How to Render With OpenGL on Tesla equiped Windows based host

I used to think that Tesla will not support OpenGL API, but recently leanred that Tesla products also can be used on visualization via OpenGL.
I have a workstation, in which there are 2 Intel E5 CPUs, and 1 Tesla C2050. According to, Tesla C2050 should support at least OpenGL version 3.
Now, I’d like to run a render service program using OpenGL 3.3 on that workstation, but without success.
The following is what I tried.
If I login through RDP remote desktop, the supported OpenGL version is 1.1 due to the virtual graphics adapter. Here, I used tscon commond to reconnect to the pysical console. As a result, the RDP connection lost. When I reconnected, I saw all the windows resized to 800*600 and the detected OpenGL support was still 1.1.
If I login with a monitor pluged to some kind of “integrated graphics adapter”, the supported OpenGL version is still 1.1, maybe because the program was started within the screen pluged to the basic adapter. BUt the Tesla GPU does not have a grpahics output port.
I wonder how should I config the host to enable the use of Tesla GPU for OpenGL based rendering.