Enough.
Srsly.
What we can do, will be done. If there is some issue with how the shaders are written that can be addressed, it will be. But do not expect miracles. Just because the drivers say the hardware can do something does not mean it will do it well.
On the issue of drivers: Yes, hardware wise, ATI makes some damn nice cards. I still own plenty as far back an the Original 32mb DDR AGP card. I still have a 3D Rage card. I still have an Xpert@Play card.
But thier drivers (and nVidia is just as at fault for this as well) are **** when it comes to the new supporting the old like it should because they want the "One driver to rule them all" approach, and it just. does. not. work.
When a completely new core line up comes out a new driver package for that cores features and cards needs to be made. The older ones can retain support and you can be sure that if/when they get updated, it will be to fix or improve something for that core/lineup. There is no guarantee of that today.
The newest drivers? DX10.1, Shader Model 4.1, OpenGL 2.1. Now, OpenGL 2.1 supports 2.0, and that is good. But the shader parser is still going to push all that it can. And it is running on a DX 9.0c, Shader Model 3.0 (first generation card to support SM3.0 and the 1900 replaced a buggy 1800 flagship card) OpenGL 2.0 card. Something in there is not necessarily going to be good, from my experience. Additionally, the 1950 is the first ATI card to support Internal crossfire. So, with revamps from the flagship of the series (and the 1950 being a complete rebuild from the ground up), it doesn't surprise me that there is _some_ sort of issue.
What is most frustrating is the irrational turns of how the issue decides it is going to exhibit itself.
Once I get the cards' I'll know more. And while we will have a patch, and while the patch will include fixed (for completely different reasons) shader files, I do not promise that they _will_ do anything for this issue. I can only hope that they will.