How to run the GPU in parallel with the CPU with the parallel computing toolbox?

I want to split the processing of a large dataset between the GPU and CPU. I tried the spmd command but the function running on the GPU crashes even though it runs fine outside of the spmd block.

My code looks like this:

 spmd
    if labindex~=12
       A=fun_CPU(x(labindex)); % run on CPU
    else
       A=fun_GPU(x(labindex)); % run on GPU
    end
 end

where fun_CPU and fun_GPU are m-file functions and fun_GPU is:

 function y=fun_GPU(x)
 % GPU processing
 y=gather(y_GPU);

The error I get states: "spmd : The client lost connection to lab 1. This might be due to network problems, or the interactive matlabpool job might have errored. This is causing: java.io.IOException: An existing connection was forcibly closed by the remote host"

Is there a problem using gather to retrieve data from the GPU inside the spmd block? Or could this be hardware related ?

I am using a demo version of the parallel processing tb with matlab R2011b. I have a Tesla C1060 with driver v4.1.