bullus you can use one sgminer instance, you need to apply a different "hamsi-expand-big" setting for the card that you want marucoin-mod-old on.
In theory it should look like [FOR EXAMPLE]
"hamsi-expand-big" : "1,1,4,1,1" for these cards "290,290,280,290,290"
I think this will still work and should work all in the same instance.
Topic: [ANN] sgminer v5 - optimized X11/X13/NeoScrypt/Lyra2RE/etc. kernel-switch miner - page 229. (Read 877859 times)
June 15, 2014, 07:08:06 PM
June 15, 2014, 05:59:53 PM
Anyone has a working bat file?
June 15, 2014, 05:42:52 PM
Everything's still working, despite total mayhem server-side 
Code:
[18:11:35] Accepted 0f216193 Diff 0.066/0.050 GPU 1 at NiceHash_X13_AS
[18:12:18] Stratum connection to NiceHash_X13_AS interrupted
[18:12:18] Lost 1 shares due to stratum disconnect on NiceHash_X13_AS
[18:12:19] NiceHash_X13_AS not responding!
[18:12:19] Switching to NiceHash_Keccak_AS
[18:12:20] NiceHash_Keccak_AS not responding!
[18:12:28] Waiting for work to be available from pools.
[18:12:51] NiceHash_X13_AS alive, testing stability
[18:12:56] NiceHash_X11_Backup extranonce change requested
[18:13:14] NiceHash_X13_AS stable for 30 seconds
[18:13:14] Switching to NiceHash_X13_AS
[18:13:23] Work available from pools, resuming.
[18:13:36] Accepted 0593bd8b Diff 0.179/0.080 GPU 1 at NiceHash_X13_AS
[18:14:51] NiceHash_X13_AS extranonce change requested
[18:15:03] Accepted 055d4376 Diff 0.186/0.080 GPU 0 at NiceHash_X13_AS
[18:16:33] Stratum connection to NiceHash_X13_AS interrupted
[18:16:33] Lost 3 shares due to stratum disconnect on NiceHash_X13_AS
[18:16:34] NiceHash_X13_AS not responding!
[18:16:34] Switching to NiceHash_X11_Backup
[18:16:35] NiceHash_X11_Backup not responding!
[18:16:43] Waiting for work to be available from pools.
[18:17:37] NiceHash_X13_AS alive, testing stability
[18:17:38] NiceHash_X11_Backup difficulty changed to 0.005
[18:17:43] Work available from pools, resuming.
[18:17:43] Initialising kernel darkcoin-mod.cl with bitalign, unpatched BFI, nfactor 10, n 1024
[18:17:43] Initialising kernel darkcoin-mod.cl with bitalign, unpatched BFI, nfactor 10, n 1024
[18:17:48] Accepted c785823a Diff 0.005/0.005 GPU 1 at NiceHash_X11_Backup
June 15, 2014, 04:27:30 PM
Test what? I went to the link and say something about a "remove-disabled" : "true" but that didn't change anything. I still get 0's hashs
We know about this issue: https://github.com/sgminer-dev/sgminer/issues/249Code:
"remove-disabled" : true (true without quotes)
Code:
"device" : "1,2,3,4,5",
Code:
"gpu-threads": "2,1,1,2,2,1",
The "device" tells it which devices to use. you still have to have all six settings in other areas, otherwise it would incorrectly assign the settings you want.
Anyhow, I got it to work by fenagling with this:
"algorithm" : "marucoin-mod",
"kernel-path": "/usr/local/bin"
I used the c:/path/to/my/bin instead of /usr/local/bin... have no idea if that was the right thing to do but I did that, deleted bins and it works now. What a PITA lol
Never mind... it worked for a few minutes and then seems to cutout when switching algos
edit: works just fine on my rig with 4 x 7970s and I didn't have to have an "algo" specified in the conf with my other device settings. But the rig with 280x and 5870/5970 is failing
Let's try to solve it step by step:
1. Is miner stable with pool-algortihm setting equal for all pools?
2. Is miner stable without device option?
3. Can you send log file when miner crashes?
4. If miner crashes only on algoswitch, can you send log file from this version: https://www.dropbox.com/s/2zb0f3sal6kkrgl/sgminer_algoswitch_addlog.zip ? I added some messages to thread restart procedure.
