Bitcoin Forum
December 09, 2016, 11:59:21 AM *
News: To be able to use the next phase of the beta forum software, please ensure that your email address is correct/functional.
 
   Home   Help Search Donate Login Register  
Pages: [1]
  Print  
Author Topic: Mining on 7970, please help!  (Read 1586 times)
splitt
Newbie
*
Offline Offline

Activity: 12


View Profile
February 16, 2012, 01:12:56 PM
 #1

Hi.
I'm trying to mine on two brand new Gigabyte 7970s, OC version.
Diablominer gives me about 30mh/s per card.
CGminer gives me about 550 mh/s per card, but one of the cards gets really hot (above 90C). The other stays around 65C.

I tried removing the OK card and using just the HOT one. Same thing - goes above 90C. I tried messing with Intensity in CGminer, it worked, but I lost about 200 mh/s.
Then I tried mining with Diablominer and the HOT card worked fine, with temperatures around 65C. But again I get only 30 mh/s.

Please help me, because I cannot post anywhere else on this forum.  Cry

Thanks very much!
Nik
1481284761
Hero Member
*
Offline Offline

Posts: 1481284761

View Profile Personal Message (Offline)

Ignore
1481284761
Reply with quote  #2

1481284761
Report to moderator
Advertised sites are not endorsed by the Bitcoin Forum. They may be unsafe, untrustworthy, or illegal in your jurisdiction. Advertise here.
matthewh3
Legendary
*
Offline Offline

Activity: 1372



View Profile WWW
February 16, 2012, 01:14:40 PM
 #2

Have you tried the 'hashkill' miner program?

splitt
Newbie
*
Offline Offline

Activity: 12


View Profile
February 16, 2012, 01:17:12 PM
 #3

Sorry, forgot to mention I'm running Windows.
I don't think there's a windows version of "hashkill".
bangra
Jr. Member
*
Offline Offline

Activity: 43



View Profile
February 16, 2012, 01:20:11 PM
 #4

What do you have controlling the fans on the card? whats the airflow like in the case?

Diablo seems to be the best miner for 79xx series cards at the moment but im not yet running any 79xx's so not sure

Have i helped? 1Gm9NMJ2qn2XU1qc4Mps9GSCGe5cdetmpv
der_meister
Full Member
***
Offline Offline

Activity: 141



View Profile
February 16, 2012, 01:22:54 PM
 #5

I had read before that phoenix miner with DiaKGCN kernel also gives good results - around 620Мhs/s @ 1050MHz.
About overheating - try to undervolt the gpu, if you can.

Moving on is a simple thing,
what it leaves behind is hard...
splitt
Newbie
*
Offline Offline

Activity: 12


View Profile
February 16, 2012, 01:25:38 PM
 #6

The fans are controlled by Catalyst Overdrive. Fan speed is 100% when mining with CGminer.
I also read that Diablominer is best for 79xx but it seems it doesn't work for me. With the two 7970s I get only 60 mh/s.
The case is open, airflow is fine, room temperature is low.
Isn't it weird that one of the cards stays cool, and the other gets hot?
jake262144
Full Member
***
Offline Offline

Activity: 210


View Profile
February 16, 2012, 02:22:17 PM
 #7

The case is open, airflow is fine, room temperature is low.
Isn't it weird that one of the cards stays cool, and the other gets hot?
Without some fans, a case with the side panel removed would imply that the airflow is nowhere near fine.
You need far more than just convection to move the air around.

Hate to break your heart but cheap PC cases suck dicks when mining comes into play.
At the very least you should point two 120mm fans at those cards, e.g. one on the side panel (the more fannage the side panel can house the better) and one blowing from the front panel or from the bottom. Also, you need as much exhaust fan-power to make sure the hot air can be removed from the case double-quick.
splitt
Newbie
*
Offline Offline

Activity: 12


View Profile
February 16, 2012, 02:35:13 PM
 #8

OK, I agree.
But airflow is not the reason for this problem.
I tried the cards one by one. And the HOT card, even with the other removed from the case, still reaches temperature around 90C.

By the way, I tried DiaKGCN kernel for Phoenix miner. It gives about the same hashrates as CGminer.
But I have the same temperature problem even with Phoenix miner. Sad((((
jake262144
Full Member
***
Offline Offline

Activity: 210


View Profile
February 16, 2012, 02:44:08 PM
 #9

Bummer.

Can you check the core voltages? The hotter card may use a lower quality GPU core requiring higher voltage to achieve the same clock as the other card.

If the voltages are exactly the same and the cards are the same make and model, the cooler may make poor contact with the GPU core as a result of manufacturing boo-boo.
Unfortunately, you'll likely void your warranty the moment you start messing with the cooler.

Who's the manufacturer of those two cards?
splitt
Newbie
*
Offline Offline

Activity: 12


View Profile
February 16, 2012, 03:02:35 PM
 #10

Both are Gigabyte R797OC.
I just bought them yesterday.

Here's what GPU-Z shows when I mine with both at the same time:

OK card
12 V  11.69
VDDC  1.029
MVDDC  1.602
 
HOT card
12 V  11.88
VDDC  1.068
MVDDC  1.602

I don't know if that's good or not.
der_meister
Full Member
***
Offline Offline

Activity: 141



View Profile
February 16, 2012, 03:24:08 PM
 #11

The hotter card has higher gpu voltage - VDDC  1.068.

