Bitcoin Forum

Alternate cryptocurrencies => Mining (Altcoins) => Topic started by: mensa84 on July 24, 2017, 01:04:27 PM



Title: Too much overclocked? How to determine which GPU did fail?
Post by: mensa84 on July 24, 2017, 01:04:27 PM
Hi,

in one of my rigs I have 1060s from different manufacturers, but all with 6GB and Samsung memory.
Now I am trying to find the correct overclocking settings for MSI Afterburner.
With following settings for all GPUs ethminer sometimes fails:

Power Limit: 65%
Core Clock: +120
Memory Clock: +750

I know from prior tests, that some of the cards can run at memory +950 without problems, so I think now, that only one or 2 specific GPUs in the rig can't run stable using the +750 setting.
Is there any way to find out which GPUs did fail, when ethminer stop working?

Then it would be easier to lower the settings for that specific cards and raise for the other, stronger cards more.


Title: Re: Too much overclocked? How to determine which GPU did fail?
Post by: dadesu on July 24, 2017, 01:11:07 PM
You can't on begining of test rise all parameters.
First go with memory, put it on +650mhz, core on 0 and power can be 65%, try like that for few hours,
and if is stable continue to rise memory but slowly to find correct size for that cards.
Then you can go on core calibration, but will be good to have instrument for mesuring consumption.


Title: Re: Too much overclocked? How to determine which GPU did fail?
Post by: Ayers on July 24, 2017, 01:42:18 PM
Hi,

in one of my rigs I have 1060s from different manufacturers, but all with 6GB and Samsung memory.
Now I am trying to find the correct overclocking settings for MSI Afterburner.
With following settings for all GPUs ethminer sometimes fails:

Power Limit: 65%
Core Clock: +120
Memory Clock: +750

I know from prior tests, that some of the cards can run at memory +950 without problems, so I think now, that only one or 2 specific GPUs in the rig can't run stable using the +750 setting.
Is there any way to find out which GPUs did fail, when ethminer stop working?

Then it would be easier to lower the settings for that specific cards and raise for the other, stronger cards more.

if the miner crash there is no way because all the gpu shut down, but if you can test one 1 by one you can find it easily, it should crash very quickly if it can't handle that speed


Title: Re: Too much overclocked? How to determine which GPU did fail?
Post by: mensa84 on July 24, 2017, 01:57:08 PM
Ok, thanks for that hint.
I will test one by one card now using option --cuda-devies in ethminer.
Do you know if the order of the cards 0,1,2,3,4,5,6,7 is the same in ethminer, in MSI Afterburner and also in GPU-Z?
Otherwise I don't know which card I am currently testing, or?