BAMT version 0.5 - Easy USB based mining Linux with farm wide management tools

[lodcrappo]

I've been running BAMT here and my Rig will restart every couple of hours by itself, I think it's due to high temp on some cards.

Problem is when it restarts, it can't boot up as some of the cards seems locked and the system freezes when loading X.

When this happens I can only make it boot by removing the locked card from the rig...

Anybody with similar issues?

anybody else who overclocks their cards way too much will have similar issues. 

reduce overclocking.

[1713949209]

1713949209

[1713949209]

1713949209

[1713949209]

1713949209

[Sant001]

Hmm, I will do that. Right now I think my system is on "auto", that means I haven't set OC on bamt.conf so I guess it's on auto-pilot.

I will be looking in the forums what are the best settings for Gigabyte 5870 that I have here.

By the way these locked cards how to make them work again? They seem to work again if I wait a long, long time before using them again. Any other way?

Almost forgot to ask, is it possible to prevent BAMT from restarting automatically?

[lodcrappo]

Hmm, I will do that. Right now I think my system is on "auto", that means I haven't set OC on bamt.conf so I guess it's on auto-pilot.

I will be looking in the forums what are the best settings for Gigabyte 5870 that I have here.

By the way these locked cards how to make them work again? They seem to work again if I wait a long, long time before using them again. Any other way?

If you are having these types of problems without any overclocking, you have one or more of the below:

A - defective hardware
B - severe thermal problems with your design
C - insufficient power supply

check all that apply.  bamt will not work properly on this system until above issue(s) are resolved.

never heard of GPUs that have to wait before they will work again.  I suspect you have some damaged goods there

[Sant001]

I have purchased my PSU's and video cards 2nd hand, so it might be possible that they're defective.

But I'm also using cable extenders and a no brand motherboard that are also candidates to be causing these issues.

For example I have one 5870 that will not work when connected via extenders, but works flawlessly when connected directly to the X16 slot. On the other hand, my 5970's work without issue on the very same extender.

So I guessed my extenders are fine, just not compatible with some cards. Perhaps due to some interference caused by longer cables?

I'm having difficulty troubleshooting this because it seems to have too many variables/candidates to be causing these issues  =/

[Sant001]

If you are having these types of problems without any overclocking, you have one or more of the below:

A - defective hardware
B - severe thermal problems with your design
C - insufficient power supply

check all that apply.  bamt will not work properly on this system until above issue(s) are resolved.

never heard of GPUs that have to wait before they will work again.  I suspect you have some damaged goods there

B - I think the system is rebooting for high temp, but even after the cards are cool system still won't start. The only way I can make it start is by removing each card until I find the one with issues.
C - My rig has 3x 5970 (3x 350w = 1050w) and 1x 5870 (190w), so 1050+190+~200 for other components totals 1440w. I'm using 2x PSU BQT-PRO8-1200W, that gives me a total of 2400W. That should be enough, only way I coud see the PSU being the problem is that it's 2nd hand and could be defective but I think that's unlikely.

Could my rig be restarting as a result of some BIOS setting? I had one time where the system crashed and restarted. Then it wouldn't boot up and I did a BIOS reset and it booted normally again. That's why I suspect the motherboard could be causing all this.

[lodcrappo]

If you are having these types of problems without any overclocking, you have one or more of the below:

A - defective hardware
B - severe thermal problems with your design
C - insufficient power supply

check all that apply.  bamt will not work properly on this system until above issue(s) are resolved.

never heard of GPUs that have to wait before they will work again.  I suspect you have some damaged goods there

B - I think the system is rebooting for high temp, but even after the cards are cool system still won't start. The only way I can make it start is by removing each card until I find the one with issues.
C - My rig has 3x 5970 (3x 350w = 1050w) and 1x 5870 (190w), so 1050+190+~200 for other components totals 1440w. I'm using 2x PSU BQT-PRO8-1200W, that gives me a total of 2400W. That should be enough, only way I coud see the PSU being the problem is that it's 2nd hand and could be defective but I think that's unlikely.

Could my rig be restarting as a result of some BIOS setting? I had one time where the system crashed and restarted. Then it wouldn't boot up and I did a BIOS reset and it booted normally again. That's why I suspect the motherboard could be causing all this.

sounds like a system that is going to take a lot of time.  if it were me, I'd tear it down to one GPU, get that working stable, and add stuff back one part at a time.  no easy fix.

[Sant001]

It seems to work well and quite stable with only 1 card, I might be putting stress on the parts by pushing the limits with 4, even 5 cards I have tried on this cheap motherboard.

I noticed something interesting, if I add 4x 5970 it won't boot up, it will always freeze at the time it's loading X. Not sure why, the kernel was supposed to support 8GPUs which is exactly how many I'd since each 5970 has 2 GPUs.

Maybe it's drawing too much power from the PCI-E slots? My system might get more stable if I use 2x 5870 + 2x 5970 instead of 4 or 3 5970 like I'm doing now. I plan to try that soon.