Also, can you post your current conf file?
Thank you Elun and Bullus.
1. wouldn't different pools have different algos?
2. I can't confirm anything as stable in this version as I've experienced far more driver glitches since trying sgminer5
3. I haven't recorded logs on this thing... I'll consider it.
4. Miner crashed other times but not very often.
Here's my currrent conf. I guess I have a lot more work to do to set various parameters on various pools. Although... I still go back to the fact that my other miner works perfectly fine. what could be the difference between a 7970 rig and a 280X/5870/5970 rig? As far as I can tell... everything else is the same on the two rigs other than the video cards. Here's the working config on my 7970 rig:
Code:
{
],
"gpu-engine" : "1065,1065,1000,1040",
"gpu-memclock": "1250",
"worksize": "256",
"intensity" : "19",
"thread-concurrency": "8192",
"gpu-threads": "2",
"lookup-gap" : "2",
"queue": "1"
}
only diff between rigs are the "device" : "1,2,3,4,5" and the algo settings.
edit... probably just something screwy with my rig
For the 280X/5870/5970 rig u need to use 2xsgminer to mine, like I did.
1 config for only the 280x and another for 5870/5970 using marucoinmod-old for x13
edit;
my config for 7950
Code:
{
"pools" : [
{
"name" : "am02 x11 multi",
"url" : "stratum+tcp://am02.eu.trademybit.com:4010",
"user" : "xxx",
"pass" : "1111",
"algorithm":"darkcoin-mod"
},
{
"name" : "am02 x13 multi",
"url" : "stratum+tcp://am02.eu.trademybit.com:4011",
"user" : "xxx",
"pass" : "1111",
"algorithm":"marucoin-mod"
}
],
"extranonce-subscribe" : true,
"remove-disabled" : true,
"api-allow" : "W:127.0.0.1",
"api-listen" : true,
"api-mcast-port" : "4028",
"api-port" : "4028",
"expiry" : "10",
"failover-only" : true,
"kernel-path" : "D:\\sgminer_v5_0_10062014",
"log" : "5",
"queue" : "1",
"scan-time" : "7",
"shares" : "0",
"failover-switch-delay" : "10",
"gpu-threads" : "2",
"gpu-dyninterval" : "7",
"gpu-engine" : "0",
"gpu-fan" : "65",
"gpu-platform" : "0",
"gpu-memclock" : "0",
"gpu-memdiff" : "0",
"gpu-powertune" : "8",
"gpu-vddc" : "0",
"intensity" : "18",
"temp-target" : "75",
"temp-overheat" : "85",
"temp-cutoff" : "95",
"temp-hysteresis" : "3",
"vectors" : "1",
"worksize" : "256",
"lookup-gap" : "0",
"shaders" : "1792",
"thread-concurrency" : "8192",
"device" : "0,2,3,4",
"no-pool-disable" : true
}6970
Code:
{
"pools" : [
{
"name" : "am02 x11 multi",
"url" : "stratum+tcp://am02.eu.trademybit.com:4010",
"user" : "xxx",
"pass" : "1111",
"algorithm":"darkcoin-mod"
},
{
"name" : "am02 x13 multi",
"url" : "stratum+tcp://am02.eu.trademybit.com:4011",
"user" : "xxx",
"pass" : "1111",
"algorithm":"marucoin-modold"
}
],
"extranonce-subscribe" : true,
"remove-disabled" : true,
"api-allow" : "W:127.0.0.1",
"api-listen" : true,
"api-mcast-port" : "4029",
"api-port" : "4029",
"expiry" : "10",
"failover-only" : true,
"kernel-path" : "D:\\sgminer_v5_0_10062014",
"log" : "5",
"queue" : "1",
"scan-time" : "7",
"shares" : "0",
"failover-switch-delay" : "10",
"gpu-threads" : "2",
"gpu-dyninterval" : "7",
"gpu-engine" : "0",
"gpu-fan" : "65",
"gpu-platform" : "0",
"gpu-memclock" : "0",
"gpu-memdiff" : "0",
"gpu-powertune" : "8",
"gpu-vddc" : "0",
"intensity" : "18",
"temp-target" : "75",
"temp-overheat" : "85",
"temp-cutoff" : "95",
"temp-hysteresis" : "3",
"vectors" : "1",
"worksize" : "64",
"lookup-gap" : "0",
"shaders" : "1536",
"thread-concurrency" : "8000",
"device" : "1",
"no-pool-disable" : true
} June 15, 2014, 02:43:14 PM
@Elun
Are u working for a fix to use only 1 sgminer when using multi-algo?
