[Mesa-dev] [PATCH] llvmpipe: Remove the ARB_draw_instanced capability.
That capability requires integer handling and that's not yet active, ending with a failure in draw-non-instanced unless you force it on. See bug 51366. Frankly, I'd rather have that patch rejected and integer/glsl 130 capability activated instead. There still are things missing, but they mostly have their own extension anyway. And the overall picture ain't so bad. Signed-off-by: Olivier Galibert galib...@pobox.com diff --git a/src/gallium/drivers/llvmpipe/lp_screen.c b/src/gallium/drivers/llvmpipe/lp_screen.c index 40037a5..5eb826e 100644 --- a/src/gallium/drivers/llvmpipe/lp_screen.c +++ b/src/gallium/drivers/llvmpipe/lp_screen.c @@ -152,8 +152,6 @@ llvmpipe_get_param(struct pipe_screen *screen, enum pipe_cap param) return 1; case PIPE_CAP_DEPTH_CLIP_DISABLE: return 0; - case PIPE_CAP_TGSI_INSTANCEID: - case PIPE_CAP_VERTEX_ELEMENT_INSTANCE_DIVISOR: case PIPE_CAP_MIXED_COLORBUFFER_FORMATS: case PIPE_CAP_CONDITIONAL_RENDER: return 1; ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
Re: [Mesa-dev] [PATCH] llvmpipe: Remove the ARB_draw_instanced capability.
On 25.06.2012 08:37, Olivier Galibert wrote: That capability requires integer handling and that's not yet active, ending with a failure in draw-non-instanced unless you force it on. See bug 51366. Frankly, I'd rather have that patch rejected and integer/glsl 130 capability activated instead. There still are things missing, but they mostly have their own extension anyway. And the overall picture ain't so bad. Signed-off-by: Olivier Galibert galib...@pobox.com diff --git a/src/gallium/drivers/llvmpipe/lp_screen.c b/src/gallium/drivers/llvmpipe/lp_screen.c index 40037a5..5eb826e 100644 --- a/src/gallium/drivers/llvmpipe/lp_screen.c +++ b/src/gallium/drivers/llvmpipe/lp_screen.c @@ -152,8 +152,6 @@ llvmpipe_get_param(struct pipe_screen *screen, enum pipe_cap param) return 1; case PIPE_CAP_DEPTH_CLIP_DISABLE: return 0; - case PIPE_CAP_TGSI_INSTANCEID: - case PIPE_CAP_VERTEX_ELEMENT_INSTANCE_DIVISOR: With VERTEX_ELEMENT_INSTANCE_DIVISOR you'd be removing GL_ARB_instanced_arrays, which doesn't need integers. case PIPE_CAP_MIXED_COLORBUFFER_FORMATS: case PIPE_CAP_CONDITIONAL_RENDER: return 1; ___ 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
Re: [Mesa-dev] [PATCH] llvmpipe: Remove the ARB_draw_instanced capability.
- Original Message - That capability requires integer handling and that's not yet active, ending with a failure in draw-non-instanced unless you force it on. See bug 51366. Frankly, I'd rather have that patch rejected and integer/glsl 130 capability activated instead. There still are things missing, but they mostly have their own extension anyway. And the overall picture ain't so bad. I'm personally also more interested in seeing llvmpipe to get the missing features for GLSL 1.30 / OGL 3. What's the overall picture of llvmpipe w/ integer/glsl 130? That is, how many piglit tests go from skipped to passed/failed? Jose ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
Re: [Mesa-dev] [PATCH] llvmpipe: Remove the ARB_draw_instanced capability.
