Bitcoin Forum
June 20, 2024, 04:14:30 AM *
News: Latest Bitcoin Core release: 27.0 [Torrent]
 
   Home   Help Search Login Register More  
Pages: [1]
  Print  
Author Topic: Strange as hell GPU temp problem  (Read 5624 times)
CryptoCluster (OP)
Member
**
Offline Offline

Activity: 84
Merit: 10



View Profile
July 22, 2013, 08:47:20 PM
 #1

I have few 4 x 7970 rigs. All of them are hashing 720-750khs (litecoin currently). But on one of the rigs 3rd GPU runs a lot hotter than on others. 1, 2 and 4 GPU temps are 65-75 but 3rd GPU is 95-105!
I initially thought it is faulty GPU, but if you swap any card in that 3rd slot is starts running hot as hell.
I have tried running it using riser as well as plugged directly into motherbord pcie.

What might be the reason?

"The cumulative development of a medium of exchange on the free market — is the only way money can become established. ... government is powerless to create money for the economy; it can only be developed by the processes of the free market." M. N. Rothbard
kuusj98
Hero Member
*****
Offline Offline

Activity: 812
Merit: 1000


I <3 VW Beetles


View Profile
July 22, 2013, 09:38:10 PM
 #2

You may want to have a look at your VRM temps, you would not be the first one to blow one out...
May be that power is not distributed right, power always takes the shortest way...
CryptoCluster (OP)
Member
**
Offline Offline

Activity: 84
Merit: 10



View Profile
July 22, 2013, 10:23:49 PM
 #3

Card on problematic pcie is now underclocked and undervolted to maintain sub-70 temps. Sadly i cannot see the VRM temps in GPU-Z or in any other monitoring software (GPUs are Gigabyta GV-R797OC-3GD voltage locked).

Is it possible that some strange mobo hardware error interferes with VRM on GPUs overvolting it somehow?

"The cumulative development of a medium of exchange on the free market — is the only way money can become established. ... government is powerless to create money for the economy; it can only be developed by the processes of the free market." M. N. Rothbard
kuusj98
Hero Member
*****
Offline Offline

Activity: 812
Merit: 1000


I <3 VW Beetles


View Profile
July 22, 2013, 10:27:57 PM
 #4

Card on problematic pcie is now underclocked and undervolted to maintain sub-70 temps. Sadly i cannot see the VRM temps in GPU-Z or in any other monitoring software (GPUs are Gigabyta GV-R797OC-3GD voltage locked).

Is it possible that some strange mobo hardware error interferes with VRM on GPUs overvolting it somehow?
Hmm, I dont think so, the clocks are handled by the gpu itself, so are the volts...

I suggest changing the PCI-e cables to that card, maybe the cable is faulty.
Does seem strange anyway...
crazyates
Legendary
*
Offline Offline

Activity: 952
Merit: 1000



View Profile
July 23, 2013, 07:12:04 PM
 #5

Does your 3rd slot have bad airflow?

Tips? 1crazy8pMqgwJ7tX7ZPZmyPwFbc6xZKM9
Previous Trade History - Sale Thread
CryptoCluster (OP)
Member
**
Offline Offline

Activity: 84
Merit: 10



View Profile
July 23, 2013, 08:39:22 PM
 #6

Does your 3rd slot have bad airflow?

Same as other slots. Today I have tried different PSU, nothing changed.

"The cumulative development of a medium of exchange on the free market — is the only way money can become established. ... government is powerless to create money for the economy; it can only be developed by the processes of the free market." M. N. Rothbard
rascal777
Full Member
***
Offline Offline

Activity: 132
Merit: 100


View Profile
July 23, 2013, 11:40:57 PM
 #7

I notice that in Cgminer that the Temps sometimes dont seem to line up with the Fan Control.

For example:  One GPU will seem to run hotter than the average, and then there will be aonther GPU  that is colder than average.

So What I think is that as CGminer turns up the fan Rate to compensate for a hotter card  it actually turns up a fan on the wrong card, thus now you see the cooler temp on another card.

I have not found a solution for this yet.  beginning efforts with gpu mapping havent seemed to work for me yet, or I haven't mapped them correctly.


BTC TIPS  19n2ienyueN4RiC38KFSZMQMgrNLgu9Uuc
Equate
Hero Member
*****
Offline Offline

Activity: 770
Merit: 500


View Profile
July 24, 2013, 08:56:15 AM
 #8

are you using risers ?
rascal777
Full Member
***
Offline Offline

Activity: 132
Merit: 100


View Profile
July 25, 2013, 04:24:39 PM
 #9

not question for me, but I am using risers

BTC TIPS  19n2ienyueN4RiC38KFSZMQMgrNLgu9Uuc
CryptoCluster (OP)
Member
**
Offline Offline

Activity: 84
Merit: 10



View Profile
July 26, 2013, 10:21:59 AM
 #10

@Equate: Some cards are on riser (not powered risers), and some are connected directly into the mainboard.
3rd slot card was tested both ways, but it does not change temp readings.

@rascal777: I have tried manually changing Fan speed to 100%, it does not help.


Today I have used IR thermometer, just in case only readings in GPU-Z and cgminer are wrong, but they are not...

"The cumulative development of a medium of exchange on the free market — is the only way money can become established. ... government is powerless to create money for the economy; it can only be developed by the processes of the free market." M. N. Rothbard
peonminer
Hero Member
*****
Offline Offline

Activity: 798
Merit: 531


Crypto is King.


View Profile
July 26, 2013, 06:10:33 PM
 #11

Does the card have a display plugged in? I have to run my display card lower than the rest.... Download CGWatcher and learn it. It'll save your rig when you're not around and even when you are.......................
Pages: [1]
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!