Topic: My first Rig, need help troubleshooting 6870 GPU problems that just sprang up (Read 3629 times)

This was helpful to me!  I'm like a more amateur version of what you're doing.

• I'm also running a 6870, but just one of them.
  
• I'm also running GUIMiner, but had no idea about how to set these flag things.  I tried the ones you suggested above, and they boosted my Mhash from about 245-250 to 260-275.

Thank you, and any other hints would be appreciated. 

Played with settings today, phatk definitely is the culprit, as soon as I fire up a miner using that kernel the problems return.  Not sure why it was able to run at all for the first 24 hours like that.  In any case I have found that I get the most stability with guiminer by running 1 gpu using Phoenix with poclbm kernel, and the other 2 with the default opencl miner.  Each gpu avg. 280+Mhash/s.  Aggression/-f values kept low.... cards clocked to 1000/525, stock volts.  Upped the fan profiles so card temps range from high 60s at the bottom card to mid 70s at the top card (sandwich configuration with MSI board).

That's probably how I will keep this rig running.  Hopefully this helps others who might run into this behavior.

Update:

Back to mining, I switched to poclbm with the -v -w 256 flags.  I also am keeping the gpu temps down between 60-70c Currently running at 900 core / 1025 mem.  All three 6870 are currently mining with no usage drop offs.  I am going to let it run like this for a few hours and see what happens.

Here is the pertinent specs of my Rig:

OS: XP SP3
Motherboard: MSI 890FXA-GD70
Ram: 1GB
CPU: AMD Sempron 140
GPU: 3x XFX Radeon 6870 (2 cards have the 5 second dummy plug with a 68 Ohm resistor)
PSU: Corsair TX850

I got all my parts on Thursday and put her together.  After some testing with guiminer, yesterday I settled on using phoenix with the following flags:

-k phatk platform=0 device=0 VECTORS BFI_INT worksize=128 AGGRESSION=08 FASTLOOP=false

I was able to have 3 cards going all out and not lock up the processor like this.

By "all out", I mean I settled on OC'ing each card to 970  Mhz on the core and under clocking the memory down to 525 using Afterburner.  The temps were in the high 80's.  Earlier today I set the fans from auto to 55 to keep it in the low 80s after reading up on acceptable temps.  So from yesterday I had a couple system crashes, but it would reboot and I had guiminer and afterburner setup to restart with windows and it would get back to mining (it was avergaing about 860 Mhash/s).  But Then a few hours ago the GPU's usage would randomly stop and go to zero, sometimes one, two or all 3 AB graphs would just drop to zero.  I could stop and restart them in guiminer but they would go for a few seconds and then drop to zero.

At his point I've set the Cards back to 775 core / 525 mem  and still no luck

I looked in the event log and I'm getting alot these messages:

Event Type: Information
Event Source: ACEEventLogSource
Event Category: None
Event ID: 0
Date:  6/26/2011
Time:  11:24:49 PM
User:  N/A
Computer: DATA
Description:
0000000223: 2011-06-26 23:24:49:187    RT_MMVideo::InitializeInternetVideo GetMMVideoAdjustInfoItem Pulldown FAILED
 Error      Called by: ATI.ACE.CLI.Aspect.MMVideo.Graphics.Runtime.RT_MMVideo::InitializeInternetVideo                  processID:00568 threadID:(                ) domainName:(ccc.exe                    ) assemblyName:(CLI.Aspect.MMVideo.Graphics.Runtime, Version=2.0.4161.39442, Culture=neutral, PublicKeyToken=90ba9c70f846762e)
------------------------------------------------------------------------------------------------------------------------
Event Type: Information
Event Source: ACEEventLogSource
Event Category: None
Event ID: 0
Date:  6/26/2011
Time:  11:24:49 PM
User:  N/A
Computer: DATA
Description:
0000000222: 2011-06-26 23:24:49:187    RT_MMVideo::InitAdvancedVideoFeature: Failed in VideoItem = CONTROLLER_VIDEO_MOSQUITONOISEREMOVAL
 Error      Called by: ATI.ACE.CLI.Aspect.MMVideo.Graphics.Runtime.RT_MMVideo::InitAdvancedVideoFeature                 processID:00568 threadID:(                ) domainName:(ccc.exe                    ) assemblyName:(CLI.Aspect.MMVideo.Graphics.Runtime, Version=2.0.4161.39442, Culture=neutral, PublicKeyToken=90ba9c70f846762e)
------------------------------------------------------------------------------------------------------------------------
Event Type: Information
Event Source: ACEEventLogSource
Event Category: None
Event ID: 0
Date:  6/26/2011
Time:  11:24:49 PM
User:  N/A
Computer: DATA
Description:
0000000221: 2011-06-26 23:24:49:187    RT_MMVideo::InitAdvancedVideoFeature: Failed in VideoItem = CONTROLLER_VIDEO_DEBLOCKING
 Error      Called by: ATI.ACE.CLI.Aspect.MMVideo.Graphics.Runtime.RT_MMVideo::InitAdvancedVideoFeature                 processID:00568 threadID:(                ) domainName:(ccc.exe                    ) assemblyName:(CLI.Aspect.MMVideo.Graphics.Runtime, Version=2.0.4161.39442, Culture=neutral, PublicKeyToken=90ba9c70f846762e)
------------------------------------------------------------------------------------------------------------------------

At this point Only  my main GPU hooked up to the monitor is mining intermittently.  Any ideas on why it has just shutdown?  Any tests I should run?