I used to mine with phoenix on Eligius. The hash rate reported by both of them did agree (in average). Now I am trying Diablo. This miner reports much higher hash rate (by 15%). However, Eligius reports only 53% of the hash rate reported by Diablo. When I count shares/blocks submitted in last 15 minutes it actually does roughly match. So it seems that despite apparent higher hash rate less shares/blocks are submitted to the pool.
I got Diablo from binary zip archive from link at the very front of this topic. It seems to be quite up to date. Timestamp on DiabloMiner.jar in archive is May 23. So it is recent version.
I was searching this forum and googled it. However, I am still in the dark abyss (and Diablo is lurking somewhere around ...) Does what I described ring a bell?
Topic: DiabloMiner GPU Miner - page 15. (Read 866226 times)
Apologies if this has been answered before, but I could not find an answer to my question on Google.
Is there a way for me to set a lower aggression for DiabloMiner ? I thought setting the -f flag to some high value might help, but not getting the desired result.
Is there a way for me to set a lower aggression for DiabloMiner ? I thought setting the -f flag to some high value might help, but not getting the desired result.
-f is fps. Set it to a multiple or divisor of 60. Higher is less aggressive.
That's what I thought
Very strange. On a multi-card setup, I've tried -f 1, 60, 600, 6000, 60000 and it's seemed to make no discernible difference in affecting hashrate I'll try snagging the latest-and-greatest DiabloMiner later this evening if see what's what.Thanks !
It doesnt effect the hashrate much, it effects the aggression. Only in badly designed miners that it greatly effects the hashrate.
On my 7979 at stock speeds -f 1000 gets me about 512. At the default of -f 30 it gets me about 556. I assume the difference is much larger in Windows.
Apologies if this has been answered before, but I could not find an answer to my question on Google.
Is there a way for me to set a lower aggression for DiabloMiner ? I thought setting the -f flag to some high value might help, but not getting the desired result.
Is there a way for me to set a lower aggression for DiabloMiner ? I thought setting the -f flag to some high value might help, but not getting the desired result.
-f is fps. Set it to a multiple or divisor of 60. Higher is less aggressive.
That's what I thought
Very strange. On a multi-card setup, I've tried -f 1, 60, 600, 6000, 60000 and it's seemed to make no discernible difference in affecting hashrate I'll try snagging the latest-and-greatest DiabloMiner later this evening if see what's what.Thanks !
Apologies if this has been answered before, but I could not find an answer to my question on Google.
Is there a way for me to set a lower aggression for DiabloMiner ? I thought setting the -f flag to some high value might help, but not getting the desired result.
Is there a way for me to set a lower aggression for DiabloMiner ? I thought setting the -f flag to some high value might help, but not getting the desired result.
-f is fps. Set it to a multiple or divisor of 60. Higher is less aggressive.
Apologies if this has been answered before, but I could not find an answer to my question on Google.
Is there a way for me to set a lower aggression for DiabloMiner ? I thought setting the -f flag to some high value might help, but not getting the desired result.
Is there a way for me to set a lower aggression for DiabloMiner ? I thought setting the -f flag to some high value might help, but not getting the desired result.
Update: Everyone effected by the problem, try the new build.
The new build works well, thanks! What was the issue?
Side note: I can now use a worksize of either 128 or 256 on OS X 10.7.4 and notice a small increase in hashrate over -w 64. May be worth mentioning in the OP instructions. I assume it's because of (one of?) the OpenCL updates Apple has pushed out.
FINALLY THEY FIXED IT! HOLY CRAP APPLE, IT ONLY TOOK TWO YEARS!
Also, the issue on AMD was AMD interprets the specification differently than I do. It costs basically nothing to fix it, and on GCN its measurably nothing.
Update: Everyone effected by the problem, try the new build.
The new build works well, thanks! What was the issue?
Side note: I can now use a worksize of either 128 or 256 on OS X 10.7.4 and notice a small increase in hashrate over -w 64. May be worth mentioning in the OP instructions. I assume it's because of (one of?) the OpenCL updates Apple has pushed out.
For me it is now double spacing instead of triple spacing.
