Re: wined3d: Reserve two varyings for ATI DX10 cards
2009/6/15 Robert Key <hpestilence(a)gmail.com>:
Changelog: * This reserves 2 varyings for gl_FrontColor and gl_FrontSecondaryColor for ATI DX10 cards. This should fix bug #17437 once the new fglrx drivers get released this week.
That doesn't look quite right. gl_FrontColor and gl_FrontSecondaryColor should only be used as a fallback when we're already out of regular varyings to use. What does the shader that triggers this look like?
On Monday 15 June 2009 14:01:33 Henri Verbeet wrote:
2009/6/15 Robert Key <hpestilence(a)gmail.com>:
Changelog: * This reserves 2 varyings for gl_FrontColor and gl_FrontSecondaryColor for ATI DX10 cards. This should fix bug #17437 once the new fglrx drivers get released this week.
That doesn't look quite right. gl_FrontColor and gl_FrontSecondaryColor should only be used as a fallback when we're already out of regular varyings to use. What does the shader that triggers this look like?
This is what it looks like fixed, before it was doing IN[15] + gl_FrontColor + gl_FrontSecondaryColor.
2009/6/15 Robert Key <hpestilence(a)gmail.com>:
This is what it looks like fixed, before it was doing IN[15] + gl_FrontColor + gl_FrontSecondaryColor.
We should probably just avoid writing to gl_FrontColor / gl_FrontSecondaryColor if they're not read. How does the attached patch work?
On Monday 15 June 2009 14:39:48 Henri Verbeet wrote:
2009/6/15 Robert Key <hpestilence(a)gmail.com>:
This is what it looks like fixed, before it was doing IN[15] + gl_FrontColor + gl_FrontSecondaryColor.
We should probably just avoid writing to gl_FrontColor / gl_FrontSecondaryColor if they're not read. How does the attached patch work? It worked perfect, everything rendered with eve-online and catalyst 9.6.
participants (2)
-
Henri Verbeet -
Robert Key