On Mon, Jun 25, 2012 at 05:34:25AM -0700, Jose Fonseca wrote: - Original Message - That capability requires integer handling and that's not yet active, ending with a failure in draw-non-instanced unless you force it on. See bug 51366. Frankly, I'd rather have that patch rejected and integer/glsl 130 capability activated instead. There still are things missing, but they mostly have their own extension anyway. And the overall picture ain't so bad. I'm personally also more interested in seeing llvmpipe to get the missing features for GLSL 1.30 / OGL 3. What's the overall picture of llvmpipe w/ integer/glsl 130? That is, how many piglit tests go from skipped to passed/failed? To failed: precision-05.vert link-mismatch-layout-02 no-redeclaration-01.vert feature-macro.vert fs-exec-after-break - general failures, everybody has them vs-clip-distance-bulk-assign vs-clip-distance-inout-param vs-clip-distance-out-param vs-clip-distance-retval - haven't checked what the problem is, softpipe has it right fs-isinf-vec2 fs-isinf-vec3 fs-isinf-vec4 vs-isinf-vec2 vs-isinf-vec3 vs-isinf-vec4 - test is iffy fs-texelFetch-2D fs-texelFetchOffset-2D - no texelFetch support yet fs-texture-sampler2dshadow-10 fs-texture-sampler2dshadow-11 - dunno what's going on, softpipe fails it too vs-attrib-ivec4-implied vs-attrib-ivec4-precision vs-attrib-uvec4-implied vs-attrib-uvec4-precision - use glVertexAttribIPointer, which is GL 3.0+ only vs-textureLod-miplevels - issue with vertex shader invalidation when sampler mode changes (as in, it's not done) vs-textureLod-miplevels-2 - you know that one, it's nowhere near fixed yet (the softpipe patch is awaiting review too :-) texel-offset-limits - no limits defined in lp_screen.c, udnno whether texture() would take it into account either To pass: 1503 total, it seems, you can be sure I'm not going to list them :-) Best, OG. ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
Re: [Mesa-dev] [PATCH] llvmpipe: Remove the ARB_draw_instanced capability.
- Original Message - On Mon, Jun 25, 2012 at 05:34:25AM -0700, Jose Fonseca wrote: - Original Message - That capability requires integer handling and that's not yet active, ending with a failure in draw-non-instanced unless you force it on. See bug 51366. Frankly, I'd rather have that patch rejected and integer/glsl 130 capability activated instead. There still are things missing, but they mostly have their own extension anyway. And the overall picture ain't so bad. I'm personally also more interested in seeing llvmpipe to get the missing features for GLSL 1.30 / OGL 3. What's the overall picture of llvmpipe w/ integer/glsl 130? That is, how many piglit tests go from skipped to passed/failed? To failed: precision-05.vert link-mismatch-layout-02 no-redeclaration-01.vert feature-macro.vert fs-exec-after-break - general failures, everybody has them vs-clip-distance-bulk-assign vs-clip-distance-inout-param vs-clip-distance-out-param vs-clip-distance-retval - haven't checked what the problem is, softpipe has it right fs-isinf-vec2 fs-isinf-vec3 fs-isinf-vec4 vs-isinf-vec2 vs-isinf-vec3 vs-isinf-vec4 - test is iffy fs-texelFetch-2D fs-texelFetchOffset-2D - no texelFetch support yet fs-texture-sampler2dshadow-10 fs-texture-sampler2dshadow-11 - dunno what's going on, softpipe fails it too vs-attrib-ivec4-implied vs-attrib-ivec4-precision vs-attrib-uvec4-implied vs-attrib-uvec4-precision - use glVertexAttribIPointer, which is GL 3.0+ only vs-textureLod-miplevels - issue with vertex shader invalidation when sampler mode changes (as in, it's not done) vs-textureLod-miplevels-2 - you know that one, it's nowhere near fixed yet (the softpipe patch is awaiting review too :-) texel-offset-limits - no limits defined in lp_screen.c, udnno whether texture() would take it into account either To pass: 1503 total, it seems, you can be sure I'm not going to list them :-) Thanks for the summary. I think that the biggest missing feature is texelFetch -- once there's a patch for it I'm happy to advertise GLSL 1.30, and treat the rest as bugs. Jose ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev