Bitcoin Forum

Hi there!

So, I got 6x R9 290 spread over 2 rigs, and for a while now, when using x-11 miner (sph, drk miner or even the new one from lasybear) I have noticed that 1 random card always have those weird fluctuations more than others and as such, hashes slower. Usually the fluctuation on that specific card just get worse in time and as such, the hash drops even further.
Now, when I say random card, I really mean it, sometimes its card number 1, sometimes number 2 and sometimes number 3. Mostly 3 though and it shows like this:

 GPU 0:  71.0C  72%    | 3.662M/3.638Mh/s | R:  3.0% HW:0 WU:0.054/m I:19
 GPU 1:  65.0C  66%    | 3.613M/3.628Mh/s | R:  1.8% HW:0 WU:0.052/m I:19
 GPU 2:  66.0C  67%    | 3.222M/3.284Mh/s | R:  2.9% HW:0 WU:0.044/m I:19

As you can see, card #3 hashes at 400kh/s less than the other 2.

Here's a screen shot of the fluctuations I'm talking about:
https://i.imgur.com/k4Flkn0.png

I tried different configurations and I would try changing some hardware, but what I didn't mention is that both rigs are completely different with the exception of the cards and the PSU (1000w Corsair).
1 rig is made out 8gb ddr2 ram, asus motherboard and amd cpu while the other rig is made out of a 8gb ddr3, gigabyte motherboard and intel cpu.

*Note: When mining Scrypt or N-factor coins there's no issue whatsoever.
Oh and I forgot to mention that while using girinos miner (the 2% fee one) than half the times I manage to get the cards to run just fine while in the other half it's similar to what I wrote.
** I've tried powertune +20,+40,+50 but haven't seen any change.

Any ideas? Been at it for some time now...

bump

I can confirm I am experiencing the exact same issues of hash degradation over a period of 24+ hours with the x13mod kernel.

Memory spikes from 150-1500 every couple of seconds.  Perhaps this was what was throwing cards into SICK status, I wonder if x11mod experiences the same issue.

Some particular function call is not properly handing itself off to the next function in this optimized kernel for x11/x13.

We need to see if we can isolate where and why the memory clock has the ability to jump down 90% then back up to OC clock...

I've heard folks are dialing back their memclock from 1500 to 1200-1300.  I currently set mine back to 1300 because I seemed to be getting glitches and sick cards.  don't know if it will help but I'm watching it.

If you'd look at the image you'd see my memclock is at 1250.

Anyway, I already solved this issue.

It was powertune not getting enabled when I was trying to use it; I managed to fix it by enabling 'Unofficial overclocking mode without PowerPlay support' in MSI AB.

Your image was flagged as invalid on my end so I couldn't see it.  Glad you got it fingered.

As i understand it's was tuning issue? Just to be clear.

Powertune issue.

Is there a solution for BAMT 1.6.2?? I'm having this issue as well, on four cards, but only one stays down at 150 and the card dies.