GPU Clock Instability

Has anyone ever have their GPU automatically "downclock" and get stuck on them before while mining? I'm on CGMiner 3.1.0 / Cat 12.8 / SDK 2.8. The rig in question consist of 3x 6870 and 1 x 7950 and is configured with a single .bat file for LTC mining. The 6870s all have identical settings while the 7950 has it's own. What happens is from random points during mining.. one of the GPU's hashrate will start to drop all the way down to <50mhs and get stuck there.

When I look into CCC, the clock/mem settings are @ idle levels and only way to remedy this is to force a restart every so often. Even if I try to overclock the GPUs with Sapphire Trixx or MSI AB it doesn't register until a restart. The operating temps are =<78c for all 6870s while the 7950 runs @ 70-73c. I've never had this issue before and nothing on the mining rig changed in terms of updates/drivers that would warrant this. Any ideas guys?