"algorithm" : "marucoinmod,marucoinmod-old,marucoinmod"
Are u working for a fix to use only 1 sgminer when using multi-algo?
"algorithm" : "marucoinmod,marucoinmod-old,marucoinmod"
June 15, 2014, 02:34:08 PM
Test what? I went to the link and say something about a "remove-disabled" : "true" but that didn't change anything. I still get 0's hashs
We know about this issue: https://github.com/sgminer-dev/sgminer/issues/249Code:
"remove-disabled" : true (true without quotes)
Code:
"device" : "1,2,3,4,5",
Code:
"gpu-threads": "2,1,1,2,2,1",
The "device" tells it which devices to use. you still have to have all six settings in other areas, otherwise it would incorrectly assign the settings you want.
Anyhow, I got it to work by fenagling with this:
"algorithm" : "marucoin-mod",
"kernel-path": "/usr/local/bin"
I used the c:/path/to/my/bin instead of /usr/local/bin... have no idea if that was the right thing to do but I did that, deleted bins and it works now. What a PITA lol
Never mind... it worked for a few minutes and then seems to cutout when switching algos
edit: works just fine on my rig with 4 x 7970s and I didn't have to have an "algo" specified in the conf with my other device settings. But the rig with 280x and 5870/5970 is failing
Let's try to solve it step by step:
1. Is miner stable with pool-algortihm setting equal for all pools?
2. Is miner stable without device option?
3. Can you send log file when miner crashes?
4. If miner crashes only on algoswitch, can you send log file from this version: https://www.dropbox.com/s/2zb0f3sal6kkrgl/sgminer_algoswitch_addlog.zip ? I added some messages to thread restart procedure.
Also, can you post your current conf file?
June 15, 2014, 02:12:06 PM
Test what? I went to the link and say something about a "remove-disabled" : "true" but that didn't change anything. I still get 0's hashs
We know about this issue: https://github.com/sgminer-dev/sgminer/issues/249Code:
"remove-disabled" : true (true without quotes)
Code:
"device" : "1,2,3,4,5",
Code:
"gpu-threads": "2,1,1,2,2,1",
Thanks, I tested it and now I use 2 sgminers to mine and it works with "remove-disabled" : true
7950,6970,7950,7950,7950
1 config with "device" : "0,2,3,4" and the other with "device" : "1"
June 15, 2014, 01:46:00 PM
Test what? I went to the link and say something about a "remove-disabled" : "true" but that didn't change anything. I still get 0's hashs
We know about this issue: https://github.com/sgminer-dev/sgminer/issues/249Code:
"remove-disabled" : true (true without quotes)
Code:
"device" : "1,2,3,4,5",
Code:
"gpu-threads": "2,1,1,2,2,1",
The "device" tells it which devices to use. you still have to have all six settings in other areas, otherwise it would incorrectly assign the settings you want.
