Timeline for Early Z-test / depth-test in DirectX 11
Current License: CC BY-SA 3.0
9 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jul 27, 2013 at 21:44 | comment | added | Maximus Minimus | [earlydepthstencil] - see msdn.microsoft.com/en-us/library/windows/desktop/… | |
| Jul 27, 2013 at 20:54 | comment | added | Nicol Bolas | @SeanMiddleditch: That sounds more like conservative depth, rather than an enforced depth test. It's a hint you use so that early depth can still happen even if you change the depth, because you promise not to make it smaller or larger than the original. | |
| Jul 27, 2013 at 19:43 | comment | added | Sean Middleditch |
Oh, I just found reference to a using float depth : SV_DepthGreaterEqual and float depth : SV_DepthLessEqual in nvidia.com/content/GTC-2010/pdfs/2157_GTC2010.pdf.
|
|
| Jul 27, 2013 at 19:40 | comment | added | Sean Middleditch |
@NicolBolas: I forwarded that question on to a friend on the DX team. I think the answer is "no" (I can't find any documentation to indicate a "yes"), though. I didn't know GL had such a feature, thanks for bringing it up! Also, I think you need to s/gl_Depth/gl_FragDepth/ in my previous comment.
|
|
| Jul 27, 2013 at 19:11 | comment | added | Nicol Bolas |
Here's a sub-question that's not worth an actual question. GL 4.2+ has a way of forcing early-z by using a specific fragment shader setting. This also turns off the ability to write to gl_Depth (so that the value that gets tested with the depth is the value that gets written if it passes). This is mainly for image load/store operations. Does D3D 11 have something similar? And if so, how does it work?
|
|
| Jul 27, 2013 at 18:05 | comment | added | Sean Middleditch |
Using it as an output parameter, yeahy. Again, it's a driver/hardware "hidden" optimization, so it's not really defined, but that's the most obvious and direct way for them to detect it. In OpenGL, the driver generally just checks during shader compilation if you assign to gl_Depth, which is OpenGL's equivalent to using semantics.
|
|
| Jul 27, 2013 at 17:58 | comment | added | aL3891 | So using the depth semantic is how Dx knows i modify the depth? i wonderd how that worked :) | |
| Jul 27, 2013 at 17:53 | vote | accept | aL3891 | ||
| Jul 27, 2013 at 17:31 | history | answered | Sean Middleditch | CC BY-SA 3.0 |