Re: [PATCH v2 0/1] MR2657: d3d9/tests: test table fog z vs rhw with shaders, depth bias, oDepth..

[Stefan Dösinger (＠stefan)]

In the test that shows the clear color the geometry ends up on the wrong side of the near clipping plane (negative Z after the perspective division). WARP differs from real drivers here - see zenable_test(), line 18770.

So I think you could get WARP behave like all other drivers if you set D3DRS_CLIPPING=TRUE in non-XYZRHW cases, but I don't think testing clipped-away geometry is what you intended.