hey guys.. need some advice. I am running my rig headless on ubuntu. I have three miners running in a separate screen session (one for each GPU). Sometimes a miner will hang up for nefarious reasons. I can still SSH into the machine, and work and the other 2 miners will run as normal. BUT... if I try to do a kill -15 or even a kill -9 on the PID of the froze miner, the entire system will hang and I cant do anything. Even if I plug a monitor and keyboard back up it will still be froze.
The same thing happens if I try to reboot the machine from CLI when a mining script is hung. Whats the best way to unstick this froze process without killing the entire system?
Is there a way to detach the particular module for the hung GPU from the kernel without killing the system?
Topic: Ubuntu Natty Narwhal 11.04 Mining Guide / HOWTO - page 18. (Read 281467 times)
I have an M4A89GTD mobo and two AMD Sapphire 5830s.
aticonfig --list-adapters
* 0. 01:00.0 ATI Radeon HD 5800 Series
1. 02:00.0 ATI Radeon HD 5800 Series
aticonfig --adapter=0 --od-gettemperature
Adapter 0 - ATI Radeon HD 5800 Series
Sensor 0: Temperature - 37.00 C
aticonfig --adapter=1 --od-gettemperature
ERROR - Get temperature failed for Adapter 1 - ATI Radeon HD 5800 Series
I switched the slots for the 5830s and get the same thing. I The drivers don't seem to be working. I've re-installed Ubuntu a couple times and get the same error.
aticonfig --list-adapters
* 0. 01:00.0 ATI Radeon HD 5800 Series
1. 02:00.0 ATI Radeon HD 5800 Series
aticonfig --adapter=0 --od-gettemperature
Adapter 0 - ATI Radeon HD 5800 Series
Sensor 0: Temperature - 37.00 C
aticonfig --adapter=1 --od-gettemperature
ERROR - Get temperature failed for Adapter 1 - ATI Radeon HD 5800 Series
I switched the slots for the 5830s and get the same thing. I The drivers don't seem to be working. I've re-installed Ubuntu a couple times and get the same error.
I have only one 6990 card on board, but when I use aticonfig --list-adapters, I got:
* 0. 08:00.0 AMD Radeon HD 6990
1. 07:00.0 AMD Radeon HD 6990
and I run the miner:
./poclbm -u ** --pass ** -o ***... -d 0
this miner give me about 330 mH/s rate
but when I try to launch another miner, using device 1, I got error:
pyopencl.LogicError: clGetPlatformIDs failed: invalid/unknown error code
any suggestions? thanks.
You will want to use screen.
I create two shell scripts, the first one is to start the miners for whatever pool I want to join and has my poclbm.py command line in it.
The second script is the script that starts all the miner script(s) that I created to start poclbm.
I then use screen like this:
screen -d -m -S gpu1 ./miner.sh
This starts a screen session named "gpu1" with the miner.sh script.
Then, to reconnect to that session, I use screen -r gpu1
To detach from that session, I use CTRL-A D
The miner will stay running even if you log out and when you log back in you can re-attach to that screen if you want, with screen -r gpu1
* 0. 08:00.0 AMD Radeon HD 6990
1. 07:00.0 AMD Radeon HD 6990
and I run the miner:
./poclbm -u ** --pass ** -o ***... -d 0
this miner give me about 330 mH/s rate
but when I try to launch another miner, using device 1, I got error:
pyopencl.LogicError: clGetPlatformIDs failed: invalid/unknown error code
any suggestions? thanks.
How do I start one miner, then start the second miner, then the third, and then able to switch back to the first to check the progress?
Also, the miner should not quit when I exit the shell or SSH.
Help please, thank you guys
Also, the miner should not quit when I exit the shell or SSH.
Help please, thank you guys
You will want to use screen.
I create two shell scripts, the first one is to start the miners for whatever pool I want to join and has my poclbm.py command line in it.
The second script is the script that starts all the miner script(s) that I created to start poclbm.
I then use screen like this:
screen -d -m -S gpu1 ./miner.sh
This starts a screen session named "gpu1" with the miner.sh script.
Then, to reconnect to that session, I use screen -r gpu1
To detach from that session, I use CTRL-A D
The miner will stay running even if you log out and when you log back in you can re-attach to that screen if you want, with screen -r gpu1
I'm getting the following error:
Traceback (most recent call last):
File "./poclbm.py", line 3, in
import pyopencl as cl
ImportError: No module named pyopencl
When I try to run either miner. I've tried repeating the steps several times but can't get anything working. Just wondering if this is a common problem where it might be something obvious?
Edit: Got it working. I had originally messed up one of the lines and redoing the steps couldn't fix it (unless there were instructions in the code whizzing by that I didn't read). Anyway I did a fresh install and it seems to be ok now.
Traceback (most recent call last):
File "./poclbm.py", line 3, in
import pyopencl as cl
ImportError: No module named pyopencl
When I try to run either miner. I've tried repeating the steps several times but can't get anything working. Just wondering if this is a common problem where it might be something obvious?
