I have a few questions about using 4 monitors on Linux (Redhat 5+) with nvidia cards. We are using Qt as our windowing system.
This is not so much games related but it is about real-time rendering and there seems to be good community knowledge on the subject here.
The issue we're facing is that we want to be able to have multiple processes running on the same system that can all potentially make use of the 4+ monitors available. We would like to determine the easiest route to doing this on RHEL5+ with opengl. Our current research has led us to the conclusion that our best option is to utilise nvidia cards for various reasons, so the question will make reference to their systems, but it is not set in stone that we only support nvidia.
The questions are assuming that we will either have two seperate cards or two cards that are connected via sli to get our 4 monitor support. It looks like the nvidia gtx 690 type cards will have 4 outputs on a single unit, but we see their usage as less likely. However, if you think this detail is important in your answer then we certainly appreciate any information.
Are there any issues with having more than 2 monitors on a RHEL5 system?
In the nvidia x configuration it talks about twinview. If there are more than 2 monitors, will I still be able to configure this through the same interface and does this depend on whether the cards are connected with sli or not?
Is this a simple case of creating a window on a specified monitor and rendering as usual? If yes, is spanning an issue, can one application use multiple windows, and what about making the program go fullscreen on a monitor or even multiple monitors.
Will my application even need to be aware that there are multiple cards, sli-ed or not, or will the driver simply take good care of me.
If using opengl, can I render to each monitor separately from one application. i.e. treat them as separate render areas, drawing separate things?
If the two cards are connected via sli, is there any change to the opengl code or will this be handled by the driver?
If using sli and rendering the same data to multiple monitors will there only be one copy of the data in vram, or will there be separate copies on each card? i.e. is the memory between the sli-ed cards shared in any way?
If the memory is not shared, do I need to pass buffers twice or can the driver just manage that for me?
Can both sli-ed cards draw to a monitor connected to the other card when doing standard desktop apps or will the jobs simply be delegated to the card that the monitor is connected to.