Starting multiple miners in Ubuntu

[um0rion]

Let me start of by saying im a complete linux noob... Ive had like 2 days of experience on it now, and Ive got an issue that I dont know how to solve.

Ive been following a guide for ubuntu 11.04 and it had me create a script called "startminer.sh" to start my miner. This works fine, for the first two GPUs. I created two more scripts, startminer1.sh and startminer2.sh with the same commands, just initializing the phoenix miner on cores 1 and 2. Now, when I start up the last miner, my system blackscreens and I have to shut off the psu and reboot.

Is this something I can troubleshoot with code, or is it some sort of instability? I dont have a kill-a-watt on the system yet, but the power suply Im using is an OCZ GameXstream 700w. Would 3 gpu's and the system take me over that much? Im running a very similar system with 4x 5850s and it only takes 650w or so, you'd think 3x 5850s would be less than that. These are reference models, but could they really suck that much power?

[1714148492]

1714148492

[1714148492]

1714148492

[hoo2jalu]

...when I start up the last miner, my system blackscreens and I have to shut off the psu and reboot.

Is this something I can troubleshoot with code, or is it some sort of instability? I dont have a kill-a-watt on the system yet, but the power suply Im using is an OCZ GameXstream 700w...

Power to 1kw will fix. If you have supported sensor on motherboard you could perhaps watch voltage drop to confirm. Check for supported lm-sensors or sensor-applet.

[Phlegmsandwich]

Try starting the third GPU first.   If it works fine, than it is probably a problem caused by the PSU.

[um0rion]

...when I start up the last miner, my system blackscreens and I have to shut off the psu and reboot.

Is this something I can troubleshoot with code, or is it some sort of instability? I dont have a kill-a-watt on the system yet, but the power suply Im using is an OCZ GameXstream 700w...

Power to 1kw will fix. If you have supported sensor on motherboard you could perhaps watch voltage drop to confirm. Check for supported lm-sensors or sensor-applet.

Try starting the third GPU first.   If it works fine, than it is probably a problem caused by the PSU.

As of right now Im pretty sure its the power supply. I did some research, and I guess the visiontek 5850 pulls a lot of voltage. Since Im going to have 6x cards on this system when Im done, Ill look into buying an 800-900w PSU to add to the system.

Its really impressive that 4x Sapphire 5850 Xtreme models can run at 650w load, and 3x Visiontek 5850s can shut down a 700w power supply. Im kinda disappointed, they dont sell the Xtremes anymore, and they're by FAR the most efficient card.

[um0rion]

Ok, new update... its NOT the power supply. I have 3 videocards on the motherboard, and when two of them are mining I use 380w. A 700w power supply should handle this with ease. Could this be something to do with the mining script im using? I used the one from a guide I found, which overclocks the card and starts mining all in one motion. Whatever it is locks up the system immediately after initializing the 3rd card.

Like I said Im completely new to this, but If I wanted to just run a terminal and start mining, how would I go about doing that? Any guides or stuff like that you guys can point me to? I can worry about overclocking later, the most important thing is to get this thing running.

Could it be the slots Im using on the motherboard? I wouldnt think so, but I guess its possible. Im using an MSI 890fxa-gd70. The board should be able to run 6 cards, right? I thought it might be a problem with the combination of slots im using, being the top one, middle one, and bottom one, and that maybe it needs 4+ to work properly.

[hugolp]

You should post the script so we can give you an opinion.

Also, did you try to launch the cards in different order? Are you sure its not a particular card failing, the one you launch last?

[um0rion]

You should post the script so we can give you an opinion.

Also, did you try to launch the cards in different order? Are you sure its not a particular card failing, the one you launch last?

Ill do that... As for launching them in different order, they all work fine. I can start 3, 2, then 1 will make it crash, or I can do 1, 2, and 3 will make it crash. Cards 1, 2, or 3 by themselves all start fine.

