Running linux-based opengl application remotely on OS X

Posted on

QUESTION :

I’d like to run some OpenGL-based visualization tools on a relatively beefy Linux-based (currently Ubuntu, but this could be changed) machine and interact with the application on a much less capable OS X laptop (MacBook Air, to be purchased). What is the best way to do this, if it is even possible, and what kind of performance should I expect, given a good Wi-Fi connection?

Update: As I mentioned in a comment, windows can be relatively small, say 1024×768, and I can live without color if necessary. The 3-D model in most cases will be more than 50% background color, so compression should be advantageous. A low frame-rate is not a problem, but shouldn’t be so slow as to be frustrating. I’m guessing 10 f/s is adequate.

ANSWER :

Doing this will require incredibly high bandwidth to get any decent quality frame rates. Modest resolution with a decent frame rate will require in the order of 5Gb/s. That means just to start off, you’ll need 10Gb interfaces on both devices. But it’s likely that you’ll run into any number of other bottlenecks on one system or the other that will severely degrade performance. The video card bus and memory are highly optimized for extremely high bandwidth and frame rates. This optimization is not likely to survive traversing the system bus, CPU, RAM, network card, wire and then all of that again on the other side.

All of that isn’t to say don’t try it. I’m very curious exactly how it would perform 🙂

On the Ubuntu server you could set up NoMachine server as described here and download the respective client from here. IMO NoMachine is highly efficient and will do the job.

Leave a Reply

Your email address will not be published. Required fields are marked *