That wasn't the bug I was talking about. That can only be fixed by going back and time and shooting whoever invented the UNIX console: DURR HURR LETS MAKE A VIRTUAL ON SCREEN TYPEWRITER, NO ONE WILL EVER KNOW
For me it is now double spacing instead of triple spacing.
Someone has confirmed the new build fixed it.
Update: Everyone effected by the problem, try the new build.
What's with the double spaced lines of output in this new version? I personally don't like it.
Sounds like your terminal isn't quite wide enough.
The default size? Yah no.
DM's console code is _very_ dumb. If output is double spaced or otherwise garbled, make your terminal wider until it stops.
What's with the double spaced lines of output in this new version? I personally don't like it.
Sounds like your terminal isn't quite wide enough.
The default size? Yah no.
This may be unrelated to typhoon's problem, but the latest version of diablominer fails to submit to and/or connect to the mining pool I've been using.
I'm running on 2011 iMacs, running OS X 10.7.4, with AMD Radeon HD 6750M (512 MB of RAM) and Apple OpenCL 1.1 (Apr 9 2012 19:41:45).
The command line arguments I'm using include:
./DiabloMiner-OSX.sh -w 64 -v 2 -na
The May 2012 version outputs the following:
The March 2012 version outputs the following:
Are there any options I can use to give more verbose messages?
On an unrelated note, I've recently noticed this model iMac go down to ~50-60 mhash/s from a previously consistent ~70 mhash/s. Could this be related to May's change(s) or the latest Lion update from Apple?
I'm running on 2011 iMacs, running OS X 10.7.4, with AMD Radeon HD 6750M (512 MB of RAM) and Apple OpenCL 1.1 (Apr 9 2012 19:41:45).
The command line arguments I'm using include:
./DiabloMiner-OSX.sh -w 64 -v 2 -na
The May 2012 version outputs the following:
Code:
[5/18/12 10:19:36 AM] Started
[5/18/12 10:19:36 AM] Connecting to: http://mining.eligius.st:8337/
[5/18/12 10:19:36 AM] Using Apple OpenCL 1.1 (Apr 9 2012 19:41:45)
[5/18/12 10:19:37 AM] Added ATI Radeon HD 6750M (#1) (6 CU, local work size of 64)
[5/18/12 10:29:40 AM] ERROR: Cannot connect to mining.eligius.st: Read timed out
[5/18/12 10:31:59 AM] ERROR: Cannot connect to mining.eligius.st: Read timed out
[5/18/12 10:38:21 AM] ERROR: Cannot connect to mining.eligius.st: Read timed out
mhash: 61.0/57.2 | accept: 0 | reject: 0 | hw error: 0
The March 2012 version outputs the following:
Code:
[5/18/12 2:09:31 PM] Started
[5/18/12 2:09:31 PM] Connecting to: http://mining.eligius.st:8337/
[5/18/12 2:09:31 PM] Using Apple OpenCL 1.1 (Apr 9 2012 19:41:45)
[5/18/12 2:09:32 PM] Added ATI Radeon HD 6750M (#1) (6 CU, local work size of 64)
[5/18/12 2:09:53 PM] mining.eligius.st accepted block 1 from ATI Radeon HD 6750M (#1)
[5/18/12 2:13:39 PM] mining.eligius.st accepted block 2 from ATI Radeon HD 6750M (#1)
[5/18/12 2:15:23 PM] mining.eligius.st accepted block 3 from ATI Radeon HD 6750M (#1)
[5/18/12 2:15:36 PM] mining.eligius.st accepted block 4 from ATI Radeon HD 6750M (#1)
[5/18/12 2:15:54 PM] mining.eligius.st accepted block 5 from ATI Radeon HD 6750M (#1)
mhash: 58.9/59.3 | accept: 5 | reject: 0 | hw error: 0
Are there any options I can use to give more verbose messages?
On an unrelated note, I've recently noticed this model iMac go down to ~50-60 mhash/s from a previously consistent ~70 mhash/s. Could this be related to May's change(s) or the latest Lion update from Apple?
Try a different pool. I have not changed anything that involves pools at all as far as I know. This has been largely the removal of code, not the changing of it.
As for OSX, its probably a Lion update. Quite a few Mac users have bitched that every time OSX puts a new update out, OpenCL apps get slower.
Unfortunately, I'm seeing the exact same problem on four different iMacs (all of the same model, OS version, etc.) regardless of whether the firewall is on/off. The only difference between it working and not is whether I run the March or May version.
Although I don't know much about the getwork protocol, eventually the March diablominer receives a response from the pool like the following (which I understand to mean that the work was accepted):
Code:
HTTP/1.1 200 OK
Content-Length: 40
X-Roll-NTime: expire=120
X-Long-Polling: /LP
Server: Eloipool
Date: Sat, 19 May 2012 17:15:11 GMT
Content-Type: application/json
{"result": true, "id": 1, "error": null}
As the result of sending a non-empty "params":[] value.
Whereas the May diablominer only receives new blocks after POSTing the following
Code:
{"method":"getwork","params":[],"id":1}I sniffed the May diablominer with wireshark for twenty minutes and never saw a non-empty "params" value sent. The old one sends one every minute to few minutes.
I'm not sure how else to troubleshoot this. Have any of the command line options changed (--url being the only other one I use that I didn't already mention)?
I suppose I could try another pool, but the only variable here seems to be the newer version.
eligius uses its own pool software that no other pool uses. Its very well possible that it isn't interacting well with a newer version of Jackson... but I upgraded that back in January. I wonder whats going on, it doesn't make sense.
What's with the double spaced lines of output in this new version? I personally don't like it.
Sounds like your terminal isn't quite wide enough.
This may be unrelated to typhoon's problem, but the latest version of diablominer fails to submit to and/or connect to the mining pool I've been using.
I'm running on 2011 iMacs, running OS X 10.7.4, with AMD Radeon HD 6750M (512 MB of RAM) and Apple OpenCL 1.1 (Apr 9 2012 19:41:45).
The command line arguments I'm using include:
./DiabloMiner-OSX.sh -w 64 -v 2 -na
The May 2012 version outputs the following:
The March 2012 version outputs the following:
Are there any options I can use to give more verbose messages?
On an unrelated note, I've recently noticed this model iMac go down to ~50-60 mhash/s from a previously consistent ~70 mhash/s. Could this be related to May's change(s) or the latest Lion update from Apple?
I'm running on 2011 iMacs, running OS X 10.7.4, with AMD Radeon HD 6750M (512 MB of RAM) and Apple OpenCL 1.1 (Apr 9 2012 19:41:45).
The command line arguments I'm using include:
./DiabloMiner-OSX.sh -w 64 -v 2 -na
The May 2012 version outputs the following:
Code:
[5/18/12 10:19:36 AM] Started
[5/18/12 10:19:36 AM] Connecting to: http://mining.eligius.st:8337/
[5/18/12 10:19:36 AM] Using Apple OpenCL 1.1 (Apr 9 2012 19:41:45)
[5/18/12 10:19:37 AM] Added ATI Radeon HD 6750M (#1) (6 CU, local work size of 64)
[5/18/12 10:29:40 AM] ERROR: Cannot connect to mining.eligius.st: Read timed out
[5/18/12 10:31:59 AM] ERROR: Cannot connect to mining.eligius.st: Read timed out
[5/18/12 10:38:21 AM] ERROR: Cannot connect to mining.eligius.st: Read timed out
mhash: 61.0/57.2 | accept: 0 | reject: 0 | hw error: 0
The March 2012 version outputs the following:
Code:
[5/18/12 2:09:31 PM] Started
[5/18/12 2:09:31 PM] Connecting to: http://mining.eligius.st:8337/
[5/18/12 2:09:31 PM] Using Apple OpenCL 1.1 (Apr 9 2012 19:41:45)
[5/18/12 2:09:32 PM] Added ATI Radeon HD 6750M (#1) (6 CU, local work size of 64)
[5/18/12 2:09:53 PM] mining.eligius.st accepted block 1 from ATI Radeon HD 6750M (#1)
[5/18/12 2:13:39 PM] mining.eligius.st accepted block 2 from ATI Radeon HD 6750M (#1)
[5/18/12 2:15:23 PM] mining.eligius.st accepted block 3 from ATI Radeon HD 6750M (#1)
[5/18/12 2:15:36 PM] mining.eligius.st accepted block 4 from ATI Radeon HD 6750M (#1)
[5/18/12 2:15:54 PM] mining.eligius.st accepted block 5 from ATI Radeon HD 6750M (#1)
mhash: 58.9/59.3 | accept: 5 | reject: 0 | hw error: 0
Are there any options I can use to give more verbose messages?
On an unrelated note, I've recently noticed this model iMac go down to ~50-60 mhash/s from a previously consistent ~70 mhash/s. Could this be related to May's change(s) or the latest Lion update from Apple?
Try a different pool. I have not changed anything that involves pools at all as far as I know. This has been largely the removal of code, not the changing of it.
As for OSX, its probably a Lion update. Quite a few Mac users have bitched that every time OSX puts a new update out, OpenCL apps get slower.
Unfortunately, I'm seeing the exact same problem on four different iMacs (all of the same model, OS version, etc.) regardless of whether the firewall is on/off. The only difference between it working and not is whether I run the March or May version.
Although I don't know much about the getwork protocol, eventually the March diablominer receives a response from the pool like the following (which I understand to mean that the work was accepted):
Code:
HTTP/1.1 200 OK
Content-Length: 40
X-Roll-NTime: expire=120
X-Long-Polling: /LP
Server: Eloipool
Date: Sat, 19 May 2012 17:15:11 GMT
Content-Type: application/json
{"result": true, "id": 1, "error": null}
As the result of sending a non-empty "params":[] value.
Whereas the May diablominer only receives new blocks after POSTing the following
Code:
{"method":"getwork","params":[],"id":1}I sniffed the May diablominer with wireshark for twenty minutes and never saw a non-empty "params" value sent. The old one sends one every minute to few minutes.
I'm not sure how else to troubleshoot this. Have any of the command line options changed (--url being the only other one I use that I didn't already mention)?
I suppose I could try another pool, but the only variable here seems to be the newer version.
What's with the double spaced lines of output in this new version? I personally don't like it.
Just reporting in that after I upgraded to the latest DiabloMiner, on both my Windows and Linux machines the miner starts as usual.. adding the devices and then hashing away.. However, shares are never submitted. I walked away for a while and saw that in fact no shares had been submitted to eligius. For proof check out: http://eligius.st/~artefact2/7/1AXiiwR4aYktPKMBMsZaP1untDgN9T8FfX
Edit: Found an older DiabloMiner and reverted (it does not include the 2 latest commits), back to mining.
Edit: Found an older DiabloMiner and reverted (it does not include the 2 latest commits), back to mining.
I cannot reproduce that bug. BTW, I've dropped OpenCL 1.0 support, so if you're on SDK 2.1... don't do that. DM is supposed to detect if you're on 1.0 and refuse to use those devices, too. Well, assuming I coded that test right.
I am on Catalyst 12.4 and the SDK that is included with the drivers. Radeon 5870s/5850s in both boxes (CentOS/Windows 7). If there is anything I can do to help you possibly pinpoint the problem I am all ears. Basically diablominer adds the OpenCL devices then starts hashing away at only god knows what.. and never submits shares, all whilst hashing at full speed and making my GPUs waste energy.
The hash meter has a number above zero, right? I'm wondering if Windows has a driver bug that Linux doesn't, because I'm on Linux /w Catalyst 12.4 using the runtime that comes with the drivers, and I can't reproduce it.
Yes the hashrate is at full speed, causing 100% gpu usage on all gpus. This happens on 2 separate systems, both with catalyst 12.4 and the included sdk. One system is windows and one is linux (CentOS).
Thats extra weird. I'm going to have to think about this one for a bit. It doesn't make sense you have the bug and I don't... I wonder if its hardware related, because I'm on a 7970, not a 58xx.
This may be unrelated to typhoon's problem, but the latest version of diablominer fails to submit to and/or connect to the mining pool I've been using.
I'm running on 2011 iMacs, running OS X 10.7.4, with AMD Radeon HD 6750M (512 MB of RAM) and Apple OpenCL 1.1 (Apr 9 2012 19:41:45).
