On 08/08/2014 07:55 AM, Roland Scheidegger wrote:
> The mesa parts of the series all look good to me.
So.. can we put your R-b on patches 1 through 7? :)
> We definitely want something like that in gallium too (draw fails the
> vertex id tests sort of on purpose right now because we needed d3d10
> behavior).
>
> Oh and sort of off-topic but since you're familiar with it do you know
> why the gl_PrimitiveIn of the geometry shader isn't a system value
> whereas all the vertex_id and friends are?
I have come vague recollection that Paul had a good reason for doing
that, but no recollection of what that reason was.
> Roland
>
>
> Am 08.08.2014 09:31, schrieb Kenneth Graunke:
>> From: Ian Romanick <ian.d.roman...@intel.com>
>>
>> v2: Additions to the documentation for SYSTEM_VALUE_VERTEX_ID. Quote
>> the GL_ARB_shader_draw_parameters spec and mention DirectX SV_VertexID.
>>
>> Signed-off-by: Ian Romanick <ian.d.roman...@intel.com>
>> Reviewed-by: Kenneth Graunke <kenn...@whitecape.org>
>> ---
>> src/mesa/main/mtypes.h | 57
>> ++++++++++++++++++++++++++++++++++++++++++++++++++
>> 1 file changed, 57 insertions(+)
>>
>> This series is available as the 'basevertex-v9' branch of ~kwg/mesa
>> (not ~idr/mesa). Ken tested this series against Piglit on Haswell and
>> Broadwell, but did not test earlier hardware, nor run the ES3 tests.
>>
>> diff --git a/src/mesa/main/mtypes.h b/src/mesa/main/mtypes.h
>> index ff130da..207be0a 100644
>> --- a/src/mesa/main/mtypes.h
>> +++ b/src/mesa/main/mtypes.h
>> @@ -2055,7 +2055,64 @@ typedef enum
>> * \name Vertex shader system values
>> */
>> /*@{*/
>> + /**
>> + * OpenGL-style vertex ID.
>> + *
>> + * Section 2.11.7 (Shader Execution), subsection Shader Inputs, of the
>> + * OpenGL 3.3 core profile spec says:
>> + *
>> + * "gl_VertexID holds the integer index i implicitly passed by
>> + * DrawArrays or one of the other drawing commands defined in section
>> + * 2.8.3."
>> + *
>> + * Section 2.8.3 (Drawing Commands) of the same spec says:
>> + *
>> + * "The commands....are equivalent to the commands with the same base
>> + * name (without the BaseVertex suffix), except that the ith element
>> + * transferred by the corresponding draw call will be taken from
>> + * element indices[i] + basevertex of each enabled array."
>> + *
>> + * Additionally, the overview in the GL_ARB_shader_draw_parameters spec
>> + * says:
>> + *
>> + * "In unextended GL, vertex shaders have inputs named gl_VertexID
>> and
>> + * gl_InstanceID, which contain, respectively the index of the vertex
>> + * and instance. The value of gl_VertexID is the implicitly passed
>> + * index of the vertex being processed, which includes the value of
>> + * baseVertex, for those commands that accept it."
>> + *
>> + * gl_VertexID gets basevertex added in. This differs from DirectX where
>> + * SV_VertexID does \b not get basevertex added in.
>> + */
>> SYSTEM_VALUE_VERTEX_ID,
>> +
>> + /**
>> + * Instanced ID as supplied to gl_InstanceID
>> + *
>> + * Values assigned to gl_InstanceID always begin with zero, regardless of
>> + * the value of baseinstance.
>> + *
>> + * Section 11.1.3.9 (Shader Inputs) of the OpenGL 4.4 core profile spec
>> + * says:
>> + *
>> + * "gl_InstanceID holds the integer instance number of the current
>> + * primitive in an instanced draw call (see section 10.5)."
>> + *
>> + * Through a big chain of pseudocode, section 10.5 describes that
>> + * baseinstance is not counted by gl_InstanceID. In that section, notice
>> + *
>> + * "If an enabled vertex attribute array is instanced (it has a
>> + * non-zero divisor as specified by VertexAttribDivisor), the element
>> + * index that is transferred to the GL, for all vertices, is given by
>> + *
>> + * floor(instance/divisor) + baseinstance
>> + *
>> + * If an array corresponding to an attribute required by a vertex
>> + * shader is not enabled, then the corresponding element is taken
>> from
>> + * the current attribute state (see section 10.2)."
>> + *
>> + * Note that baseinstance is \b not included in the value of instance.
>> + */
>> SYSTEM_VALUE_INSTANCE_ID,
>> /*@}*/
>>
>>
>
> _______________________________________________
> mesa-dev mailing list
> mesa-dev@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/mesa-dev
>
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
