We are in need of a relatively high-end OpenGL-accelerating
board, and knowing relatively little about what's available,
we thought we ask for a little help. Any recommendations?
Our application is probably pretty bizarre compared to most:
every graphics primitive (vertices, triangles, and texture
maps) has a lifetime of exactly one rendered frame, so
the ability to stream and dispose of textures is critical
to good performance. The textures are large, approaching
textures, we could use that very well -- are there any
out there that do?
We also are looking at a need to read the rendered frame
out of the framebuffer as quickly as possible.
Any pointers to information, as well as recommendations,
would be greatly appreciated!