Anyhow, I got it to work by fenagling with this:
"algorithm" : "marucoin-mod",
"kernel-path": "/usr/local/bin"
I used the c:/path/to/my/bin instead of /usr/local/bin... have no idea if that was the right thing to do but I did that, deleted bins and it works now. What a PITA lol
Never mind... it worked for a few minutes and then seems to cutout when switching algos
edit: works just fine on my rig with 4 x 7970s and I didn't have to have an "algo" specified in the conf with my other device settings. But the rig with 280x and 5870/5970 is failing
June 15, 2014, 01:26:45 PM
Today we introduced another advanced feature for GPU miners - Fine tuning with custom speed factors and daily costs. You can now set custom hashing rate factors and even your electricity power consumption as well! This will make sure your miner mines on absolutely most profitable algorithm. You can also configure your miner, for example, to only switch between X11 and X13 (whichever is more profitable) even if globally some other algorithm is more profitable. Visit https://www.nicehash.com/multialgo/#finetuning and play around I'm sure this is one very nice tool for GPU users to play with!
p.s.: this is why you may have notice several disconnects today while we were deploying this new features
I'm sure this is one very nice tool for GPU users to play with!p.s.: this is why you may have notice several disconnects today while we were deploying this new features
Awesome.
edit: I think this is most useful for limiting miners to only specific algorithms. If you use ALL algorithms, chances are the defaults will work fine, unless you have a bum card that eats up more power than it should.
June 15, 2014, 01:13:36 PM
Test what? I went to the link and say something about a "remove-disabled" : "true" but that didn't change anything. I still get 0's hashs
We know about this issue: https://github.com/sgminer-dev/sgminer/issues/249Code:
"remove-disabled" : true (true without quotes)
Code:
"device" : "1,2,3,4,5",
Code:
"gpu-threads": "2,1,1,2,2,1",
The "device" tells it which devices to use. you still have to have all six settings in other areas, otherwise it would incorrectly assign the settings you want.
Anyhow, I got it to work by fenagling with this:
"algorithm" : "marucoin-mod",
"kernel-path": "/usr/local/bin"
I used the c:/path/to/my/bin instead of /usr/local/bin... have no idea if that was the right thing to do but I did that, deleted bins and it works now. What a PITA lol
June 15, 2014, 12:24:51 PM
Test what? I went to the link and say something about a "remove-disabled" : "true" but that didn't change anything. I still get 0's hashs
We know about this issue: https://github.com/sgminer-dev/sgminer/issues/249Code:
"remove-disabled" : true (true without quotes)
Code:
"device" : "1,2,3,4,5",
Code:
"gpu-threads": "2,1,1,2,2,1",
June 15, 2014, 12:14:20 PM
June 15, 2014, 12:00:55 PM
Today we introduced another advanced feature for GPU miners - Fine tuning with custom speed factors and daily costs. You can now set custom hashing rate factors and even your electricity power consumption as well! This will make sure your miner mines on absolutely most profitable algorithm. You can also configure your miner, for example, to only switch between X11 and X13 (whichever is more profitable) even if globally some other algorithm is more profitable. Visit https://www.nicehash.com/multialgo/#finetuning and play around I'm sure this is one very nice tool for GPU users to play with!
p.s.: this is why you may have notice several disconnects today while we were deploying this new features
I'm sure this is one very nice tool for GPU users to play with!p.s.: this is why you may have notice several disconnects today while we were deploying this new features
June 15, 2014, 11:12:36 AM
Going back to the CGWatcher question, It works perfectly fine for me, but I am not letting it launch sgminer. I have sgminer.exe and the conf file defined in the current profile, but i did not create the conf file with CGWatcher. CGWatcher strips the pool-algo entry from the conf file and adds pool priority.
Instead, Launch CGwatcher pointed to the sgminer.exe and conf, and then launch sgminer out of a bat file that points to the conf file separate. CGWatcher *should* detect that its mining and being to display stats.
Instead, Launch CGwatcher pointed to the sgminer.exe and conf, and then launch sgminer out of a bat file that points to the conf file separate. CGWatcher *should* detect that its mining and being to display stats.
June 15, 2014, 09:23:53 AM
With the same locked clock settings (defined highs/lows) interalgorithm, I've successfully gotten this for the past hour.
It's been back and forth a few times with nary a hiccup.
