running a graphical application over ssh - where is rendering done?

[Steve O'Hara-Smith]

On Thu, 20 Aug 2015 10:07:18 +0100 (BST)
Anton Shterenlikht <mexas at bris.ac.uk> wrote:

> If I conect from my laptop to a server via ssh -XY,
> and then launch a 3D graphical application on the server,
> and view the results back on my laptop, which side
> is doing the rendering?

	With X the rendering is always done by the X server (the thing
with the screen, keyboard and pointer) what goes over the network are
commands to the server (this can include bitmaps to be placed in windows).

> So does the graphical application send the data over the
> network back to my laptop, where I render it using the local resources?

	The graphical application communicates with the display server
using X11 protocol (and whatever extensions are available which includes
GLX and DRI on most systems these days) the rendering is done by the display
server. It doesn't have to be, the application could present a bitmap and
update it leaving the server to place the bitmap into a window. A high
resolution, 24 bit colour animation can eat a lot of network bandwidth done
this way, and network latency can really kill interactive behaviour.

> Is there any control over where the rendering is actually done?

	You could disable the GLX extension and see if the application has
some other way of getting the data across - it is likely to be painful
unless you have a very fast network, and the chances are that the
application will simply fail.

-- 
Steve O'Hara-Smith <steve at sohara.org>