Edit: Got it working. I had originally messed up one of the lines and redoing the steps couldn't fix it (unless there were instructions in the code whizzing by that I didn't read). Anyway I did a fresh install and it seems to be ok now.
How do I start one miner, then start the second miner, then the third, and then able to switch back to the first to check the progress?
Also, the miner should not quit when I exit the shell or SSH.
Help please, thank you guys
Also, the miner should not quit when I exit the shell or SSH.
Help please, thank you guys
You will want to use screen.
I create two shell scripts, the first one is to start the miners for whatever pool I want to join and has my poclbm.py command line in it.
The second script is the script that starts all the miner script(s) that I created to start poclbm.
I then use screen like this:
screen -d -m -S gpu1 ./miner.sh
This starts a screen session named "gpu1" with the miner.sh script.
Then, to reconnect to that session, I use screen -r gpu1
To detach from that session, I use CTRL-A D
The miner will stay running even if you log out and when you log back in you can re-attach to that screen if you want, with screen -r gpu1
Ok, I tried to install on a fresh Natty Narwhal 64bit USB instance (created with Universal-USB-Installer-1.8.4.7.exe)
Fails on step 4 - just says screen is terminating.
When I run without screen it says - E: unable to locate libboost-all-dev
Fails on step 4 - just says screen is terminating.
When I run without screen it says - E: unable to locate libboost-all-dev
Thx Inaba, I frequently encountered this problem as I was also using LiveCD installation. You'll have to go to System Settings > Software/Update Manager > Open Source > Check the 'Universe' thing.
How do I start one miner, then start the second miner, then the third, and then able to switch back to the first to check the progress?
Also, the miner should not quit when I exit the shell or SSH.
Help please, thank you guys
Also, the miner should not quit when I exit the shell or SSH.
Help please, thank you guys
how big should the usb stick be?
4 Gig would probably be sufficient, but 8 GB probably better.
I use 8 GB Patriot Rage XT sticks (http://www.amazon.com/gp/product/B003WUX6PG)
I guess I will need 8GB then.
What is the best partition for a 8GB USB?
how big should the usb stick be?
4 Gig would probably be sufficient, but 8 GB probably better.
I use 8 GB Patriot Rage XT sticks (http://www.amazon.com/gp/product/B003WUX6PG)
i followed this guide and at first i got some problems, but later when i tried again after fresh reinstall, i didnt get any problems.
but i have tow question. i have installed SSH, how do i connect to my mining rig with SSh on my main computer? is it possible to start mining with SSH when the mining rig restarts?
but i have tow question. i have installed SSH, how do i connect to my mining rig with SSh on my main computer? is it possible to start mining with SSH when the mining rig restarts?
You need an SSH client, such as PuTTY. I use SecureCRT but it costs money, PuTTY is free. To answer your question, yes, you can start mining with SSH, that's what I do. None of my mining boxes have monitors on them (or dummy plugs).
thats great, i was worried to have monitor, keyboard and mouse in the kitchen hehe. i have used putty, but i used it on a server i owned a year ago, i just inserted the IP for my server and i could see the desktop and connect. how do i conned to linux when the machine is in LAN network? is there a guide?
how big should the usb stick be?
4 Gig would probably be sufficient, but 8 GB probably better.
I use 8 GB Patriot Rage XT sticks (http://www.amazon.com/gp/product/B003WUX6PG)
i followed this guide and at first i got some problems, but later when i tried again after fresh reinstall, i didnt get any problems.
but i have tow question. i have installed SSH, how do i connect to my mining rig with SSh on my main computer? is it possible to start mining with SSH when the mining rig restarts?
but i have tow question. i have installed SSH, how do i connect to my mining rig with SSh on my main computer? is it possible to start mining with SSH when the mining rig restarts?
You need an SSH client, such as PuTTY. I use SecureCRT but it costs money, PuTTY is free. To answer your question, yes, you can start mining with SSH, that's what I do. None of my mining boxes have monitors on them (or dummy plugs).
i followed this guide and at first i got some problems, but later when i tried again after fresh reinstall, i didnt get any problems.
but i have tow question. i have installed SSH, how do i connect to my mining rig with SSh on my main computer? is it possible to start mining with SSH when the mining rig restarts?
but i have tow question. i have installed SSH, how do i connect to my mining rig with SSh on my main computer? is it possible to start mining with SSH when the mining rig restarts?
how big should the usb stick be?
I'm going to be using this tutorial later tonight. Just to confirm - this is using the *Desktop* version of Natty, correct? (and NOT the server version?)
I assumed that the Desktop version was required in order to take advantage of Xserver and such. (needed for GPU activities)
I assumed that the Desktop version was required in order to take advantage of Xserver and such. (needed for GPU activities)
That's correct, use the desktop version.
Ok, I tried to install on a fresh Natty Narwhal 64bit USB instance (created with Universal-USB-Installer-1.8.4.7.exe)
Fails on step 4 - just says screen is terminating.
When I run without screen it says - E: unable to locate libboost-all-dev
Fails on step 4 - just says screen is terminating.
