DiabloMiner GPU Miner

[nyargh]

Stalling cores could very well be a "normal" driver bug. It happens to some OpenCL users (isn't just my miner, but requires multiple GPUs), and I'm not sure why. Although almost every report I've heard I think involved using aticonfig to poll temps, so it might be related.

I used to have code that detected the condition, but I can't stop the threads for that GPU and it can lead to needing to reboot the machine because everything locked up if I try to "solve" it.

Interesting...  My cooling situation is handled now so I'm no longer polling temps, but I am polling clock speed and load.  In any case, I get less than a minute of downtime while the script kills and restarts the miner, so I have a solution that works for me.  An added bonus is that it also detects when a pool isn't handing out work and will switch pools when it restarts.

I may run it without any monitoring sometime and see if it fails, would be interesting if that was the cause.

[1714196215]

1714196215

[1714196215]

1714196215

[1714196215]

1714196215

[BOARBEAR]

Is there any way for this miner to work with intel opencl?

If Intel's implementation follows the specification, it should already work fine.

The miner crashes with Intel opencl, have you tried?

[DiabloD3]

Is there any way for this miner to work with intel opencl?

If Intel's implementation follows the specification, it should already work fine.

The miner crashes with Intel opencl, have you tried?

I haven't tried, I don't know anyone who owns brand new Intel CPUs and isn't already mining on Radeons.

This leads me to believe Intel did not follow the specification.

What sort of error?

[enthusiast]

First off let me say thank you to DiabloD3 for putting so much time into this and offering so much help.  What a great service you have provided here!

I first learned of BitCoin about a week ago and I'm hooked and I'm determined to get my miner working.  I've spent hours reading first before posting my question...as any n00b should do.  So here's my question...

I'm getting this error: OpenCL platform AMD Accelerated Parallel Processing contains no devices

I've searched through the thread and found references to "ATI Stream contains no devices" but nothing like what I've pasted here.  The PC I'm running this on has an ATI Radeon 4670 but that doesn't seem to be the issue.

Any help or further documentation to what I currently have would be immensely appreciated!

You need to make sure fglrx functions correctly, and that you've ran cd /; sudo tar zxvf /path/to/AMD-APP-SDK/icd-registration.tgz and that if you're running through ssh that you've made sure X is running as your user and you exported DISPLAY=:0.

My other usual advice doesn't apply to you because the SDK itself seems to be working fine.

I should have said I'm running Windows 7 I suppose...

Ahh, in that case. If you're trying to use SDK 2.4, you need at least Catalyst 11.1 (preferably whatevers newest, which is 11.5b I think)

thanks...I'll try that

[DiabloD3]

Update: Fixed the Windows shits out lots of white space on device name bug, no more 2 line log entries on Windows

[Tril]

Diablo,

First off excellent miner and great work.  Java has far better multi-platform support than python, and the boost libraries that that pycl requires totally suck and are huge.  I hate boost.

Anyhow,  I'm consistently running into a strange issue with my dual 5830 setup using your miner.  After 4-5 hours of mining, one of the cores (random) will just stop.  The miner won't report any errors and the cumulative hashing average will continue to slowly drop.  In fact I get no other errors, can use all the aticonfig tools to manage the cores and simply restarting the miner seems to fix the problem.  The cores are overclocked (temps are ~70C under load) and I poll temps every minute using aticonfig.

Would it be worth adding some code to detect this stall?  Right now I just parse the GPU load output from aticonfig and restart the miners if it drops below a certain threshold.

I'm using the latest drivers and SDK.  I could not get the 2.1 SDK to work for the life of me in my environment, always "No OpenCL devices found" (or something to that effect).  

Also, to anyone else using the 5830: I seem to get best results with: -w 128 -f 1 -v 18


Thanks again!

Stalling cores could very well be a "normal" driver bug. It happens to some OpenCL users (isn't just my miner, but requires multiple GPUs), and I'm not sure why. Although almost every report I've heard I think involved using aticonfig to poll temps, so it might be related.