Code:
[06:44:13] Stratum connection to NiceHash_X11_AS interrupted
[06:44:23] NiceHash_X11_AS not responding!
[06:44:23] Switching to NiceHash_Scrypt_Backup
[06:44:24] Waiting for work to be available from pools.
[06:44:35] NiceHash_Scrypt_Backup not responding!
[06:44:36] NiceHash_Scrypt_Backup difficulty changed to 1024
[06:44:38] Work available from pools, resuming.
[06:44:39] Initialising kernel zuikkis.cl with bitalign, unpatched BFI, nfactor 10, n 1024
[06:44:39] Initialising kernel zuikkis.cl with bitalign, unpatched BFI, nfactor 10, n 1024
[06:44:54] NiceHash_X11_AS alive, testing stability
[06:45:03] Stratum connection to NiceHash_X11_AS interrupted
[06:45:03] Reconnect requested from NiceHash_Scrypt_Backup to stratum.nicehash.com:3333
[06:45:13] Waiting for work to be available from pools.
[06:45:16] Rejected untracked stratum share from NiceHash_Scrypt_Backup
[06:45:16] Stratum connection to NiceHash_Scrypt_Backup interrupted
[06:45:23] NiceHash_Scrypt_Backup difficulty changed to 16
[06:45:23] Work available from pools, resuming.
[06:45:24] Accepted 0f1241e5 Diff 16/16 GPU 0 at NiceHash_Scrypt_Backup
[06:45:24] Accepted 0e3d19d6 Diff 17/16 GPU 0 at NiceHash_Scrypt_Backup
[06:45:25] Accepted 09802c89 Diff 26/16 GPU 1 at NiceHash_Scrypt_Backup
It's been back and forth a few times with nary a hiccup.
So maybe we can assume that this is running okay since it is on the same nf10, and not changing to nf11?
EDIT: Reporting in confirming that I can effectively Switch Pools back and forth between x11/x13/scrypt on an nf10, my bin file is ckolivasHawaiigw128l4lg2tc32765nf10 so the TC is not throwing a -4 Error so far.
June 15, 2014, 08:48:22 AM
With the same locked clock settings (defined highs/lows) interalgorithm, I've successfully gotten this for the past hour.
It's been back and forth a few times with nary a hiccup.
Code:
[06:44:13] Stratum connection to NiceHash_X11_AS interrupted
[06:44:23] NiceHash_X11_AS not responding!
[06:44:23] Switching to NiceHash_Scrypt_Backup
[06:44:24] Waiting for work to be available from pools.
[06:44:35] NiceHash_Scrypt_Backup not responding!
[06:44:36] NiceHash_Scrypt_Backup difficulty changed to 1024
[06:44:38] Work available from pools, resuming.
[06:44:39] Initialising kernel zuikkis.cl with bitalign, unpatched BFI, nfactor 10, n 1024
[06:44:39] Initialising kernel zuikkis.cl with bitalign, unpatched BFI, nfactor 10, n 1024
[06:44:54] NiceHash_X11_AS alive, testing stability
[06:45:03] Stratum connection to NiceHash_X11_AS interrupted
[06:45:03] Reconnect requested from NiceHash_Scrypt_Backup to stratum.nicehash.com:3333
[06:45:13] Waiting for work to be available from pools.
[06:45:16] Rejected untracked stratum share from NiceHash_Scrypt_Backup
[06:45:16] Stratum connection to NiceHash_Scrypt_Backup interrupted
[06:45:23] NiceHash_Scrypt_Backup difficulty changed to 16
[06:45:23] Work available from pools, resuming.
[06:45:24] Accepted 0f1241e5 Diff 16/16 GPU 0 at NiceHash_Scrypt_Backup
[06:45:24] Accepted 0e3d19d6 Diff 17/16 GPU 0 at NiceHash_Scrypt_Backup
[06:45:25] Accepted 09802c89 Diff 26/16 GPU 1 at NiceHash_Scrypt_Backup
It's been back and forth a few times with nary a hiccup.
