CUDA deployment on SERVER for Clients..

Helloo Guys…
M a kinda newbie in GPU computing… i bought two gtx 980 GPUs. Now i want to deploy them on my server so that multiple clients can access their computation process to compile parrallel codes SIMULTANEOUSLY i browsed much but could not found any accurate way to deploy the GPU on server.can any one help me out to deploy it in optimize way. i.e installing CUDA toolkit on server. or using any 3rd Party software or wrapper… ???

For that I would suggest following the getting started guide appropriate for your OS: