Need Tech Support for mining >.<

[Soaron123]

Hi, I'm just having a bit of an issue running 4x6990 (8 gpus - 2 power supplies - 2400W) on GUI miner.

Running Win7x64, All 8 GPU's show up on CCC, device manager and GUI miner (using catalyst 11.7 with SDK 2.5)

However, only 6 of the gpu's (any 6) mine at a given time. The other two just hang with "connection problems."

There have been occasions where they do connect, however, although they do show a hash rate and an increase in total hash rate, they aren't actually contributing anything.

Is this a problem with GUIminer? The miners are all openCL (poclbm). Sometimes I get the windows error message "Poclbm has stopped responding" - but GUIminer seems unaffected.

Was thinking of going to 11.6b as people say thats more stable. Was also thinking of making a .bat file instead of using GUIminer.

anyone know what the problem could be?

[hmblm1245]

Not 100 percent on this but windows may not be able to see all 8 gpus. i think that was fixed with 11.7 but its worth a search. Windows does not like more than 6ish gpus in general. I have had issues with win7 and 8 gpus and switched to 4 to be more stable. just a thought.

[Soaron123]

nah, pretty sure thats not the case. I can see all in device manager, they all show up in GUIMiner as well as MSI after burner and CCC. I've got dummy plugs, but it doesn't matter if I have them in or not (11.7).

[hmblm1245]

in the ccc do you have the power control at +20?

[Soaron123]

all the GPU's are clocked at 925mhz, with the Mem clock at 1250mhz (lowest).

I've left the power at 0%

[hmblm1245]

I was referring to the power control setting which allows additional power to the cards when OCed.  http://www.legitreviews.com/article/1488/4/

See the third slider in the ccc is that set to +20. i noticed i had to make this change to +20 so all my cards receive enough power when needed.

[Soaron123]

yeh I know what you're referring to, I've got it set to default which it states as 0%. I'll try fiddle around with it now and see how I go.

[hmblm1245]

Also your temps ok? I assume they are on risers.

[Soaron123]

All the temps are fine.

I tried a few things:

- I set the cards to standard clocks (830mhz) with 0 power control - still only 6 are able to run at a time
- clocks @ 830mhz, power control @ +20 - still only 6
- clocks @ 925mhz, power control @ +20 - still only 6
- remove all dummy plugs - all 8 run, however one of them will show a false positive (give a m/hash, but not actually be doing anything)

[hmblm1245]

Let me see if i got this.

so if all the dummy plugs are removed, guiminer sees 8 gpus. as you plug in the monitor to each video card to start mining one of the 8 gpus are not mining at all.

You may have already tried this but its worth a shot.

when you say "however one of them will show a false positive (give a m/hash, but not actually be doing anything" you know this by looking in MSI AB gpu usage graphs?

can you find out which card and gpu is not doing anything and then try swapping physically slots on the mobo  and 6 pin power to that card with a working one. Just to eliminate any potential defects with that card.

[km3b]

What motherbord do you have? Some, if not most, does not have 4 PCI-e slots that can be run all at once. Sometimes they are shared.

[jonand]

Try this:

Downloading the linuxcoin USB image, install it on a USB stick, boot from the USB stick and see if the miners work from there.

I'm not saying you have to run linux - but by doing that you can rule out hardware or win7 as the problem.

Good luck.

[toliman]

if i remember, the ATI temperature/fan monitors i downloaded, have cards 0 to 5 in win7 x64 and cat 11.7.

this could be the reason why it can't talk to cards 7 and 8.

i'll see if i can find a link to confirm it's not a driver/interface problem ... perhaps it's the reason why GUIminer only sees cards 1-6.

[Soaron123]

I'll get to trying the Linuxcoin solution straight away.

It seems people are misunderstanding the problem. ALL cards SHOW up. This isn't an issue of one card not working or not, it seems to be more of an ISSUE with GUIminer.

ALL graphics cards WORK and can MINE.

ALL 8 GPUS show up. When trying to mine, however, ANY SIX will mine, and the last TWO to press 'start' will have the message "Connection problems."
If i start gpus 1 through to 8 in that order, 1,2,3,4,5,6 will MINE, while 7 and 8 will say "connection problems."
If I do it in reverse, starting with 8 and finishing at 1 - 8,7,6,5,4,3 will MINE while 2 and 1 will say "connection problems."

Eventually I was able to fix this, but when they do ALL mine, one of them isn't actually mining even though GUIminer says it is. The M/Hash increases to around 350 and then just stays fixed at that number, never fluctuating like the other GPUS and shows 0% activity on CCC and MSI afterburner.

my next steps are trying Linuxcoin, and then trying the command line of poclbm instead of GUIminer. However, while using GUIminer, every now and then I get "Poclbm.exe has stopped responding".

[Soaron123]

Bump and update.

Tried several things.

-Linuxcoin seemed too a bit too complicated to get to run as I'm not too familar with the Ubuntu architecture and theres no GUI for mining.
-Tried using different frontend, AOCLBF - couldn't get this to work in general. Had a problem with the OSD.
-Uninstalled drivers, reinstalled, tried lower version catalyst drivers.
-Messed around with the CPU affinities for each core, although not too sure if this did anything at all.
-Tried having 8 miners, where one miner is a CPU - same issue as if I was trying to have 8gpu miners leading to my conclusion.
-Tried having 7 openCL miners with 1 pheonix miner - same issue as above

Conclusion:

My current rig only has 2gb ram, so this might be an issue, but I'm pretty sure it isnt as everything is stable with 7 miners. Almost unnoticeable lag.
Not sure what this problem is related to, possibly poclbm.exe (as it seems to always crash for the last miner), or maybe it has to do with 'conhost.exe.' I noticed with each miner started, there is an additional poclbm.exe or pheonix.exe (depending on miner type) opened as a process. Along with this is conhost.exe process. Maybe windows is unable to have too many of these open at a given time?

I'm 100% sure this is not a gpu issue, rather having 8 independent miners on the same rig is the problem.

[max in montreal]

I am having the same problems with4 cards on my MSI 890FXA-GD70.

5870
5970
5970
5970

[Elder III]

I had the identical problem that you have had, except I was trying to get 5 cards to work instead of 4.  I finally chalked it up to being a Windows Xp thing.  I've tried to confirm that XP can or cannot support more than 4 video cards (even with latest drivers) but I have not been able to either confirm nor deny that so far.  I do wonder if perhaps your motherboard cannot supply enough power to all the video cards.... do you have some of them with risers that have the molex adapter so that they draw all their power from the PSU?