^^^^^^^^^^^
635
636Create a pipe_context.
637
638**priv** is private data of the caller, which may be put to various
639unspecified uses, typically to do with implementing swapbuffers
640and/or front-buffer rendering.
641
642is_format_supported
643^^^^^^^^^^^^^^^^^^^
644
645Determine if a resource in the given format can be used in a specific manner.
646
647**format** the resource format
648
649**target** one of the PIPE_TEXTURE_x flags
650
651**sample_count** the number of samples. 0 and 1 mean no multisampling,
652the maximum allowed legal value is 32.
653
654**bindings** is a bitmask of :ref:`PIPE_BIND` flags.
655
656**geom_flags** is a bitmask of PIPE_TEXTURE_GEOM_x flags.
657
658Returns TRUE if all usages can be satisfied.
659
660
661can_create_resource
662^^^^^^^^^^^^^^^^^^^
663
664Check if a resource can actually be created (but don't actually allocate any
665memory).  This is used to implement OpenGL's proxy textures.  Typically, a
666driver will simply check if the total size of the given resource is less than
667some limit.
668
669For PIPE_TEXTURE_CUBE, the pipe_resource::array_size field should be 6.
670
671
672.. _resource_create:
673
674resource_create
675^^^^^^^^^^^^^^^
676
677Create a new resource from a template.
678The following fields of the pipe_resource must be specified in the template:
679
680**target** one of the pipe_texture_target enums.
681Note that PIPE_BUFFER and PIPE_TEXTURE_X are not really fundamentally different.
682Modern APIs allow using buffers as shader resources.
683
684**format** one of the pipe_format enums.
685
686**width0** the width of the base mip level of the texture or size of the buffer.
687
688**height0** the height of the base mip level of the texture
689(1 for 1D or 1D array textures).
690
691**depth0** the depth of the base mip level of the texture
692(1 for everything else).
693
694**array_size** the array size for 1D and 2D array textures.
695For cube maps this must be 6, for other textures 1.
696
697**last_level** the last mip map level present.
698
699**nr_samples** the nr of msaa samples. 0 (or 1) specifies a resource
700which isn't multisampled.
701
702**usage** one of the PIPE_USAGE flags.
703
704**bind** bitmask of the PIPE_BIND flags.
705
706**flags** bitmask of PIPE_RESOURCE_FLAG flags.
707
708
709
710resource_destroy
711^^^^^^^^^^^^^^^^
712
713Destroy a resource. A resource is destroyed if it has no more references.
714
715
716
717get_timestamp
718^^^^^^^^^^^^^
719
720Query a timestamp in nanoseconds. The returned value should match
721PIPE_QUERY_TIMESTAMP. This function returns immediately and doesn't
722wait for rendering to complete (which cannot be achieved with queries).
723
724
725
726get_driver_query_info
727^^^^^^^^^^^^^^^^^^^^^
728
729Return a driver-specific query. If the **info** parameter is NULL,
730the number of available queries is returned.  Otherwise, the driver
731query at the specified **index** is returned in **info**.
732The function returns non-zero on success.
733The driver-specific query is described with the pipe_driver_query_info
734structure.
735
736get_driver_query_group_info
737^^^^^^^^^^^^^^^^^^^^^^^^^^^
738
739Return a driver-specific query group. If the **info** parameter is NULL,
740the number of available groups is returned.  Otherwise, the driver
741query group at the specified **index** is returned in **info**.
742The function returns non-zero on success.
743The driver-specific query group is described with the
744pipe_driver_query_group_info structure.
745
746
747Thread safety
748-------------
749
750Screen methods are required to be thread safe. While gallium rendering
751contexts are not required to be thread safe, it is required to be safe to use
752different contexts created with the same screen in different threads without
753locks. It is also required to be safe using screen methods in a thread, while
754using one of its contexts in another (without locks).
755