Add GraphicsSystemValues.test for graphics-stage system-value semantics#1142
Add GraphicsSystemValues.test for graphics-stage system-value semantics#1142EmilioLaiso wants to merge 2 commits into
GraphicsSystemValues.test for graphics-stage system-value semantics#1142Conversation
|
I like the things we are testing, I am just wondering if it is better from a compiler developer's point of view if we have a test for each one of these or have more elements mixed in one test like we did here. |
Considering |
EmilioLaiso
force-pushed
the
graphics-semantics
branch
from
b4fa460 to
425d666
Compare
| # This test exercises: | ||
| # * SV_VertexID - VS input, forwarded via a `nointerpolation` varying | ||
| # * SV_PrimitiveID - PS input | ||
| # * SV_IsFrontFace - PS input | ||
| # * SV_POSITION - VS output / PS input, asserted at pixel center | ||
| # | ||
| # SV_VertexID: https://github.com/llvm/wg-hlsl/issues/151 | ||
| # SV_PrimitiveID: https://github.com/llvm/wg-hlsl/issues/160 | ||
| # SV_IsFrontFace: https://github.com/llvm/wg-hlsl/issues/162 | ||
| # SV_POSITION: https://github.com/llvm/wg-hlsl/issues/141 | ||
| # Clang's HLSL -> DXIL lowering does not yet implement these graphics-stage SVs. | ||
| # | ||
| # Also indirectly exercises the `nointerpolation` modifier, which Clang's HLSL | ||
| # frontend does not yet parse. | ||
| # https://github.com/llvm/wg-hlsl/issues/303 |
There was a problem hiding this comment.
Would it be better to put these test explanations at the top of the file so that it's the first thing you see when opening the file?
3 participants
Adds a new test that bundles every per-pixel, observable graphics-stage system-value semantic the framework can express today into one shader pass.
Modeled on the existing
ComputeSystemValues.test.Covers
SV_VertexID,SV_PrimitiveID,SV_IsFrontFace, andSV_POSITIONpass-through.