[Sant001]

If you are having these types of problems without any overclocking, you have one or more of the below:

A - defective hardware
B - severe thermal problems with your design
C - insufficient power supply

check all that apply.  bamt will not work properly on this system until above issue(s) are resolved.

never heard of GPUs that have to wait before they will work again.  I suspect you have some damaged goods there

B - I think the system is rebooting for high temp, but even after the cards are cool system still won't start. The only way I can make it start is by removing each card until I find the one with issues.
C - My rig has 3x 5970 (3x 350w = 1050w) and 1x 5870 (190w), so 1050+190+~200 for other components totals 1440w. I'm using 2x PSU BQT-PRO8-1200W, that gives me a total of 2400W. That should be enough, only way I coud see the PSU being the problem is that it's 2nd hand and could be defective but I think that's unlikely.

Could my rig be restarting as a result of some BIOS setting? I had one time where the system crashed and restarted. Then it wouldn't boot up and I did a BIOS reset and it booted normally again. That's why I suspect the motherboard could be causing all this.

sounds like a system that is going to take a lot of time.  if it were me, I'd tear it down to one GPU, get that working stable, and add stuff back one part at a time.  no easy fix.

Thanks so much for all the insights lodcrappo and for BAMT as well of course 

[tosku]

I updated BAMT with the latest version of cgminer. This is how I did it. I'm sure there are more elegant or "correct" ways, but this worked for me. If I'm doing something horribly wrong, please point it out.

Logged in as root, type this into a terminal to download the latest cgminer and prepare it with the ADL SDK:

Code:

cd /usr/share
git clone https://github.com/ckolivas/cgminer.git
cd cgminer/ADL_SDK
wget http://download2-developer.amd.com/amd/GPU/zip/ADL_SDK_4.0.zip
unzip ADL_SDK_4.0.zip
cd include
mv adl* ..
cd ..
cd ..

Keep going. The following instructions on compiling cgminer are taken from http://ck.kolivas.org/apps/cgminer/README

Code:

cd /opt
tar xf /opt/AMD-APP-SDK-v2.4-lnx32/icd-registration.tgz
ln -s /opt/AMD-APP-SDK-v2.4-lnx32/include/CL /usr/include
ln -s /opt/AMD-APP-SDK-v2.4-lnx32/lib/x86/* /usr/lib/
ldconfig

cd /usr/share/cgminer
./autogen.sh
CFLAGS="-O2 -Wall -march=native" ./configure
make

Now that cgminer is compiled, you need to replace the cgminer that came with BAMT with the latest version. It's a good idea to stop mining before doing that, and to make a backup of the old cgminer:

Code:

gpumon
D
q
cd /opt/miners
mkdir cgminer_backup
cd cgminer
cp * ../cgminer_backup/

cd /usr/share/cgminer
cp * /opt/miners/cgminer

Now restart mining and verify that things work as they should:

Code:

gpumon
S
0

You should see the latest version of cgminer running in a screen.

[bulanula]

I updated BAMT with the latest version of cgminer. This is how I did it. I'm sure there are more elegant or "correct" ways, but this worked for me. If I'm doing something horribly wrong, please point it out.

Logged in as root, type this into a terminal to download the latest cgminer and prepare it with the ADL SDK:

Code:

cd /usr/share
git clone https://github.com/ckolivas/cgminer.git
cd cgminer/ADL_SDK
wget http://download2-developer.amd.com/amd/GPU/zip/ADL_SDK_4.0.zip
unzip ADL_SDK_4.0.zip
cd include
mv adl* ..
cd ..
cd ..

Keep going. The following instructions on compiling cgminer are taken from http://ck.kolivas.org/apps/cgminer/README

Code:

cd /opt
tar xf /opt/AMD-APP-SDK-v2.4-lnx32/icd-registration.tgz
ln -s /opt/AMD-APP-SDK-v2.4-lnx32/include/CL /usr/include
ln -s /opt/AMD-APP-SDK-v2.4-lnx32/lib/x86/* /usr/lib/
ldconfig

cd /usr/share/cgminer
./autogen.sh
CFLAGS="-O2 -Wall -march=native" ./configure
make

Now that cgminer is compiled, you need to replace the cgminer that came with BAMT with the latest version. It's a good idea to stop mining before doing that, and to make a backup of the old cgminer:

Code:

gpumon
D
q
cd /opt/miners
mkdir cgminer_backup
cd cgminer
cp * ../cgminer_backup/

cd /usr/share/cgminer
cp * /opt/miners/cgminer

Now restart mining and verify that things work as they should:

Code:

gpumon
S
0

You should see the latest version of cgminer running in a screen.

Thanks for these instructions. Always wondered how to upgrade cgminer to latest version in BAMT !

[philips]

You can also use the tarball from Inaba:
https://bitcointalk.org/index.php?topic=65915.msg901078;topicseen#msg901078
You just need to run the update script.

