Looks like drivers reset because OC too high. Dial back the OC.
I second that. When overclocking settings are too high, I get "NVML error: cannot get temperature" and "NVML error: cannot get fan speed" in Claymore's log file, and the hashrate becomes null on that GPU.
chanyewhau,
Ethash algorithm is more memory-intensive than Equihash and memory instability manifests itself faster. What is the exact memory frequency - check GPU-Z or another Afterburner skin, that shows the actual frequency? Try to run the miner at stock clocks and PL 80. Sometimes it happens that cards do not hold even a 10% clock increase.
BTW, what memory size and type is installed on your GPUs?
Fans running at 100% will not shorten the life of your GPU, though the fans might fail sooner due to increased mechanical stress (mainly the bearings run dry and might jam).
The memory size is 6144 MB, memory type is Hynix.
My current OC setting is Power Limit 75%, Core Clock +50, Memory Clock +300, Fan Speed 65%.
With the current OC setting, the cards are running at GPU Clock 1566 MHz, Memory 2152 Mhz. (Based on GPU-Z)
The default clock is 1506 Mhz, Memory 2002 Mhz.
The problematic fan that runs at 100% automatically has the lowest temperature even it is running at 65% like the other cards.
So i dont think the fan switch to 100% due to overheat.