I used to have code that detected the condition, but I can't stop the threads for that GPU and it can lead to needing to reboot the machine because everything locked up if I try to "solve" it.

I'm having the same issue with random GPU stalls on a dual 6990 system.  I don't remember if I noticed it before I started polling temps with aticonfig, so I turned off the polling to see if it helps.  If not, I would very much like DiabloMiner to go ahead and reboot the machine (or run a script, or just exit) if it can detect the condition.

[jimmyw]

When I run ./DiabloMiner-YourOS.sh -u youruser -p yourpass (with the proper fields completed) I get this error:

Code:

Error: Can't connect to Bitcoin: Connection refused

I'm running Natty on Kubuntu, any ideas on how to solve this?

[Jaime Frontero]

When I run ./DiabloMiner-YourOS.sh -u youruser -p yourpass (with the proper fields completed) I get this error:

Code:

Error: Can't connect to Bitcoin: Connection refused

I'm running Natty on Kubuntu, any ideas on how to solve this?

are you connecting to a pool, or to a local Bitcoin client mining solo?

[jimmyw]

When I run ./DiabloMiner-YourOS.sh -u youruser -p yourpass (with the proper fields completed) I get this error:

Code:

Error: Can't connect to Bitcoin: Connection refused

I'm running Natty on Kubuntu, any ideas on how to solve this?

are you connecting to a pool, or to a local Bitcoin client mining solo?

I'm solo.

[cschmitz]

tried alot of settings, am about 10-20 mh/s slower than phoenix with phatk kernel, sorry, too slow.

[DiabloD3]

tried alot of settings, am about 10-20 mh/s slower than phoenix with phatk kernel, sorry, too slow.

What settings are you using with Phoenix, and what settings did you try with DiabloMiner? Because I've repeatedly proven Phoenix, even with phatk, is damned slow.

[Jaime Frontero]

When I run ./DiabloMiner-YourOS.sh -u youruser -p yourpass (with the proper fields completed) I get this error:

Code:

Error: Can't connect to Bitcoin: Connection refused

I'm running Natty on Kubuntu, any ideas on how to solve this?

are you connecting to a pool, or to a local Bitcoin client mining solo?

I'm solo.

so you run ./bitcoind, and in the ~/./bitcoin/bitcoin.conf file you set rpcuser=username, and rpcpassword=userpassword - and then you use those same credentials in the command line invocation of DiabloMiner (not your operating system credentials)?  with nothing else in the .conf file?

[Tril]

I'm having the same issue with random GPU stalls on a dual 6990 system.  I don't remember if I noticed it before I started polling temps with aticonfig, so I turned off the polling to see if it helps.  If not, I would very much like DiabloMiner to go ahead and reboot the machine (or run a script, or just exit) if it can detect the condition.

It doesn't seem to help to avoid checking temperature.  Random GPUs stall out randomly every few hours.  This might help, though: GPU #2 stopped mining (I could tell by checking the temperature...it was low). Instead of restarting DiabloMiner, like I usually do, this time I tried running a second copy of DiabloMiner with -D 2 to only mine on the stalled GPU.  This new copy got 11Khash/sec, but something amazing happened!  The first DiabloMiner started mining on GPU #2 again (I could tell by hashrate going up)!  I'll try this again next time to see if it's consistent.

[DiabloD3]

So I played around with optimum settings at different clock rates on SDK 2.4

-v  -w   mhash

725/1000
 1  64  243
 2 128  278
 3 192  263
18 192  285
19 192  287 <-- (~= 364 @ 918)
36 192  244

765/1000
 1  192  243
 2  128  291
 3  192  277
18  128  302
19  192  304 <-- (~= 364 @ 918)
36  192  258

918/1000
 1  128  293
 2  128  351
 3  192  329
18   64  361
19  128  364 <--
36  192  308

918/305
 1  256  293
 2  128  350
 3  128  357 <--
18  128  293
19  128  293
36   64  343


This is vs SDK 2.1 doing 367 on 918/305 using -v 2 -w 256

[organofcorti]

So I played around with optimum settings at different clock rates on SDK 2.4

...and I thought, 'great idea!