June 15, 2014, 07:37:52 AM
Right off the bat, I'm getting "thread was not cancelled in 60 seconds after restart_mining_threads"
you need to remove pool-gpu-threads from your config and use gpu-threads in the global config part
Ok got it. Wait, wasn't this figured out ten pages ago? How on earth did it end up in kenshirothefist's '100% working' example config? (no worries if it was just an oversight)
Well it looks like the "thread was not cancelled in 60 seconds after restart_mining_threads" is another issue that is platform/os/drivers/dependent as it mainly tend to show on Windows; in our testing on Linux rigs we never encountered this issue. It is discussed here: https://github.com/sgminer-dev/sgminer/issues/246. So, yes, the solution is to not use "pool-gpu-threads" and use only "gpu-threads" in the global config part.
Is it possible we're experiencing SICK/DEAD cards and shut-offs every 6 or so hours on the Windows builds primarily?
Any way we can mitigate this? As stated above, removing 1 environment variable from Windows appears to increase stability in the short term.
kenshiro's config has switched and run all night on Windows for me (sans pool-gpu-threads), but I have yet to see the switch to Scrypt/NScrypt. I suspect the key is probably keeping engine/memclock locked solid between algorithms. Will continue to report.
June 15, 2014, 05:12:38 AM
Right off the bat, I'm getting "thread was not cancelled in 60 seconds after restart_mining_threads"
you need to remove pool-gpu-threads from your config and use gpu-threads in the global config part
Ok got it. Wait, wasn't this figured out ten pages ago? How on earth did it end up in kenshirothefist's '100% working' example config? (no worries if it was just an oversight)
Well it looks like the "thread was not cancelled in 60 seconds after restart_mining_threads" is another issue that is platform/os/drivers/dependent as it mainly tend to show on Windows; in our testing on Linux rigs we never encountered this issue. It is discussed here: https://github.com/sgminer-dev/sgminer/issues/246. So, yes, the solution is to not use "pool-gpu-threads" and use only "gpu-threads" in the global config part.
Is it possible we're experiencing SICK/DEAD cards and shut-offs every 6 or so hours on the Windows builds primarily?
Any way we can mitigate this? As stated above, removing 1 environment variable from Windows appears to increase stability in the short term.
June 15, 2014, 04:33:26 AM
Right off the bat, I'm getting "thread was not cancelled in 60 seconds after restart_mining_threads"
you need to remove pool-gpu-threads from your config and use gpu-threads in the global config part
Ok got it. Wait, wasn't this figured out ten pages ago? How on earth did it end up in kenshirothefist's '100% working' example config? (no worries if it was just an oversight)
Well it looks like the "thread was not cancelled in 60 seconds after restart_mining_threads" is another issue that is platform/os/drivers/dependent as it mainly tend to show on Windows; in our testing on Linux rigs we never encountered this issue. It is discussed here: https://github.com/sgminer-dev/sgminer/issues/246. So, yes, the solution is to not use "pool-gpu-threads" and use only "gpu-threads" in the global config part.
June 15, 2014, 02:45:48 AM
yes i am setting that variable.. but in my case its not ALWAYS the last, most of the time its GPU 0 and GPU 2, but i have seen GPU 1 and GPU 3 do it also sometimes.
I have switched the rig that experiences DEAD cards over to STOCK values on the questionable card, and I deleted the system environment variable GPU_USE_SYNC_OBJECTS 1 and there has been absolutely no drop in hashrate so far, and I also omitted no-restart:true so that the miner would try and restart threads again.
I will report back with findings on uptime (Windows binaries from 6/11 off nh site), gpu-threads 1, no nscrypt flip.
EDIT: Noticing about a 1-2% increase in hashrate actually, after removing GPU_USE_SYNC_OBJECTS.
This is for x11/x13 ONLY this has not been tested on scrypt or nscrypt.
EDIT 2: Not a huge deal, but should be incredibly easy to implement --> https://github.com/sgminer-dev/sgminer/issues/268
This would let us cover all bases in case for some reason NIST5 gets a moment in the sun, mining-algo-wise...
Jump to: