Any solution that involves capturing the screen and streaming it through network is bound to be processor-intensive. However, if your hardware is powerful enough, you could get satisfactory performance.
The company I work for has a screen capture video source and a multimedia toolkit with C++ demos for capturing, streaming and playback.
If you would like to try, download the free evaluation edition from our website and check out the C++ demos for the Capture module and the streaming module.
The evaluation is fully-functional (but time-limited) and comes with unlimited free technical support through email and chat. This means if you need help modifying the demos to do what you want, send us the full details and our engineers will work with you to see if it’s possible to do and how.
Update in response to comment:
I experimented a bit on a test PC here (Core i7-870, about 8 years old) and tried both uncompressed and compressed full-screen capture (1680x1050 resolution) at 2 capture rates (15 and 30 frames per second).
Here are the results I got:

In some cases, using compression increased CPU usage, but in others, it actually reduced it, probably because the smaller data size reduced copying load off of the CPU.
Also, using compression almost always improves effective frame rate.
Other compressors, such as H264, might require more processing power, but again they can make use of the GPU.
I expect you will get better results if you use better hardware, that’s why I recommended trying the free evaluation in your actual environment.