Quote from: ccccccc7 on January 23, 2017, 04:41:27 PM

Quote from: malandante on January 23, 2017, 03:32:24 PM

Do you think these spikes can be the reason why my rig freezes? I will describe my problem a little.

I have a 7970, a 280x and two fury nitros. During the last couple of weeks I have been trying to run my system stable with no luck. Freezes can happen at any time after starting to mine. Sometimes minutes, sometimes hours.

Rig specs: Asrock 970 extreme4. Sempron 145. 8GB ram. EVGA 1300 G2. Sapphire 7970 ghz edition, sapphire 280x vapor-x, 3x sapphire fury nitro OC. Tahiti cards undervolted on bios, fiji cards on msi afterburner. No OC on furys. Win 10. 12.6.3. CM 11.1.

First I changed all risers to v6 ones.

Next I upgraded the processor to athlon x2 because I noticed that I was getting less hashes when MSI afterburner was on and I thought the system would be more stable.

After a couple of days more I changed the mobo to asrock h81 pro btc with celeron 1840.

Still no luck.

I thought that the power was insufficient so I added a corsair rx1000 linked with an add2psu device.

No luck, freezes persist.

Changed EVGA to corsair hx1200i. This psu is powering mobo, raisers and tahiti cards.

Nothing changed.

I stopped undervolting. Still same freezes.

One fury died the day before yesterday. No fans, not detected in win 10, not detected on ubuntu, mobo not booting when the card is alone.

I changed risers again to the furys.

I used separate cables for each riser.

I powered the offending fury card with one cable for each connector. Still the same.

I tried catalyst 15.12. Same effect.

Sometimes I see 511 degrees in the fury card on CM log prior to freeze, sometimes it just stops hashing without any error and the watchdog doesn't restart the miner, sometimes all 4 cards go to 0.000 hashing without any error, sometimes after gpu reported incorrect temperature 511°C.

I'm already out of ideas.

Any hint, anyone?

Edit: typo

I have the exact same issue as you describe.

2 x 7990 & 2 x 7950
Enermax 1500w Maxrevo
Celeron cpu
Asrock Z97 Extreme 4

The worst part is it cuts out randomly, sometimes it can be 10mins and others 12hours.

When I have seen the cut outs it happens exactly the same as if the PSU cuts from being over loaded.

I had also observed the power spikes in GPUZ and it normally happens when a share is accepted.

I had guessed that it might cut out, say, in the rare instance that 5 or 6+ shares are found simultaneously which causes a huge spike cutting the PSU which would explain why it cuts after an arbitrary amount of time.

I have no watt meter, however, I assume my cards are running the PSU near to the limit but but I can't check.

I have never really got to the bottom of the problem, however, it seems running Afterburner -20 power limit core at 1050mv leaves my system stable indefinitely.

I thought one cable is insufficient for one card with power spikes. For some reason it could be that the psu is unable to deliver the necessary power if the 12v line is weak. I mean, it could be that from the six or eight pcie connectors on the psu, some can deliver less current?

When I wrote my post, I had started to try with two cables on one card and I thought it crashed because I was at work and the pool made me thought the miner crashed. But my wife checked it and it was still running. I'll post if that is the solution.

If it still fails, I'll boot on linux and test more.