Try to lower it. Usually cards can mine at lower voltage than defaults. If you play games you must get it back to 1.068, but for mining I think these cards will run ok even under 1 volt. Just set lower and lower voltage until the cards begin to crash, then go back one level up and all should be ok.

For example my 5770 defaults are 850mhz@1.125 v, but mine like a charm at 930mhz@1.075. The electricity bill is smaller and temperatures are great Smiley

Moving on is a simple thing,
what it leaves behind is hard...
jake262144
Full Member
***
Offline Offline

Activity: 210


View Profile
February 16, 2012, 03:34:47 PM
 #12

The hotter card has higher gpu voltage - VDDC  1.068.
Precisely.

OP, there's nothing you can do about the GPU core quality. You can only try to drop the voltage manually, e.g. using cgminer, afterburner, or trixx.
Try to take the vddc downto 1.050V, then if it's still stable further down to 1.040, 1.030,...
Power consumption, and hence temperature, drops proportionally to voltage reduction squared - every milivolt helps.

By all means, you can try undervolting the second GPU too - you can't break it this way and may be able to drop the power usage by a nice margin.

Make sure to drop memory clocks as low as they'll go - you don't need fast memory for mining and the temperature drop can be significant when going from factory 1375 to 880 MHz.


portron
Member
**
Offline Offline

Activity: 106


View Profile
February 16, 2012, 03:35:57 PM
 #13



This style of fan design is not optimal for being in a case with multiple GPU's.

It spews hot air all over, and if you have more then one of them in the case, they get HOT.

The "OC" version is already clocked to 1000mhz, to achieve this I'm assuming the voltage was upped slightly to achieve a stable result.



Drop your memory clocks, undervolt your card, get a delta fan pointing directly at both of the cards and fans sucking the hot air out too.  If that one STILL runs hot, I'd just return it.  Get a reference design 7970 that blows the hot air out the back.  It's not really worth voiding your warranty to replace the thermal paste.  I'd be worried about the VRM's with the temps reaching 90c.
waterboyserver
Member
**
Offline Offline

Activity: 78


View Profile
February 20, 2012, 02:58:25 AM
 #14

Like the others have already mentioned, this type of card just blows the hot air from the graphics card around your computer case. A reference designed 7970 with a blower fan is quite loud, but gets the job done in displacing the heat produced by the card. Another consideration (which I currently use) is to get a full card waterblock (universal blocks do not cool 7970 mosfets, so the board will crash in minutes at high clocks under load).
MajorMiner13
Newbie
*
Offline Offline

Activity: 10


View Profile
February 20, 2012, 03:58:42 PM
 #15

Like the others have already mentioned, this type of card just blows the hot air from the graphics card around your computer case. A reference designed 7970 with a blower fan is quite loud, but gets the job done in displacing the heat produced by the card. Another consideration (which I currently use) is to get a full card waterblock (universal blocks do not cool 7970 mosfets, so the board will crash in minutes at high clocks under load).

I agree, but this doesn't seem to account for the problem the OP is having. Placing each card individually in his rig still results in one card running at much higher temps than the other.

I would avoid this design from the start. Apparently since they aren't reference, they also change the voltages as they please to achieve their meager overclock.

In my experience, reference design is always a superior solution.

splitt
Newbie
*
Offline Offline

Activity: 12


View Profile
February 20, 2012, 04:08:12 PM
 #16

Thanks to all for their advice.
I just returned it as faulty and am currently waiting for a replacement card withing two weeks.
I did a Furmark stress test on both cards separately and it was clear one of them was not OK, so I decided not to take chances with it.

I just hope they improve the miners for the new SDK 2.6, so that 79xx cards mine faster.
MajorMiner13
Newbie
*
Offline Offline

Activity: 10


View Profile
February 20, 2012, 04:16:44 PM
 #17

I just hope they improve the miners for the new SDK 2.6, so that 79xx cards mine faster.

With the newest cgminer (2.2.7) and the latest AMD drivers (12.2 pre-cert) I'm getting around 668 Mh/s with 1150/150, stock voltages. Temp is 64 degrees with 3000 rpm fan. Reference design.
Kettenmonster
Sr. Member
****
Offline Offline

Activity: 420


bool eval(bool b){return b ? b==true : b==false;}


View Profile
February 23, 2012, 08:02:15 PM
 #18

OK card
12 V  11.69
 HOT card
12 V  11.88
While you are waiting for replacement, did you check the 12V with just one card?

The paining (sic!) is done with the QPainter class inside the paintEvent() method.
(source: my internet)
deepceleron
Legendary
*
Offline Offline

Activity: 1470



View Profile WWW
February 23, 2012, 10:43:43 PM
 #19

The heatsink was likely not mounted correctly on the card that was too hot. If pervasive warranty-voiding anti-tamper stickers were not used, it likely could have been removed and reinstalled with new thermal paste and the problem fixed.

Pages: [1]
  Print  
 
Jump to:  

Sponsored by , a Bitcoin-accepting VPN.
Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!