When I run without screen it says - E: unable to locate libboost-all-dev
Nowhere in the guide does it say to use "Universal-USB-Installer," whatever that is. I just finished installing Natty on another USB stick on a different machine. Zero Problems.
FOLLOW THE GUIDE.
I'm going to be using this tutorial later tonight. Just to confirm - this is using the *Desktop* version of Natty, correct? (and NOT the server version?)
I assumed that the Desktop version was required in order to take advantage of Xserver and such. (needed for GPU activities)
I assumed that the Desktop version was required in order to take advantage of Xserver and such. (needed for GPU activities)
Ok, I tried to install on a fresh Natty Narwhal 64bit USB instance (created with Universal-USB-Installer-1.8.4.7.exe)
Fails on step 4 - just says screen is terminating.
When I run without screen it says - E: unable to locate libboost-all-dev
Fails on step 4 - just says screen is terminating.
When I run without screen it says - E: unable to locate libboost-all-dev
step 25 failed with:
/bin/sh: ctags: not found
fixed by doing:
sudo apt-get install exuberant-ctags
/bin/sh: ctags: not found
fixed by doing:
sudo apt-get install exuberant-ctags
Glad it's working for you, but I feel the need to reiterate the fact that if you are installing extra packages because of a problem and it's not listed in the guide, something is wrong elsewhere and installing extra packages is a band aid to the problem, not a solution. The guide as written works if the steps are followed exactly. Installing extra packages or making changes not in the guide indicate that at some point there was a deviation from the guide.
Persistence setting is only for a live installation. I don't see any reason why you'd create an install like that, especially on a USB stick.
Hi Inaba, so what u're saying is that I could install the full OS onto the USB stick. Sorry cuz when I searched for USB installation the first thing I got was on LiveCDs, didn't know u can do that.
I'll try it out today on my new prototype rigYes, that's what I'm saying. You may have to change a setting in your bios so that your motherboard recognizes a USB stick as a hard drive and not a floppy, but all my motherboards have this as a default and I did not have to adjust the setting.
Persistence setting is only for a live installation. I don't see any reason why you'd create an install like that, especially on a USB stick.
Hi Inaba, so what u're saying is that I could install the full OS onto the USB stick. Sorry cuz when I searched for USB installation the first thing I got was on LiveCDs, didn't know u can do that.
I'll try it out today on my new prototype rig
step 25 failed with:
/bin/sh: ctags: not found
fixed by doing:
sudo apt-get install exuberant-ctags
/bin/sh: ctags: not found
fixed by doing:
sudo apt-get install exuberant-ctags
Found a typo in the 2.4 guide on step 22 for the 32-bit version:
32-bit 22. ./configure.py --cl-inc-dir=/opt/AMD-APP-SDK-v2.4-lnx6432/include/ --cl-lib-dir=/opt/AMD-APP-SDK-v2.4-lnx32/lib/x86
the highlighted 64 should be removed.
I was having the same problem as Schizo during the make -j3 step:
src/wrapper/wrap_cl.hpp:20:19: fatal error: CL/cl.h: No such file or directory
compilation terminated.
error: command 'gcc' failed with exit status 1
make: *** [all] Error 1
After fixing the typo, this error went away, and now everything seems to be working great!
32-bit 22. ./configure.py --cl-inc-dir=/opt/AMD-APP-SDK-v2.4-lnx6432/include/ --cl-lib-dir=/opt/AMD-APP-SDK-v2.4-lnx32/lib/x86
the highlighted 64 should be removed.
I was having the same problem as Schizo during the make -j3 step:
src/wrapper/wrap_cl.hpp:20:19: fatal error: CL/cl.h: No such file or directory
compilation terminated.
error: command 'gcc' failed with exit status 1
make: *** [all] Error 1
After fixing the typo, this error went away, and now everything seems to be working great!
I've got all 4 of my miners up and running on my 6990's no problem. The ati-config commands work great. (I do have to sudo for all of them, but that's fine)
Now, I'd like to use AMDOverdriveCtrl to set my memory clock below what the aticonfig program allows.
I've downloaded & installed the package and am running it from another linux machine with: ssh -X remotebox AMDOverdriveCtrl
It's telling me 'This program depends on the proprietary Catalyst drivers. Please check the driver installation.'
Do I really want to download the 11.5 drivers and install them or will this hose everything else?
thanks!
Now, I'd like to use AMDOverdriveCtrl to set my memory clock below what the aticonfig program allows.
I've downloaded & installed the package and am running it from another linux machine with: ssh -X remotebox AMDOverdriveCtrl
It's telling me 'This program depends on the proprietary Catalyst drivers. Please check the driver installation.'
Do I really want to download the 11.5 drivers and install them or will this hose everything else?
thanks!
Awesome, anyway I think in this guide you've already got the files required to get the mem clock down to where AMDOverdriveCtrl can. So you could essentially do the clocking using ATIconfig. Except, I hasn't got to that part just yet.
Not so much - the aticonfig only allows lowering the memory clock rate to 1250 - amdoverdrivectrl will let me take it down to 625... i may clone this install for safe-keeping and do another 'different' setup to see what happens...
Jump to: