floats vs. doubles

OpenGL Development JOGL Development
#1

Up until now, I have just been using doubles everywhere. Mostly, I have done this because I’m far too lazy to type a lot of "f"s after numbers. Now that I am beginning to deal with larger volumes of data going from the client to the GPU, I’ve noticed that glDrawElements() is the biggest pig at the CPU trough. So, I’ve set everything up to leverage glInterleavedArrays(); but I noticed that it only supports floats ( from what I can tell ). I can see that if throughput becomes an issue, I would want to send floats across to the GPU to reduce the amount of data. So, how do JOGLrs out there deal with this?

Work in double format on client side, then convert when you send to GPU?
Work only in float format?
Work only in double format – just use multiDrawElements(), make your data as sparse as possible, and just deal with that?

cheers

#2

The GPU is not even able to do math in 64 bits.

So it makes no sense to send it double values.

#3

ah. Thank you Riven. OK, so sounds like either work in floats, or convert to floats then use glInterleavedArrays(). I know this stuff is sort of obvious to an experienced person, so thanks for the response.

cheers