small gpu load (poclbm.py, 3x6950 under ubuntu 11.04)

[Laminator]

Hi
I have a problem with gpu load under ubuntu 11.04 with catalyst 11.5 and ati stream 2.3
The problem is when I start mining with first gpu I got ~350 mh/s and around 95% gpu load
But when I start second and third gpus I've got this:

Code:

Adapter 0 - AMD Radeon HD 6900 Series 
                            Core (MHz)    Memory (MHz)
           Current Clocks :    870           1250
             Current Peak :    870           1250
  Configurable Peak Range : [500-900]     [1250-1325]
                 GPU load :    52%

Adapter 1 - AMD Radeon HD 6900 Series 
                            Core (MHz)    Memory (MHz)
           Current Clocks :    870           1250
             Current Peak :    870           1250
  Configurable Peak Range : [500-900]     [1250-1325]
                 GPU load :    54%

Adapter 2 - AMD Radeon HD 6900 Series 
                            Core (MHz)    Memory (MHz)
           Current Clocks :    870           1250
             Current Peak :    870           1250
  Configurable Peak Range : [500-900]     [1250-1325]
                 GPU load :    57%

and hashrate is about 190000 kh/s
Can anyone help me?

[1714098200]

1714098200

[1714098200]

1714098200

[1714098200]

1714098200

[1714098200]

1714098200

[1714098200]

1714098200

[colorprint]

I have a similar problem with radeon 6990 card... it has 2 GPUs...
With mining with 1 GPU I get 99% GPU load and 400MH/s.
With both GPUs load is 80% and total 620-660MH/s

[dc_lose]

i have the same problem! 10 btc if you help me to solve it

[Meatball]

Do you have enough power on your PSU for the three cards?  You probably need at minimum a 750W or so to handle that, preferably with a single rail or you'll overload the rails.  I've seen GPU's throttle if they aren't getting enough power.

[Laminator]

Do you have enough power on your PSU for the three cards?  You probably need at minimum a 750W or so to handle that, preferably with a single rail or you'll overload the rails.  I've seen GPU's throttle if they aren't getting enough power.

I have 1200W PSU and everything works fine in win7

[dc_lose]

Do you have enough power on your PSU for the three cards?  You probably need at minimum a 750W or so to handle that, preferably with a single rail or you'll overload the rails.  I've seen GPU's throttle if they aren't getting enough power.

I have 1200W PSU and everything works fine in win7

Yea PSU is not the problem

[dc_lose]

What are your miner command line settings?

Also, is your power supply large enough?


If possible, try downclocking the memory to 300 MHz, that might lift some strain on the PSU.

miner settings -v -w128
How to downcloack the memory in ubuntu 1104?

[Inaba]

Do you have the crossfire bridges connected?  If so, disconnect.

Have you disabled CF with aticonfig?  If not, aticonfig --crossfire=off

[Laminator]

Do you have the crossfire bridges connected?  If so, disconnect.

Have you disabled CF with aticonfig?  If not, aticonfig --crossfire=off

CF is disabled bridges are disconnected

[fasti]

I had this problem with my 2x 6950, I majorly fixed by you just leaving it alone. Unplug screen or log off or something.

Setting the miner very aggressive (-f 0) for secondary cards will fix it slightly also. It doesn't bother desktop responsiviness, only settings on the primary card effects it. Stacking up more miners per GPU also should make it use the GPU more, but I didn't need to do it.

So if it's headless computer, this should fix it.

If problem is CPU related with multiple cards, this might help:
"export GPU_USE_SYNC_OBJECTS=1" in terminal.

[fasti]

Ye using AMDOverdriveCtrl aswell. 880core 760memory(same temps as ~300-400, left it high for more stableness perhaps)

For multiple instances of the software for each GPU you need to start it with "-i = number" flag. You can get the info from "AMDOverdriveCtrl -h". For me the first card was "0", second "6".