This was the guide: http://forum.bitcoin.org/index.php?topic=9239.0
This is the script, I made 2 more copies but adjusted for adapter 0, 1, and 2. This is the script for adapter 0.

cd ~
   sudo touch /usr/local/bin/startminer.sh
   sudo chmod 755 /usr/local/bin/startminer.sh
   sudo gedit /usr/local/bin/startminer.sh

      # ${1} is used as a variable for the username, password and for the gpu device number.  Ex. mineruser0, minerpass0, Device=0 or mineruser1, minerpass1, Device=1
      HOMEDIR=/home/matt/pyopencl-0.92
      MINERSERV=mining.bitcoin.cz:8332
      MINERUSER=xxxxxxxxxxxxxxxxxx
      MINERPASS=xxxxxxxxxxxxxxxxxx

      export AMDAPPSDKROOT=${HOMEDIR}/AMD-APP-SDK-v2.4-lnx64/
      export AMDAPPSDKSAMPLESROOT=${HOMEDIR}/AMD-APP-SDK-v2.4-lnx64/
      export LD_LIBRARY_PATH=${AMDAPPSDKROOT}lib/x86_64:${LD_LIBRARY_PATH}

      #Overclock GPU to 775Mhz
      DISPLAY=:0 aticonfig --od-enable --adapter=all
      DISPLAY=:0 aticonfig --od-setclocks=775,1000 --adapter=0
      cd ${HOMEDIR}/phoenix-1.48
      echo "Startming Miner: 0"
      ${HOMEDIR}/phoenix-1.48/phoenix.py -u http://${MINERUSER}:${MINERPASS}@${MINERSERV} -k phatk VECTORS BFI_INT AGGRESSION=8 DEVICE=0

[hugolp]

I dont see nothing wrong, but there is something weird:

DISPLAY=:0 aticonfig --od-enable --adapter=all

Why adapter=all? You want adapter 0 right? Also, once you do od-enable once you dont need to do it again, ven after rebooting the computer. I think you can delete that line. Try without it, maybe it does a difference although I doubt it.

[um0rion]

Guys, I cant get this working right. I bought another motherboard thinking it was the problem (MSI suggested an RMA) and the new motherboard does the same thing.

The first two cards start fine, and mine happily away, but I start the 3rd card, and get a message that the phoenix core is starting, then it IMMEDIATELY shuts down. Power supply shuts right off. Is there some way to troubleshoot this problem? I dont think im doing anything fancy that would make it crap out like that, but I really dont know my way around linux...

How would I just launch a miner from the terminal, is there just 1 command I can run (length doesnt matter) that would start a miner? I kind of want to try that rather than the script Im using. Windows is looking really tempting, but Im trying to run 6 cards eventually on this, and just getting 2 to work isnt cutting it.

[detroit]

Maybe check out that 700W power supply on sites like johnnyguru.com or hardocp.com, and see if it's solid enough to actually handle the load.  A lot of respectable sounding power supplies are actually pretty cheap and deceiving.
You seem to have a Kill-a-Watt now, how much are each of those cards pulling?  You'll know how much one of those cards draws you can make a guess where you'll be with all 3 started.

[um0rion]

Maybe check out that 700W power supply on sites like johnnyguru.com or hardocp.com, and see if it's solid enough to actually handle the load.  A lot of respectable sounding power supplies are actually pretty cheap and deceiving.
You seem to have a Kill-a-Watt now, how much are each of those cards pulling?  You'll know how much one of those cards draws you can make a guess where you'll be with all 3 started.

I think Ive found the issue and Im working on resolving it. It WAS respectable when I bought it, but It wasnt designed with 3-4 cards in mind. Its got 2x 6 pin connectors for starters, plus its got like 4 +12v power rails, rated I think at 18 amps, but the power layout is sketchy. Here's the power setup...


With 2 cards full load and 1 idle it pulls 356w from the wall. An almost identical rig I have downstairs runs 4 cards and pulls 680w from the wall...  Same motherboard, same chip, same ram. Im thinking that Ive maxed out a rail and its shutting down the power supply for that reason. Does that make sense, or ?

[detroit]

That makes a lot of sense.
It's a hassle, but the cheap test is to swap the known workable power supply into the troublesome machine.
Or you can go buy another power supply and hope that takes care of it!