[Sant001]

I'm getting this error:

Code:

xauth: (argv):1:  merge:  unable to open file /home/user/.Xauthority

And when this happens, gpumon shows no card at all. And load unusually high at 8..

Sounds like my usb flash drive is failing?

[mc_lovin]

I'm getting this error:

Code:

xauth: (argv):1:  merge:  unable to open file /home/user/.Xauthority

And when this happens, gpumon shows no card at all. And load unusually high at 8..

Sounds like my usb flash drive is failing?

No, it's not failing.. I don't fully understand it, but X is what BAMT uses for it's video display..  Your best bet is flash the USB again and try again.  Also, are you using a monitor?  Your best bet is flash the USB fresh, plug it in, and turn on the rig with no monitor connected and use PuTTY to SSH in and configure it.  If it still says Xauthority, flash it again. 

[Inaba]

That error is a result of your SSH client trying to forward X packets, not any particular setting in BAMT.

[Cray-1]

I am having trouble with one of my rigs.  Approximately once per 24 hours it stops mining and displays the “Red Screen” stating  "A fault has been found in gpu x" for all five GPU’s.

I am running a msi 890fxa-gd70 with 5 Saphire 5830’s and a Cooler Master 1200w under BAMT v0.5c.  I have 3 other near identical rigs that do not seem to be having any problems.  The rig will be running fine and then just stop mining and give me the Red Screen.  I can determine no pattern as to when it stops working.  I have been having this problem for a while so the first thing I did was stop all overclocking.  This did not help.

Here are some clues, any help would be appreciated.

1.   While in the red screen I went to BAMT control and clicked on Stop Mining (this seemed to act normal).  Then I clicked on Start Mining and the start screen seemed to be acting normal.  When I went to GPU status screen it displayed all 5 cards producing 0 M/H.
2.   While in the red screen I went to a windows computer on the same network and ran Putty.  Putty was not able to communicate with the problem rig.

I have switched out the BAMT usb drive – does not help.  I am not overclocking.  It must be hardware (motherboard?) or the power supply, I guess.  The bios on the mobo has been updated to the latest approved version.  What tests can I run to narrow this down?  Thanks

[seriouscoin]

I havent checked BAMT for a while,... sorry if this is already covered.

Has anyone successfully run 7970s with BAMT?

Even with latest cgminer? I guess you also need to update to the latest SDK? (is it possible to update SDK in Bamt?)

[BitMinerN8]

I am having trouble with one of my rigs.  Approximately once per 24 hours it stops mining and displays the “Red Screen” stating  "A fault has been found in gpu x" for all five GPU’s.

I am running a msi 890fxa-gd70 with 5 Saphire 5830’s and a Cooler Master 1200w under BAMT v0.5c.  I have 3 other near identical rigs that do not seem to be having any problems.  The rig will be running fine and then just stop mining and give me the Red Screen.  I can determine no pattern as to when it stops working.  I have been having this problem for a while so the first thing I did was stop all overclocking.  This did not help.

Here are some clues, any help would be appreciated.

1.   While in the red screen I went to BAMT control and clicked on Stop Mining (this seemed to act normal).  Then I clicked on Start Mining and the start screen seemed to be acting normal.  When I went to GPU status screen it displayed all 5 cards producing 0 M/H.
2.   While in the red screen I went to a windows computer on the same network and ran Putty.  Putty was not able to communicate with the problem rig.

I have switched out the BAMT usb drive – does not help.  I am not overclocking.  It must be hardware (motherboard?) or the power supply, I guess.  The bios on the mobo has been updated to the latest approved version.  What tests can I run to narrow this down?  Thanks

When it "goes red" do you still have access to the internet? For example, can you open Iceweasel and go to Google? I was having this issue too, turned out to be this strange "lost the DHCP lease" issue that has been going around but hard to pin down. I put an old Intel Pro/100 card in and the issue has left that rig.

[BitMinerN8]

I havent checked BAMT for a while,... sorry if this is already covered.

Has anyone successfully run 7970s with BAMT?

Even with latest cgminer? I guess you also need to update to the latest SDK? (is it possible to update SDK in Bamt?)

I think there was a bounty, don't know where it's at though. (% Funded)

[Joshwaa]

I think Lod said it was funded. Im waiting to hear from him I got a Daimond HD7970 I was gonna send him for 80BTC (well 72BTC now since price went up). Unless some one promised to pay and has not come thru yet.

[mc_lovin]

I havent checked BAMT for a while,... sorry if this is already covered.

Has anyone successfully run 7970s with BAMT?

Even with latest cgminer? I guess you also need to update to the latest SDK? (is it possible to update SDK in Bamt?)

I think there was a bounty, don't know where it's at though. (% Funded)

I have two 7970s running in BAMT 0.5c but I can't seem to get the BAMT scripts to start the mining, I have to do it manually but it still reports the speeds to mgpumon... however the 7970 integration mining fund has been completed and lodcrappo will have the updated version soon.