python OpenCL bitcoin miner

[molecular]

ImportError: No module named pyopencl[/pre]

Could anyone help me?

You need to install pyopencl.

[drgr33n]

I'm sorry if this has been asked but I've been fooling around with this all day.  Basically I have a machine I run slackware on...yes yes not your standard distro.  I built all the necessary packages installing stream 2.1 sdk, pyopencl, rpc, etc.  Everything compiled fine and runs without a hitch.  I'd like to run poclbm in INIT 3 vs running it in INIT 4 so im not sharing resources with kde.   Everytime i try to run poclbm in init 3 it tells me ati needs x windows.  I never had this issue with nvidia which i can start poclbm in init 3, i took the same script and ran it on this box and basically it cant find the gpus unless x is running.  Anyone found a way to bypass this?  I exported my library and the display in the begining of the script.

Create an .xinitrc that just runs a window manager (XMonad or something lightweight) and startx from init 2.

Thanks Ill try that I tried making a script to modprobe the driver in init 3(default multiuser in slackware) with no success. Ill try make a xinitrc and run fluxbox maybe DAMN YOU ATI WHY DO YOUR DRIVERS SUCK!   GRRR lol thanks everyone for the input if i get it working ill post the scripts on here so someone else can use them.  I don't think posting a howto for slackware is necessary as if you use slackware you can probably already figure it out as its quite similar to ubuntu or any other distro.   

Sucks don't it !! Nvidia have a kernel module that the GPGPU as a resource without xorg but ATI needs xorg to function. Is there any reason in particular that you would want to run this without xorg ?

If it's a speed issue you can just get GDM to load and and ssh  into the machine to start mining ?   

[marcus_of_augustus]

No it's stability issue. Since Xorg is always querying the default display adapter GPU, the system will hang more often if you load up with mining the or OC too much the default GPU core. Basically, there will always be one core that is not a dedicated compute node, the default display adapter, since Xorg is always sticking its nose in and talking with that core.

Maybe a simple hacky hardware work around it is to get a cheap lightweight graphics card that can run as the default display (or better still a mobo with an on-board GPU) and then Xorg will leave alone the compute GPUs and you can load them right-up and OC to the max. It doesn't matter if you run as headless since as long as Xorg is running (which you need it to be to load the drivers) then it will be talking unnecessarily with that default display adapter.

There was some discussion on AMDs OpenCL forum for a way to set the default display to none or some such in xorg.conf but I never had any success with it ...

[d3c0n808]

I ended up activating my onboard gpu and using that to load kde.  With the onboard being used as the default device I have no problem using -f 0 on the 5870s also the overclocks are more steady the only thing i need to worry about is the temp.  When i was using one of the cards as the display card one or both would crash after a few hours due to the overclock.   This may not be advisable to most but my hash rate when up about 10mhash percard with the higher oc and f0 flag

[marcus_of_augustus]

I ended up activating my onboard gpu and using that to load kde.  With the onboard being used as the default device I have no problem using -f 0 on the 5870s also the overclocks are more steady the only thing i need to worry about is the temp.  When i was using one of the cards as the display card one or both would crash after a few hours due to the overclock.   This may not be advisable to most but my hash rate when up about 10mhash percard with the higher oc and f0 flag

Thanx for confirming that for me ... i suspected it would be the case since i have found not max-loading up the default display adapter with mining reduces number of kernel hangs markedly ... now need to get hold of cheap mobo with onboard graphics chip and 4 or more PCIe slots ....

[Cryptoman]

Thanx for confirming that for me ... i suspected it would be the case since i have found not max-loading up the default display adapter with mining reduces number of kernel hangs markedly ... now need to get hold of cheap mobo with onboard graphics chip and 4 or more PCIe slots ....

Make sure the onboard graphics is ATI and not nvidia.  The ATI and nvidia drivers do not play well together.  Already been down that path myself.

[marcus_of_augustus]

Thanx for confirming that for me ... i suspected it would be the case since i have found not max-loading up the default display adapter with mining reduces number of kernel hangs markedly ... now need to get hold of cheap mobo with onboard graphics chip and 4 or more PCIe slots ....