I run it with: -v -w 64 -f 10(primary)/0(secondary)

[dc_lose]

What are your miner command line settings?

Also, is your power supply large enough?


If possible, try downclocking the memory to 300 MHz, that might lift some strain on the PSU.

miner settings -v -w128
How to downcloack the memory in ubuntu 1104?

Looks like your cards are highly restricted in over/underclocking range, aticonfig won't do. AMDOverdriveCtrl, on the other hand, might just do that.
Get it here:
http://sourceforge.net/projects/amdovdrvctrl/files/C%2B%2B%20sources/AMDOverdriveCtrl.1.2.0.tar.bz2/download

If you don't run headless, just start it and adjust the values. If you do, post, and I'll be back

http://img43.imageshack.us/img43/3699/screenshotfxx.jpg
with 3 x 6970

[Atroxes]

Didn't the Hardware Mining comparison wiki page state that SDK's 2.2 and 2.3 had a bug causing them to hog CPU resources?

Check your CPU load while mining, and change to 2.4 if possible.

[dc_lose]

Didn't the Hardware Mining comparison wiki page state that SDK's 2.2 and 2.3 had a bug causing them to hog CPU resources?

Check your CPU load while mining, and change to 2.4 if possible.

Ok I'll check it later

[dc_lose]

CPU usage is about 50%

I'm trying to install SDK 2.4

Code:

wget http://download2-developer.amd.com/amd/APPSDK/AMD-APP-SDK-v2.4-lnx64.tgz
sudo tar zxfv AMD-APP-SDK-v2.4-lnx64.tgz -C /opt
export LD_LIBRARY_PATH=/opt/AMD-APP-SDK-v2.4-lnx64/lib/x86_64/:$LD_LIBRARY_PATH
wget http://pypi.python.org/packages/source/p/pyopencl/pyopencl-0.92.tar.gz
tar zxfv pyopencl-0.92.tar.gz
cd pyopencl-0.92
./configure.py --cl-inc-dir=/opt/AMD-APP-SDK-v2.4-lnx64/include/ --cl-lib-dir=/opt/AMD-APP-SDK-v2.4-lnx64/lib/x86_64
make
sudo make install
cd poclbm
./poclbm.py

Code:

/poclbm$ ./poclbm.py
Traceback (most recent call last):
  File "./poclbm.py", line 27, in 
    platforms = cl.get_platforms()
pyopencl.LogicError: clGetPlatformIDs failed: invalid/unknown error code

What am I doing wrong?

[fasti]

http://forum.bitcoin.org/index.php?topic=7514.msg110334#msg110334

try that one for installing 2.4

[mikegogulski]

This is the fix for the high CPU utilization issue (assuming you don't have something else really wacky going on):

Code:

echo "export GPU_USE_SYNC_OBJECTS=1" >> ~/.bashrc
source ~/.bashrc

Then restart your miners.

[dc_lose]

http://forum.bitcoin.org/index.php?topic=7514.msg110334#msg110334

try that one for installing 2.4

thanks mate
I've installed SDK 2.4 but gpu usage is still ~50%

[fasti]

Did you try leaving it alone(logoff or unplugmonitor or..) and calculate from "2^32 * share/s = hash/s" from your pool site(~3hours to get most accurate). I know there's some variance, but should clearly see if they are still running 50% or 99%.

[Laminator]

http://forum.bitcoin.org/index.php?topic=7514.msg110334#msg110334

try that one for installing 2.4

thanks mate
I've installed SDK 2.4 but gpu usage is still ~50%

Crap. Now try underclocking your memory to 300 MHz, maybe that helps...

Ok, will try it later

Did you try leaving it alone(logoff or unplugmonitor or..) and calculate from "2^32 * share/s = hash/s" from your pool site(~3hours to get most accurate). I know there's some variance, but should clearly see if they are still running 50% or 99%.

yep, I've tried that, hashrate is 490mh/s instead of ~1200mh/s