The command line arguments I'm using include:
./DiabloMiner-OSX.sh -w 64 -v 2 -na
The May 2012 version outputs the following:
Code:
[5/18/12 10:19:36 AM] Started
[5/18/12 10:19:36 AM] Connecting to: http://mining.eligius.st:8337/
[5/18/12 10:19:36 AM] Using Apple OpenCL 1.1 (Apr 9 2012 19:41:45)
[5/18/12 10:19:37 AM] Added ATI Radeon HD 6750M (#1) (6 CU, local work size of 64)
[5/18/12 10:29:40 AM] ERROR: Cannot connect to mining.eligius.st: Read timed out
[5/18/12 10:31:59 AM] ERROR: Cannot connect to mining.eligius.st: Read timed out
[5/18/12 10:38:21 AM] ERROR: Cannot connect to mining.eligius.st: Read timed out
mhash: 61.0/57.2 | accept: 0 | reject: 0 | hw error: 0
The March 2012 version outputs the following:
Code:
[5/18/12 2:09:31 PM] Started
[5/18/12 2:09:31 PM] Connecting to: http://mining.eligius.st:8337/
[5/18/12 2:09:31 PM] Using Apple OpenCL 1.1 (Apr 9 2012 19:41:45)
[5/18/12 2:09:32 PM] Added ATI Radeon HD 6750M (#1) (6 CU, local work size of 64)
[5/18/12 2:09:53 PM] mining.eligius.st accepted block 1 from ATI Radeon HD 6750M (#1)
[5/18/12 2:13:39 PM] mining.eligius.st accepted block 2 from ATI Radeon HD 6750M (#1)
[5/18/12 2:15:23 PM] mining.eligius.st accepted block 3 from ATI Radeon HD 6750M (#1)
[5/18/12 2:15:36 PM] mining.eligius.st accepted block 4 from ATI Radeon HD 6750M (#1)
[5/18/12 2:15:54 PM] mining.eligius.st accepted block 5 from ATI Radeon HD 6750M (#1)
mhash: 58.9/59.3 | accept: 5 | reject: 0 | hw error: 0
Are there any options I can use to give more verbose messages?
On an unrelated note, I've recently noticed this model iMac go down to ~50-60 mhash/s from a previously consistent ~70 mhash/s. Could this be related to May's change(s) or the latest Lion update from Apple?
Try a different pool. I have not changed anything that involves pools at all as far as I know. This has been largely the removal of code, not the changing of it.
As for OSX, its probably a Lion update. Quite a few Mac users have bitched that every time OSX puts a new update out, OpenCL apps get slower.
Just reporting in that after I upgraded to the latest DiabloMiner, on both my Windows and Linux machines the miner starts as usual.. adding the devices and then hashing away.. However, shares are never submitted. I walked away for a while and saw that in fact no shares had been submitted to eligius. For proof check out: http://eligius.st/~artefact2/7/1AXiiwR4aYktPKMBMsZaP1untDgN9T8FfX
Edit: Found an older DiabloMiner and reverted (it does not include the 2 latest commits), back to mining.
Edit: Found an older DiabloMiner and reverted (it does not include the 2 latest commits), back to mining.
I cannot reproduce that bug. BTW, I've dropped OpenCL 1.0 support, so if you're on SDK 2.1... don't do that. DM is supposed to detect if you're on 1.0 and refuse to use those devices, too. Well, assuming I coded that test right.
I am on Catalyst 12.4 and the SDK that is included with the drivers. Radeon 5870s/5850s in both boxes (CentOS/Windows 7). If there is anything I can do to help you possibly pinpoint the problem I am all ears. Basically diablominer adds the OpenCL devices then starts hashing away at only god knows what.. and never submits shares, all whilst hashing at full speed and making my GPUs waste energy.
The hash meter has a number above zero, right? I'm wondering if Windows has a driver bug that Linux doesn't, because I'm on Linux /w Catalyst 12.4 using the runtime that comes with the drivers, and I can't reproduce it.
Yes the hashrate is at full speed, causing 100% gpu usage on all gpus. This happens on 2 separate systems, both with catalyst 12.4 and the included sdk. One system is windows and one is linux (CentOS).
Thats extra weird. I'm going to have to think about this one for a bit. It doesn't make sense you have the bug and I don't... I wonder if its hardware related, because I'm on a 7970, not a 58xx.
This may be unrelated to typhoon's problem, but the latest version of diablominer fails to submit to and/or connect to the mining pool I've been using.
