<div dir="ltr">On 10 September 2013 16:28, Chad Versace <<a href="mailto:chad.versace@linux.intel.com" target="_blank">chad.versace@linux.intel.com</a>> wrote: <div class="gmail_extra"><div class="gmail_quote"> <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div class="HOEnZb"><div class="h5">On 09/04/2013 02:22 PM, Paul Berry wrote: <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> Most geometry shaders access their input arrays using a constant index (or an index that is constant after loop unrolling). This test checks that we get the right results if we access a geometry shader input array using a non-constant index. --- .../geometry/dynamic_input_array_index.shader_test | 67 ++++++++++++++++++++++ 1 file changed, 67 insertions(+) create mode 100644 tests/spec/glsl-1.50/execution/geometry/dynamic_input_array_index.shader_test diff --git a/tests/spec/glsl-1.50/execution/geometry/dynamic_input_array_index.shader_test b/tests/spec/glsl-1.50/execution/geometry/dynamic_input_array_index.shader_test new file mode 100644 index 0000000..685a66a --- /dev/null +++ b/tests/spec/glsl-1.50/execution/geometry/dynamic_input_array_index.shader_test @@ -0,0 +1,67 @@ +# Verify that a geometry shader can access an input array using a +# dynamic index rather than a constant index. + +[require] +GL >= 3.2 +GLSL >= 1.50 + +[vertex shader] +#version 150 + +in vec4 vertex; +in vec4 color; +in int index; +out vec4 vertex_to_gs; +out vec4 color_to_gs; +out int index_to_gs; + +void main() +{ + vertex_to_gs = vertex; + color_to_gs = color; + index_to_gs = index; +} + +[geometry shader] +#version 150 + +layout(triangles) in; +layout(triangle_strip, max_vertices = 3) out; + +in vec4 vertex_to_gs[3]; +in vec4 color_to_gs[3]; +in int index_to_gs[3]; +out vec4 color_to_fs; + +void main() +{ + for (int i = 0; i < 3; i++) { + gl_Position = vertex_to_gs[i]; + color_to_fs = color_to_gs[index_to_gs[i]]; + EmitVertex(); + } +} + +[fragment shader] +#version 150 + +in vec4 color_to_fs; +out vec4 color; + +void main() +{ + color = color_to_fs; +} + +[vertex data] +vertex/float/2 color/float/4 index/int/1 +-1.0 -1.0 1.0 0.0 0.0 1.0 1 + 1.0 -1.0 0.0 1.0 0.0 1.0 1 + 1.0 1.0 0.0 0.0 1.0 1.0 1 +-1.0 -1.0 1.0 0.0 0.0 1.0 2 + 1.0 1.0 0.0 0.0 1.0 1.0 2 +-1.0 1.0 0.0 1.0 0.0 1.0 2 + +[test] +draw arrays GL_TRIANGLES 0 6 +probe all rgba 0.0 1.0 0.0 1.0 </blockquote> </div></div> Seeing that the index attribute has the same value for all vertices that belong to the same primitive, I can imagine some simple code generation bugs under which this test would produce a false positive. I'd like to see each index attribute have a unique value, and guarantee that the test fails if a single wrong index value is chosen. Here's a partial example that I think accomplishes that: [geometry shader] #version 150 layout(triangles) in; layout(triangle_strip, max_vertices = 3) out; in vec4 vertex_to_gs[3]; in ivec4 color_to_gs[3]; in ivec4 modulus_to_gs[3]; int index_to_gs[3]; out vec4 color_to_fs; void main() {<div class="im"> for (int i = 0; i < 3; i++) { </div> gl_Position = vertex_to_gs[i]; color_to_fs = vec4(mod(color_to_gs[i], modulus_to_gs[index_to_gs[i]])); EmitVertex(); } } [vertex data] vertex/float/2 color/int/4 modulus/int/4 index/int/1 -1.0 -1.0 5 7 9 10 4 7 5 2 2 1.0 -1.0 2 3 4 3 2 2 2 2 1 1.0 1.0 8 15 15 15 5 6 9 9 0 -1.0 -1.0 ... 1.0 1.0 ... -1.0 1.0 ... </blockquote></div> </div><div class="gmail_extra">Sure, I can go along with this. It's definitely a more thorough test, and it didn't take me too long to think through all the math and convince myself that it produces the correct results. v2 on its way. </div></div>