3 Cards burned after 14hrs?

[ivakar]

I never raise the temperature of my GPU above 58 degrees. 70 degrees is the limit for GPU. But I think that is not the case. Did you have any problems with the power supply. If the power supply also worked to the limit could cause a power surge. This is enough to fail a large part of your equipment.

you are wrong, the limit set to 82-84C depending on the model
once card reaches that temperature it starts to trottle and reduce the core clock trying to shut the temp off.
of course you could still to bur the card... but it is more difficult now

[RoyaLunique]

O.c so hig

[QuintLeo]

On both the Zotac Mini 1070 ti and my EVGA SC 1070 ti cards, I use a power setting of about 105 watts (I vary this up to 5 watts either side to hit my "target total machine power consumption", with very small effect on efficiency), +200 core +700 memory using nvidia-settings.
 At 105 watts, all of my 1070 ti cards hit right around 465 sol/s for 4.4ish efficiency.
 50% fan setting in a room that is 80F was putting the card temps in the low 50s (EVGA) or mid 50s (Zotac) as long as I wasn't putting the card air intake right next to the CPU on the motherboard.

 I'm not sure how nvOS settings compare to nvidia-settings though, I've NEVER seen a card that worked at all with +1000 in nvidiasettings on the memory clock.

 The memory speed DOES make a noticeable difference in hashrate.

https://bitcointalk.org/index.php?topic=2324651.msg23996225#msg23996225 for my initial testing under Windows - I never did figure out why the "60% TDP" setting in Afterburner was only giving me 103 watts though (it should have been 106).

[Cereberus]

If the USB is a Lexar I think that might be the fault. I bought a Lexar USB 64 GB 3.0 for my work and it burned out within 2 days, I only did a few copies and never put that USB outside work computers which are all protected from viruses. It used to just pulse and showing nothing on screen. I think a cheap SSD 60 GB can solve your problem. I don't see any Linux OS causing trouble, they are well known for running years without a restart and not restarting all the time so something must be wrong with that USB drive of yours.

And lower the memory clock to +500 max and core clock +100 max. Pushing your cards too much in Windows causes nvidiaklm.sys error or Thread Stuck in Device Driver in AMD cards, don' know about Linux though if this maybe a reason.

[QuintLeo]

You are pushing those cards too hard with the overclocks, most cards don't run with clocks too high.

I get over 510 sols on my EVGA 1070 Ti with these settings below with the temp at 70c using the zm miner.

Set the memory to -303 and the core to +140 and power to 85 - 90.

Lower your overclocks and keep the fans above 90% to keep them cool.

 I was seeing 509 sols/s at 70% power with +200 core +700 memory on my EVGA SC 1070 ti cards using EBWF - I think DSTM was a hair higher at that setting but can't remember for sure.

 They also run cooler than at 85-90% power with lower fan settings - 56C at 52% fan when I was testing that setting in an 80F ambient room with my standard Afterburner fan profile.

 I don't have any of the MSI blower models, but I do have a pair of ASUS (one 1080 ti one 1080) that are a very similar design - they DO run quite a bit hotter than any of my "fan-type" designs, but it helps a lot to remove the mounting bracket for better airflow.

[STSMiner]

DSTM makes the cards run hotter, the EWBF miner works best for lower heat.

[QuintLeo]

15-20% less power lowers heat a LOT more than changing miner software.

 8-P