I'm running on 2011 iMacs, running OS X 10.7.4, with AMD Radeon HD 6750M (512 MB of RAM) and Apple OpenCL 1.1 (Apr 9 2012 19:41:45).
The command line arguments I'm using include:
./DiabloMiner-OSX.sh -w 64 -v 2 -na
The May 2012 version outputs the following:
Code:
[5/18/12 10:19:36 AM] Started
[5/18/12 10:19:36 AM] Connecting to: http://mining.eligius.st:8337/
[5/18/12 10:19:36 AM] Using Apple OpenCL 1.1 (Apr 9 2012 19:41:45)
[5/18/12 10:19:37 AM] Added ATI Radeon HD 6750M (#1) (6 CU, local work size of 64)
[5/18/12 10:29:40 AM] ERROR: Cannot connect to mining.eligius.st: Read timed out
[5/18/12 10:31:59 AM] ERROR: Cannot connect to mining.eligius.st: Read timed out
[5/18/12 10:38:21 AM] ERROR: Cannot connect to mining.eligius.st: Read timed out
mhash: 61.0/57.2 | accept: 0 | reject: 0 | hw error: 0
The March 2012 version outputs the following:
Code:
[5/18/12 2:09:31 PM] Started
[5/18/12 2:09:31 PM] Connecting to: http://mining.eligius.st:8337/
[5/18/12 2:09:31 PM] Using Apple OpenCL 1.1 (Apr 9 2012 19:41:45)
[5/18/12 2:09:32 PM] Added ATI Radeon HD 6750M (#1) (6 CU, local work size of 64)
[5/18/12 2:09:53 PM] mining.eligius.st accepted block 1 from ATI Radeon HD 6750M (#1)
[5/18/12 2:13:39 PM] mining.eligius.st accepted block 2 from ATI Radeon HD 6750M (#1)
[5/18/12 2:15:23 PM] mining.eligius.st accepted block 3 from ATI Radeon HD 6750M (#1)
[5/18/12 2:15:36 PM] mining.eligius.st accepted block 4 from ATI Radeon HD 6750M (#1)
[5/18/12 2:15:54 PM] mining.eligius.st accepted block 5 from ATI Radeon HD 6750M (#1)
mhash: 58.9/59.3 | accept: 5 | reject: 0 | hw error: 0
Are there any options I can use to give more verbose messages?
On an unrelated note, I've recently noticed this model iMac go down to ~50-60 mhash/s from a previously consistent ~70 mhash/s. Could this be related to May's change(s) or the latest Lion update from Apple?
Just reporting in that after I upgraded to the latest DiabloMiner, on both my Windows and Linux machines the miner starts as usual.. adding the devices and then hashing away.. However, shares are never submitted. I walked away for a while and saw that in fact no shares had been submitted to eligius. For proof check out: http://eligius.st/~artefact2/7/1AXiiwR4aYktPKMBMsZaP1untDgN9T8FfX
Edit: Found an older DiabloMiner and reverted (it does not include the 2 latest commits), back to mining.
Edit: Found an older DiabloMiner and reverted (it does not include the 2 latest commits), back to mining.
I cannot reproduce that bug. BTW, I've dropped OpenCL 1.0 support, so if you're on SDK 2.1... don't do that. DM is supposed to detect if you're on 1.0 and refuse to use those devices, too. Well, assuming I coded that test right.
I am on Catalyst 12.4 and the SDK that is included with the drivers. Radeon 5870s/5850s in both boxes (CentOS/Windows 7). If there is anything I can do to help you possibly pinpoint the problem I am all ears. Basically diablominer adds the OpenCL devices then starts hashing away at only god knows what.. and never submits shares, all whilst hashing at full speed and making my GPUs waste energy.
The hash meter has a number above zero, right? I'm wondering if Windows has a driver bug that Linux doesn't, because I'm on Linux /w Catalyst 12.4 using the runtime that comes with the drivers, and I can't reproduce it.
Yes the hashrate is at full speed, causing 100% gpu usage on all gpus. This happens on 2 separate systems, both with catalyst 12.4 and the included sdk. One system is windows and one is linux (CentOS).
Thats extra weird. I'm going to have to think about this one for a bit. It doesn't make sense you have the bug and I don't... I wonder if its hardware related, because I'm on a 7970, not a 58xx.
Jump to: