Re: X11 forwarding issue

Hi Pekka,
If I get it right, what you say is that NVIDA might then be able of
forwarding the buffers (in version 2.0 as Michael said) without caring
about the GLX library?
I don't agree that transmitting video is always more efficient. Consider
our case: our data is made of 80% static maps (vertex buffers that almost
never change) and 20% texture buffers that only change a couple of times in
a minute. At least theoretically, it would be for us far more efficient to
transmit the buffers only once through the network (or at very low refresh
rates) and then only use their 32 bit identifiers to issue a drawing call.
For the zooming and panning refresh calls (where almost no data needs to be
updated) the gains should be significant. And indeed, while using the
legacy mode we achieved an excellent bandwidth consumption using the GL
lists, it was an amazing improvement compared to X11 forwarding using Qt or
Cairo calls.
I get the feeling that only a few people are interested in efficiently
exporting through the network nowadays, so the development is going in a
different direction. Also, the bandwidth has become less critical for most
people, so probably no one care. However, there are still some areas where
keeping a quite network is considered very important.

_______________________________________________
mesa-users mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/mesa-users