Hi people, I''ve got a NVidia GeForce and an ATI Rage Pro. My D3DXComputeBoundingBox code works correctly with the ATI card but not with the NVIDIA card (getting wrong values of the mesh). Is this ComputeBoundingBox function of DX driver specific?? Did anyone got the function work under NVidia if the function is driver specific?.... Bye, Alex

There shouldn''t be ANY difference between how that function behaves on different gfx hardware. Gfx hardware is "one way", it can''t be used to be compute arbitrary things like bounding boxes so there is no way D3DX is using it internally.

Are the values really far out on the nVidia chip or just a small amount ?

Two thoughts:

1. I''m assuming you''ve been testing on two different machines. With different processors ?. There may be differences between values returned AMD and Intel chips. There are also some minor bugs in the D3DX PSGP for AMD CPUs. These are fixed IIRC in the 8.1b runtime.

2. Where are the source vertices you''re passing into the function coming from. If the pointer you are passing is to a locked vertex buffer, then that is likely the reason (remember T&L gfx hardware is one way, reading back can be flaky!). Particularly if you haven''t locked in the correct way or have set the buffer up as write only.


--
Simon O''Connor
Creative Asylum Ltd
www.creative-asylum.com