triggering gc based on gpu memory

[denizyuret]

Is there any progress on making gc sensitive to remaining gpu memory? The following example still fails with an out-of-memory error. It works if you uncomment the manual gc() line.

using CUDArt
for i=1:1000
    a=CudaArray(Float64,1000000)
    # i%100==0 && gc()                                                                                                     
end