How much extra load on the video card is there when using the geometry
shader to generate geometry as aposed to just sending a static lot of data
into it.
And is there the 1,000,000 or 2,000,000 max triangle in one batch still
applying?
2
The new geometry shader.
Started by rouncer, Mar 11 2008 02:13 PM
3 replies to this topic
#3
-
- Administrators
-
- 4979 posts
DevMaster Staff
- LocationBellevue, WA
Posted 13 March 2008 - 07:12 PM
I don't think anyone hates you, there just may not be anyone here who knows that much about geometry shaders. I know I don't know anything about them...
reedbeta.com - developer blog, OpenGL demos, and other projects
#4
-
- Members
-
- 911 posts
Senior Member
Posted 13 March 2008 - 11:26 PM
IIRC the geometry shader is only allowed to output 1024 floats per triangle.
In terms of load, I would try to consider the savings you could make. One of the Ruby demos from ATI (Video, Terrain only) gained a 6X speedup and memory savings of 44MB on the terrain rendering, due to using the GPU for tesselating data. The paper has a lot more detail.
In terms of load, I would try to consider the savings you could make. One of the Ruby demos from ATI (Video, Terrain only) gained a 6X speedup and memory savings of 44MB on the terrain rendering, due to using the GPU for tesselating data. The paper has a lot more detail.
"Stupid bug! You go squish now!!" - Homer Simpson
1 user(s) are reading this topic
0 members, 1 guests, 0 anonymous users
