General rule of thumb on acceptable num crashes per day? GTX 1070ti DSTM

[minernew2000]

Hello All,

I have a 1 card rig running a Gigabyte 1070 ti Gaming OC edition.

Running HiveOS , I set the OC settings to +200core +1400 mem power max=115 watts as per the general recommendations I have read. This seems to give me the best sol/watt but I have not tweaked it too much to try other settings.

Been getting on average 490 sol/second at around 4.26 sol/watt, but the card does seem to crash 2x daily.

2018-04-21 10:40:25 AM|gpu_id 0 73 1 an illegal memory access was encountered
2018-04-21 10:40:25 AM|gpu 0 unresponsive - check overclocking
2018-04-21 10:40:25 AM|cudaMemcpy 1 failed

My questions for the more experienced folks: Is it OK to accept with the associated 40 seconds downtime between crash and miner restart 2 times per day?

Do you guys run OCed to get better hash and efficiency and accept that cards will go up and down or do you accept worse hash rates/efficiency to keep the cards up at all times?

Can these sorts of crashes damage the cards themselves?

Thanks

[minernew2000]

Sorry forgot to mention that the card is running at 60 degrees temp or less with the fans at 30% - so it is not an overheating issue.

Thanks.

[Tidsdilatation]

Ofcourse these crashes could hurt the GPU in the long run. You should never accept any amount of crashes. You should lower your overclocking and fine a nice stable amount that you can run 24/7. You are losing more money when you restart the crashes then you would lowering your overclock.

[minernew2000]

OK, thanks for the info.  I figured that would be the expert response.

I have since tried lowering the memory oc to stock and leaving the core clock to +200.  No crashes yet and no change in hash rate.  Seems like the memory oc does not affect the dstm hash rate.

I wonder if the #s are a bit off for me as this is the version of the card that came factory oced from gigabyte.

Thanks.