Black screen on a 2x6990 rig - Please help

[flower]

Hello, beautiful people of Bitcoin.org.

I come to you for help regarding the problems I've been experiencing with my rig:

2x HD6990 (Asus)
Asus Crosshair formula IV
Corsair 1200W PSU
OS: Lubuntu 11.04 (Ubuntu with an LXDE desktop)
I am using poclbm as the miner.

I have read through and through all the Wiki, the forum and many obscure google searches and have not yet been able to find a solution.

Let me explain.

Before I bought the second 6990, it worked perfectly.
I was able to follow the brilliant guides on this forum, and set up a rig that produces about 1.5 BTC/day, so I went ahead and bought another 6990.
Thats when the problem started.
When I plugged the second 6990 to the motherboard (I didn't crossfire, obviously), Linux was not booting to desktop. I was getting just a black screen, and was unable to do anything except a hard reset to the computer.
I went ahead and started again, this time in recovery mode. I tried starting X server on failsafe, but that didn't work either, so I went ahead and deleted whichever xorg.conf files I had in /etc/X11/  .
This time the system did start, but since the xorg.conf file was gone, poclbm wasn't able to find and use my two 6990 for mining (it only saw the processor).
At this stage I ran "sudo aticonfig --initial -f --adapter=all" (with or without the -f, it didn't matter), and sure enough, xorg.conf was generated in all its beauty. (I did not yet reboot at this point.)
Trying to run "./poclbm.py" I saw that it was able to see the two Cayman cores (0 and 1) on the first 6990, but not the second one.
I rebooted the machine and was greeted by that same black screen from before.

Currently I have tried, numerous times, to install different ATI drivers. Both from ati.com and the repo's.
I have tried to find some reference as to X server making these problems with dual 6990 on ubuntu, but found nothing similar.

Running "aticonfig --list-adapters" does give me a list of all four 6990 cores on the system.
Running "lspci", I can see that the PCI bus given to the 6990s is the same as the one written in xorg.conf

At this point I'm stuck. The only way I'm able to log back into linux is if I delete the xorg.conf and start fresh.

Anyone here with two or three 6990, and mining on linux that can help?

Thanks for reading this far.

If there's any more info you need, let me know.

[flower]

With respect,

Can any of the mods help me out here by moving this thread to the technical support forum?

[Wynand]

cd to /etc/X11, re-enter the aticonfig command and chmod the /etc/X11/xorg.conf file to 644.
Easiest way to prevent that black screen is by logging in using SSH and executing the required commands, as you indeed won't have any graphics after this.

If problems continue to exist, check the slot configuration for two cards in the manual, the required slotconfig is a bit silly on that mainboard. :-)
(card 1 in slot 1, 1 slot empty, card 2 in slot 3, although you couldn't even put 2x6990 next to each other..

[Obsi]

Do you have 2 monitors hooked up or just 1?

[shat]

Hey flower, one thing I noticed a long time ago with Ubuntu 11.04 at least is that at times, with numerous cards, depending on board, you just see a black screen and the monitor will goto power save mode.  Push ALT-F1, SHIFT-ALT-F1, CTRL-ALT-F1.  Usually alt-f1 will work, but it is just wonkiness with the graphics display xvga settings in Grub.

You should get a screen.  You can test if the box is truly locked up by once the black screen happens, push your caps lock key, if light no keyboard lights up, you're not locked up* so to speak.

Try the method I mention above and just push ALT-F1 or other combination(s) a few times and see what happens

[Wynand]

@shat: I tried changing inits on those mainboards, doesn't work. The only way in is SSHing to it.

[coini]

Hi,

did you finally find a solution for you black screen problem on you minig rig? I have this problem, too.  It drives my crasy. I am searching for a solution since 2 days now, I am nearly about to use windows.

[drawoc]

coini, could you post your xorg.log here?
(The following steps assume you use Ubuntu or Debian)
After you receive the black screen, hit Ctrl-Alt-F1. Login to the terminal. Type:

Code:

sudo apt-get install pastebinit
pastebinit /var/log/Xorg.0.log

The first command will probably ask you to enter your password.
The second will output a URL. Could you post the URL here? Thanks.

