Texture memory, nvidia, and opengl
So I'm working in a embedded like env. I have a nVidia Geforce 4 card
with 128 MB of ram and a linux system with 128 MB of ram. Due to the
"embeded like" env, the linux system does not have a swap file system.
Which would be fine except the nVidia Open GL driver seems to keep 2
copies of each texture I load in memory. One copy in system memory
and one copy in card memory. Meaning I malloc some memory to read in
the texture file from disk, hand it to open gl, then free the memory I
malloc'ed to hold the texture. It seems as if the driver keeps one
copy of the texture in system memory and keeps one in the card memory.
I am working with OpenGL 1.3 drivers on linux from nVidia's web site.
I would like to know if anyone else has seen this problem? And is
there a way to stop this from happening.
To be completely fair, I'm about 70% confidient that this is my
problem, so my description above may be incorrect. But nVidia does
not open source their open GL drivers so without dumping assembler and
reverse engineer the driver, I think my decription above is accurate.
Any help or insite would be greatly appreciated.