Are your GPU all Identical and same brands? I was thinking it is due to mem problem and -h issues, use -di to separate -h values to each card
i also experience gpu0 with 0 hashrate after few minutes of having like 600+ hash.
gpu 0 hangs in opencl call exit
then miner will restart again. only restarting pc will make the gpu0 mine again with 600+ hashrate and will repeat same error.
what's your GPU memory? have you tried to decrease your -h value? I often encountered this and decreasing the -h value solve it.