So here's my table of hashrate versus changes in v and w for a stock 6990 on DiabloMiner (based on 20mins to 2 hours of averaging):

                           
v     w       hashrate  % invalid GPU1 temp GPU2 temp fan speed
1      256      657015      0.7%         74.0         78.0         3905
1      128      659598      0.9%         74.0         77.5         3931
1      64      658721      2.7%         74.5         78.0         3940
2      256      589505      1.6%         73.0         77.0         3880
2      128      653134      0.0%         75.5         79.0         3980
2      64      661726      0.0%         75.5         79.0         3980
3      256      649722      1.1%         76.0         80.0         4070
3      128      661191      2.4%         76.0         80.0         4080
3      64      570522      0.0%         73.0         77.0         3880
4      256      518821      2.3%         70.5         74.5         3670
4      128      523139      9.1%         70.5         74.5         3700
4      64      660914      0.6%         76.0         79.5         3959
5      256      518460      2.0%         70.5         74.0         3670
5      128      522742      3.1%         70.0         73.5         3670
5      64      490119      3.4%         69.5         73.0         3620
6      256      522556      5.6%         70.0         73.5         3670
6      128      525496      2.4%         70.0         74.0         3670
6      64      525356      0.0%         70.5         74.0         3670
18      256      505655      1.6%         69.0         72.5         3600
18      128      511409      0.8%         69.0         73.0         3600
18      64      584084      0.0%         72.0         75.5         3770
19      256      498097      1.7%         69.0         71.0         3606
19      128      512743      1.2%         69.2         71.8         3628
19      64      587775      4.0%         72.0         75.5         3746
20      256      133321      1.4%         65.5         68.8         3294
20      128      131673      1.9%         65.7         68.7         3346
20      64      134202      0.0%         65.1         69.0         3359
21      256      143255      0.0%         66.2         69.0         3348
21      128      142799      1.2%         66.1         69.6         3358
21      64      143776      2.2%         66.5         69.7         3370


So default flags (-v1 -w256) works pretty well, and v1 w128 and w256 have the lowest temps of the settings that will give the highest rate. It's a shame I can't post my chart porn since I have a really nice density distribution plots showing bands of hashrate at: 660000, 582000, 520000, 495000, 132000. Odd. Nearly multiples of 11.

Also, the second GPU is consistently higher temp than the first one - dud core?

Cheers for the miner - that averager is the only way you can really get an accurate idea of what changing the flags actually does.

Next: changing clock speeds for the highest scoring flags, then finally changing -f.            

[DiabloD3]

Update: I improved SDK 2.4 performance and fixed emails in --url (use \\@ or +++++ instead of @)

Also, organofcorti, I just made your chart of doom outdated. 69xx should, theoretically, perform better with the new update I just pushed.

[Landci]

hello i am trying to run the DiabloMiner on my MAC with localhost,
but how is this working?
why do i have to put in a username and password when i am doing localhost?

do i need to configure the bitcoin programm?

thanks

[Jaime Frontero]

hello i am trying to run the DiabloMiner on my MAC with localhost,
but how is this working?
why do i have to put in a username and password when i am doing localhost?

do i need to configure the bitcoin programm?

thanks

try here:

https://en.bitcoin.it/wiki/Running_Bitcoin#Bitcoin.conf_Configuration_File

[Landci]

i cant access the wiki, it always tells me the 403 Forbidden html error...

[Jaime Frontero]

i cant access the wiki, it always tells me the 403 Forbidden html error...

eh.  they're probably screwing with the server.  it'll be back up by and by.  it's the best resource for what you want.

mining is a game of patience...