Make sure the onboard graphics is ATI and not nvidia.  The ATI and nvidia drivers do not play well together.  Already been down that path myself.

Thnx, good point ... amd all the way.

[d3c0n808]

Indeed Im running a asus m3a78-t with 3800 hd onboard gpu so I've had no problem the only issue was i had to write a new xorg config to send the display to the onboard gpu but other than that it was easy to implement.  My hash rate on both 5870s went from about 400mhash to 420 mhash by chaning the f flag and also switching to onboard gpu.   Also as I said earlier much more stable without increasing voltage as well.  Also when trying to resume from sleep screen while using one of the video cards always caused the system to lock up where as now with the onboard gpu active i can let the monitor go to sleep and not have to worry about the whole system crashing when trying to resume.

[saadtariq30]

can it be done using onboard intel? ive a 6950 unlocked and clocked at 950mhz..

[rage911]

what is optimal settings for HD4850?

[Scrappi]

I registered with Bitcoin and downloaded it to this computer.   I also signed up for a pool.   From what I have read about using it you need to make a special folder.  It also looked like some kind of coding is used to set it up.  I am internet and computer illiterate and have no idea how to do anything like that.   How do I complete setting it up?   

[d3c0n808]

can it be done using onboard intel? ive a 6950 unlocked and clocked at 950mhz..

I want to say yes it could be possible but don't hold me to it. 

[saadtariq30]

can it be done using onboard intel? ive a 6950 unlocked and clocked at 950mhz..

I want to say yes it could be possible but don't hold me to it. 

ok i tried..when i choose dual mode from the bios, i have to connect the monitor to my onboard vga port...in windows device manager both intel gm4500 and radeon 6900 series show up..but catalyst says the card installed is not supported..and it doesnt show up in the miner as well...when i disable intel gm4500 from the device manager, i have to connect the monitor to the radeon and everything starts working again...what i want to do is use the onboard for display and the radeon for mining..so that i can mine with -v -w 128 -f 1 without inducing any lag...suggestions?

[d3c0n808]

can it be done using onboard intel? ive a 6950 unlocked and clocked at 950mhz..

I want to say yes it could be possible but don't hold me to it. 

ok i tried..when i choose dual mode from the bios, i have to connect the monitor to my onboard vga port...in windows device manager both intel gm4500 and radeon 6900 series show up..but catalyst says the card installed is not supported..and it doesnt show up in the miner as well...when i disable intel gm4500 from the device manager, i have to connect the monitor to the radeon and everything starts working again...what i want to do is use the onboard for display and the radeon for mining..so that i can mine with -v -w 128 -f 1 without inducing any lag...suggestions?

Ok since your using windows you may need a dummy plug in order for catalyst to see the gpu.   You probably wont be able to use the onboard to process bitcoins as I don't believe intel supports open cl.  Windows is problematic because you have to use dummy plugs in order for the drivers to activate the card where in linux this is not necessary as we just export the display.  if this is a dedicated miner i would consider following the ubuntu guide as it is well written and quite easy to setup and you will have much more stable hash rates in linux then windows as well not to mention you wont have to make a dummy plug.

[saadtariq30]

can it be done using onboard intel? ive a 6950 unlocked and clocked at 950mhz..

I want to say yes it could be possible but don't hold me to it. 

ok i tried..when i choose dual mode from the bios, i have to connect the monitor to my onboard vga port...in windows device manager both intel gm4500 and radeon 6900 series show up..but catalyst says the card installed is not supported..and it doesnt show up in the miner as well...when i disable intel gm4500 from the device manager, i have to connect the monitor to the radeon and everything starts working again...what i want to do is use the onboard for display and the radeon for mining..so that i can mine with -v -w 128 -f 1 without inducing any lag...suggestions?

Ok since your using windows you may need a dummy plug in order for catalyst to see the gpu.   You probably wont be able to use the onboard to process bitcoins as I don't believe intel supports open cl.  Windows is problematic because you have to use dummy plugs in order for the drivers to activate the card where in linux this is not necessary as we just export the display.  if this is a dedicated miner i would consider following the ubuntu guide as it is well written and quite easy to setup and you will have much more stable hash rates in linux then windows as well not to mention you wont have to make a dummy plug.

thanks it worked...i made a dummy plug..the only reason i did this was so that that pc wouldnt lag with f 1...hash rates up by 15M/hashs..and i can watch movies/youtube over the onboard chip..no lag at all ...thanks man

[jkminkov]

thanks it worked...i made a dummy plug..the only reason i did this was so that that pc wouldnt lag with f 1...hash rates up by 15M/hashs..and i can watch movies/youtube over the onboard chip..no lag at all ...thanks man

no lag?? just press right mouse button - any program or desktop

[FireFox31]

Hello, I recently upgraded from poclbm_20110325 to poclbm_20110428 and I'm now receiving "long poll exceptions" once every hour:

10/05/2011 14:05:19, long poll exception:
Traceback (most recent call last):
  File "BitcoinMiner.pyo", line 259, in longPollThread
  File "BitcoinMiner.pyo", line 222, in request
  File "httplib.pyo", line 974, in getresponse
  File "httplib.pyo", line 391, in begin
  File "httplib.pyo", line 349, in _read_status
  File "socket.pyo", line 397, in readline
timeout: timed out

On the older version, I had no problems, though I was generating about 4% stale shares.  I hoped the newer version would reduce stale shares.  I'm running Windows XP, a GTX460, and poclbm on GPU mining.  Is this a problem?  Is there a fix?  Thanks for any help you can provide.

[Pygy]

I didn't read the whole thread, so, please forgive me if this problem has already been addressed (in that case, I'll look for the solution by myself).

I'm mining on a Debian box with a 5970. I'm prudently trying to overclock it, but I run into a problem.

The hashrate in the shell of the second core sometimes gets stuck. CTRL+C is ineffective to quit (I get the "Bye" message, but I dont go back to the prompt).

If I close the shell and confirm that I want to kill the child process, the whole graphical interface freezes. It happened twice at two different clock rates (800 and 780) even though the temperature monitors reported ~66 °C. The second time I actually tried to kill the process from the shell.

At the time, I didn't think of pressing CTRL+ALT+F1 to launch a shell outside of X, and rebooted.

I went back to a lower clock rate, and everything is fine, but I'm curious about the cause of the freezing of the process, and the of the whole GUI.

[rtsome]

i've got a bug to report.
so there was an exception

Code:

14/05/2011 22:52:28, 36a8d1c6, accepted                     
14/05/2011 22:52:30, dfd608d5, accepted                     
14/05/2011 22:53:35, cd65a2af, accepted       
53409 khash/sException in thread Thread-2:                  
Traceback (most recent call last):
  File "/usr/lib/python2.6/threading.py", line 532, in __bootstrap_inner
    self.run()
  File "/usr/lib/python2.6/threading.py", line 484, in run
    self.__target(*self.__args, **self.__kwargs)
  File "/home/diz/Desktop/bitcoinshit/poclbm/BitcoinMiner.py", line 336, in miningThread
    queue.finish()
LogicError: clFinish failed: invalid command queue

14/05/2011 22:55:35, long poll: new block 00003873116dcf9f  
14/05/2011 22:58:28, long poll: new block 000016bc6ee42ac4  
14/05/2011 22:59:41, long poll: new block 0000071527b5e52e  
14/05/2011 23:00:13, long poll: new block 0000594949d6ec0d  

and mining stopped, had to restart it.
probably relevant dmesgs:

Code:

[30953.851867] NVRM: Xid (0000:01:00): 3, C 00000004 SC 00000000 M 00001010 Data 00000000
[30955.858750] NVRM: os_schedule: Attempted to yield the CPU while in atomic or interrupt context

poclbm commit 06cf386a358a416b4119482bdfaf0d4511ac91b7
ubuntu 10.10@amd64; NVIDIA-Linux-x86_64-270.41.06; GTX 280

[sniper_sniperson]

If I put & symbol in the password after --pass option the script says Bad password. Should I try the password with quotes?