[Mesa-dev] Request for sponsered development...

Peter Robinson pbrobinson at gmail.com
Tue Mar 8 17:23:57 PST 2011

On Tue, Mar 8, 2011 at 3:55 PM, Jason D. Clinton <me at jasonclinton.com> wrote:
> On Mon, Mar 7, 2011 at 08:56, Christian Weiß <christian.weisz at spoc.at>
> wrote:
>> Consider an installation of about ~600 low budget thin-clients (with
>> almost no 3D support from the graphics chip) running as X-Terminals. Those
>> thin-client stations are serviced by a host computer for 25-30 stations
>> each. This infrastructure should be the basis of an architectural/interior
>> planning system with serious demands in terms of 3D rendering. It's all
>> clear that the client hardware will not be able to provide the power, so it
>> comes down to a server-based rendering approach.
> While your willingness to fund this kind of development is laudable, this
> sounds rather like an organization doing LTSP and grappling with the
> widespread proliferation of compositing window managers. If that's the case,
> you should probably reevaluate the current state of thin client hardware.
> Four years ago, when I last did this work, thin clients with 512MB of RAM
> and an 945 Intel graphics chip were widely available for under $300. I'm
> sure that things have improved. Any modern Intel graphics chipset >965 is
> more than enough to run a compositor and 512 MB is plenty of RAM for a
> desktop shell. This is essentially what Dave Richards at City of Largo is
> doing.
> The best way to run things in this configuration is to make the thin clients
> root from NFS root or run a Live-CD like image on them and then run a script
> in rc.local which drops your own customized .desktop files in
> ~/.local/share/applications in the Live user's home directory. These
> .desktop files Exec= lines are SSH w/ remote X11 forwarding back to your
> application servers. For example: Exec="ssh $username at libreoffice-server -Y
> /usr/bin/oowriter".
> It does require a little bit of initial scripting and figuring out how to
> handle user accounts is specific to your environment but it puts all that
> computing power on the thin clients, including their 3D hardware, to use.
> When I last did this, I used ~/.local/share/autostart shell script to launch
> a username and password prompt after the desktop was up to mount directories
> and make the SSH w/ username links proper.
> The other way to go about it is to get thin clients with 2GB of RAM and no
> harddisk and just boot NFS root and run them as full-blown disk-less
> workstations. Then all you need is a big file server. And less bandwidth
> than LTSP requires (which is a huge plus.)

The biggest issue I see in that is that most thin style corp
deployments now days are moving towards (or at least looking closely
at) a VDI style deployment and most of those techs don't support 3D


More information about the mesa-dev mailing list