[aerion]

Hi,

I am experiencing similar behaviour with two ATI HD6990 network cards, tested in two different motherboards with the same result. The machine will begin to boot, than when it attempts to boot the gui, the machine simply locks up with nothing on the screen, and requires a hard power reset to reboot it.

The machine is running Ubuntu 11.04 32bit mobo is Gigabyte Z68A-D3-B3, I also followed Inaba's Ubuntu 11.04 howto for 6xxx series video cards.

I started the machine with gdm disabled so that I could boot it up, than ran a tail -f on all the important logs (syslog,dmesg,Xorg.0.log, Xorg.1.log). I than started gdm, and only syslog outputed something before the machine locked up. I have copied it to pastebin, link below.

http://pastebin.com/qWHtPa9U

Would appreciate if anyone could provide any sugestions on investigating this further.

[drawoc]

Aerion, try running:

Code:

lspci

Are your cards listed there? If not, It's a hardware problem.

[aerion]

Yeah they are there.

I decided to try Debian Wheezy, and I am experiencing similar lock ups.

What I have tried next is uninstalled the Debian packaged fglrx driver, and downloaded the latest ATI proprietary driver and installed that instead, and this appears to have fixed the problem

However I now have another problem When SSHing to the machine and trying to run commands like:

export DISPLAY=:0; aticonfig --odgt --adapter=all
No protocol specified
aticonfig: This program must be run as root when no X server is active

I have made sure that I am logged into Gnome. Also running the same command via an Xterminal works fine.

[drawoc]

Maybe it's not on display 0? From the xterm, run:

Code:

echo $DISPLAY

and see what it outputs. If it doesn't output :0, try using whatever value it does output for running it over ssh.

[aerion]

ran echo $DISPLAY from an xterm, and found that its actually :0.0.

however the command via SSH still yields the same

export DISPLAY=:0.0; aticonfig --odgt --adapter=all

[drawoc]

Are you logged in as a different user with ssh as you are in X? If not, you need to be using the same user for both.

[Nivco]

Hey there !
It seems I'm in same sh*thole 

My rig is 4x 5870s and fresh Ubuntu Natty.
I followed Inaba's HowTo.
Everything went fine (from steps 1 to 31).
On step #32 ($ sudo reboot), the rig reboot but won't load Ubuntu GUI (obviously an issue with X). Instead I got the Ubuntu logo and 5 red dots and automatic reboot or simple frozen black screen. If the latter appears then I still can access terminal with Ctrl+Alt+F1. The system asks for login+password. Once looged I try to start phoenix which only detect my CPU.
$ fglrxinfo returns
"Error: unable to open display (null)"
$ startx gives me a black screen again.

Help please !

[drawoc]

My rig is 4x 5870s and fresh Ubuntu Natty.
I followed Inaba's HowTo.
Everything went fine (from steps 1 to 31).
On step #32 ($ sudo reboot), the rig reboot but won't load Ubuntu GUI (obviously an issue with X). Instead I got the Ubuntu logo and 5 red dots and automatic reboot or simple frozen black screen. If the latter appears then I still can access terminal with Ctrl+Alt+F1. The system asks for login+password. Once looged I try to start phoenix which only detect my CPU.
$ fglrxinfo returns
"Error: unable to open display (null)"
$ startx gives me a black screen again.

It seemed to help aeron when he installed fglrx from ATI instead of from the ubuntu package repo, that might help.

[Nivco]

I've just found the issue (after 3 days of troubleshooting) : I used a faulty PCI-e 16x extender that crashed the rig. Yes, it was a cheap eBay cable from KongKong BUT the other cables work (PCI-e 1x extender modded by me... which is basically cutting a little end of plastic) ! 
Thanks for the help I received from all members of this great forum and other fellow mining adventurers !

[aerion]

Are you logged in as a different user with ssh as you are in X? If not, you need to be using the same user for both.

I made sure I am logged in as the same user both in Gnome and SSH. I ended up just giving up on Debian for now and installing Ubuntu 11.04 and ATI Proprietary driver instead, and I am having no issues

[drawoc]

Are you logged in as a different user with ssh as you are in X? If not, you need to be using the same user for both.

I made sure I am logged in as the same user both in Gnome and SSH. I ended up just giving up on Debian for now and installing Ubuntu 11.04 and ATI Proprietary driver instead, and I am having no issues

Glad you fixed your issue.