I agree that if a texture was created as GL_TEXTURE_RECTANGLE, meaning that all shaders use the TGSI_TEXTURE_RECT sampler type, it makes sense to use unnormalized coordinates. However if a texture is NPOT as in OpenGL, meaning that all shaders use the TGSI_TEXTURE_2D sampler type, we shouldn't force unnormalized texture coordinates everywhere and trade effectivity and simplicity of one driver (nvfx/nv30) at the expense of another (r300). Also I don't like changing pipe_resource::flags after a texture is created. Some time ago I think Brian suggested to add PIPE_TEXTURE_RECT as another texture type and make it match the meaning of GL_TEXTURE_RECTANGLE. We would do this change but we had no time back then. This seems to be the solution that would work best for everybody. -Marek <div class="gmail_quote">On Wed, Aug 11, 2010 at 6:27 PM, Luca Barbieri <<a href="mailto:luca@luca-barbieri.com" target="_blank">luca@luca-barbieri.com</a>> wrote: <blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;"> <div>> I am not sure if the following is legal, but I think we will just mask this > unnormalized flag out in resource_create, because r300 doesn't like (read: > doesn't support in hardware) unnormalized texture coordinates, and switching > shaders just because of the flag is silly. </div>Hmm I thought it support GL_TEXTURE_RECTANGLE natively. This is making me think that my patchset is wrong in having blitter/blit set the flag themselves on NPOT textures. It might even be a good idea to split it in two flags, one for state tracker->driver communication, and another for the driver->state tracker. This way, r300g won't have to do anything. Not sure on how to call the flags though, since the current name is already too long. </blockquote></div>