GTX 1060 3GB

[storm5510]

Would someone be so kind as to look at the attached and tell me what this issue is. This is on Windows 7 Pro x64.

[kriesel]

Hi,

I've never yet seen that on my GTX1060 3GB, or other gpus. Perhaps there's a clue in the event log.

Maybe something at https://www.lifewire.com/43-errors-explained-2619238 would help.
Or at http://www.tomshardware.com/answers/...card-code.html

Good luck...

[storm5510]

I downloaded and installed a BIOS update. No changes. I am afraid I may have something here that I cannot use.

[potonono]

I notice that's a fairly recent driver date. Did you just update the driver recently, or allow it to automatically update? Have you tried uninstalling and reinstalling, or uninstalling and installing an older driver version?

[Mark Rose]

There's a known issues with GTX 1060 and the latest drivers on Windows. I'd downgrade until they fix it.

[kriesel]

This one seems to work

[mackerel]

397.31 has known problems. They released a hotfix 397.55 which is reported to solve those problems, but I found both of these give problems for gaming. I haven't crunched on them yet. Reverting to the next earlier release seems to be the solution many are going for.

[storm5510]

A bug in 397.31. I had some really strange behavior using this one. Just as a test, I went to NVidia's web site and downloaded the oldest one I could find. In the low 380's if memory serves. The card would function, barely. 102 GHz-d/day running mfaktc. I have not tried the one kriesel uses, but will do so.

Thank you all for you comments.

[storm5510]

Now, I know what the problem is, and I am going to have somebody's liver for breakfast.

[kladner]

I know there are differences between 3GB and 6GB cards as to CUDA cores, but this is obscene. Do you have access to said liver?

[kriesel]

Originally Posted by storm5510

A bug in 397.31. I had some really strange behavior using this one. Just as a test, I went to NVidia's web site and downloaded the oldest one I could find. In the low 380's if memory serves. The card would function, barely. 102 GHz-d/day running mfaktc. I have not tried the one kriesel uses, but will do so.

Thank you all for you comments.

Wow, 102 is comparable to an RX550. More normal would be around 400-500 in TF for a GTX1060. Higher Ghzd/day at lower p.
http://www.mersenne.ca/mfaktc.php

On my GTX1060 3GB:

Starting trial factoring M88007021 from 2^75 to 2^76 (86.95 GHz-days)  
k_min =  214635897418320  k_max =  429271794837336 
Using GPU kernel "barrett76_mul32_gs" 
Date    Time | class   Pct |   time     ETA | GHz-d/day    Sieve     Wait 
Mar 19 13:26 |    0   1.0% | 161.09   4h15m |    485.77    82485    n.a.% 
Mar 19 13:29 |    3   2.1% | 161.08   4h12m |    485.79    82485    n.a.%

I've seen the same RX550 have more than 2:1 variation in throughput depending on exponent and bit level. (~90-211 Ghzd/day)