[OS] nvOC easy-to-use Linux Nvidia Mining

[fullzero]

So I'm pretty new to mining, have 2 rigs. 1 running 2 1080ti's and 1 1060 mini. And then the other rig has 1 1080ti and one 1060 mini (other 1080ti is bunk) I tried downloading nvoc18 but it didn't cooperate and the copy just crashed upon boot. So I imaged back to nvoc17 and it's running well other than this. Stuff about the individual power settings and teamviewer etc those show up every time I've run 17, even with multiple re image attempts. My buddy voskcoin has been helping me troubleshoot my problems and getting me started, but he wasn't even sure about this.

Your photo link doesn't work.

[1714208498]

1714208498

[1714208498]

1714208498

[fullzero]

I ran the 2.2 version of TPruvot ccminer w/ skunk and got horrible hashrates.  I will recompile and test it again, but for now, krnlx is the fastest for me.  I'm not sure why there is a perceived drop of 30% in his setup, but I'm not seeing that at all.  I'm mining exactly at the expected coin rate per day based on both whattomine.com and crypto-economy.com calculators (450mh/s nets me around 550 coins a day +/- 25).  I will note that when I tried changing clock settings a little higher it really affected hashrates, dropping them over time considerably.  There is definitely a 'sweet spot' clock rate for skunk, and deviation appears to create greatly reduced production.  Also mining at -i 26 seems to have reduced my hashrates, so -i 25 is the proper setting for 1070 / 1080 Ti at least.  Lower settings or leaving -i blank greatly reduce hashrates, as is to be expected.

I'll update with the results of a recompile with tpruvot's most recent ccminer release on 0018.

Thanks for keeping an eye on this.

EDIT: the latest tpruvot ccminer release runs at just about the same hashrate as krnlx within 1-2mh/s (463mh/s vs. 462mh/s) and in a 2 hour test it mined 26-27 coins an hour on sigt.pool.mn which is a small number more than I was mining before, but it's a shjort test length so I'll update again after 24 hours to get a better average.  I was able to run tpruvot with skunk from command line in guake, but it crashes for some reason when running from the script almost immediately after it starts mining (before watchdog kicks in).  Not sure why that's happening, the command line is the same in both as I have it configured.

In the end, I don't see a dramatic difference between the two miners' performance except that I know krnlx runs from the script properly, whereas TPccminer 2.2 does not work in the script yet.  I'll keep playing around with it to see if some code in the script I added has an error. It's strange because the miner actually starts mining properly, but then errors out, rather than just not running at all which is normally how that goes.

Yeap skunk have sweet spot for every brand i think. U have to manage to find the fine tunning. I am not sure about this but i bet 1080ti can grab better performance on -i 26, i think u assign intensity per gpu like -d 0,1,2,3,4,5,6 -i 25 -d 7,8,9,10,11 -i 26
Anyway i get a bad experience with krnlx, before ccminer update, i was thinking about going back to palginmod with -20% hashrate ( not the new one )
I try to fix my variation GPU volatility, i cant run watchdog without lot of reboot by GPU sometime go to 0% for 1 second, everything is fine but watchdog dont like this
When u try to reboot miner via quake, i dont know why but on ccminer build, script cant kill "irq" is state in zombie mode. So when u launch miner again, GPU cant assign correct value cause if this "irq" zombie, just do "sudo pkill -e irq" and u can launch it

Are you using the newest watchdog; the one from the purple download link on the OP?

[BigSmurf]

Hi, i managed to connect 13*p106-100 gpu-z with no display output but when i log on via teamviewer the resolution is too small to do anything, could you fix this so that the resolution is normal ? Then i can do everything easily and test to my will:)

Edit: i get this in restart log :
Sun Aug 20 18:10:35 CEST 2017 - Utilization is too low: reviving did not work so restarting system in 10 seconds

Sun Aug 20 18:13:03 CEST 2017 - Starting miner restart script.

what does it mean

[TenaciousJ]

Yeap skunk have sweet spot for every brand i think. U have to manage to find the fine tunning. I am not sure about this but i bet 1080ti can grab better performance on -i 26, i think u assign intensity per gpu like -d 0,1,2,3,4,5,6 -i 25 -d 7,8,9,10,11 -i 26
Anyway i get a bad experience with krnlx, before ccminer update, i was thinking about going back to palginmod with -20% hashrate ( not the new one )
I try to fix my variation GPU volatility, i cant run watchdog without lot of reboot by GPU sometime go to 0% for 1 second, everything is fine but watchdog dont like this
When u try to reboot miner via quake, i dont know why but on ccminer build, script cant kill "irq" is state in zombie mode. So when u launch miner again, GPU cant assign correct value cause if this "irq" zombie, just do "sudo pkill -e irq" and u can launch it

Ahh... I was wondering how to set individual difficulty .. I'll give that a shot to see if I can squeeze some better perf. out of it.  Thanks for the tip.

I've just gone back to using krnlx in the script since the performance is nearly identical as far as i can tell and it's stable without having to worry whether watchdog will kill it or not.

[Bibi187]

I ran the 2.2 version of TPruvot ccminer w/ skunk and got horrible hashrates.  I will recompile and test it again, but for now, krnlx is the fastest for me.  I'm not sure why there is a perceived drop of 30% in his setup, but I'm not seeing that at all.  I'm mining exactly at the expected coin rate per day based on both whattomine.com and crypto-economy.com calculators (450mh/s nets me around 550 coins a day +/- 25).  I will note that when I tried changing clock settings a little higher it really affected hashrates, dropping them over time considerably.  There is definitely a 'sweet spot' clock rate for skunk, and deviation appears to create greatly reduced production.  Also mining at -i 26 seems to have reduced my hashrates, so -i 25 is the proper setting for 1070 / 1080 Ti at least.  Lower settings or leaving -i blank greatly reduce hashrates, as is to be expected.

I'll update with the results of a recompile with tpruvot's most recent ccminer release on 0018.

Thanks for keeping an eye on this.

EDIT: the latest tpruvot ccminer release runs at just about the same hashrate as krnlx within 1-2mh/s (463mh/s vs. 462mh/s) and in a 2 hour test it mined 26-27 coins an hour on sigt.pool.mn which is a small number more than I was mining before, but it's a shjort test length so I'll update again after 24 hours to get a better average.  I was able to run tpruvot with skunk from command line in guake, but it crashes for some reason when running from the script almost immediately after it starts mining (before watchdog kicks in).  Not sure why that's happening, the command line is the same in both as I have it configured.

In the end, I don't see a dramatic difference between the two miners' performance except that I know krnlx runs from the script properly, whereas TPccminer 2.2 does not work in the script yet.  I'll keep playing around with it to see if some code in the script I added has an error. It's strange because the miner actually starts mining properly, but then errors out, rather than just not running at all which is normally how that goes.

Yeap skunk have sweet spot for every brand i think. U have to manage to find the fine tunning. I am not sure about this but i bet 1080ti can grab better performance on -i 26, i think u assign intensity per gpu like -d 0,1,2,3,4,5,6 -i 25 -d 7,8,9,10,11 -i 26
Anyway i get a bad experience with krnlx, before ccminer update, i was thinking about going back to palginmod with -20% hashrate ( not the new one )
I try to fix my variation GPU volatility, i cant run watchdog without lot of reboot by GPU sometime go to 0% for 1 second, everything is fine but watchdog dont like this
When u try to reboot miner via quake, i dont know why but on ccminer build, script cant kill "irq" is state in zombie mode. So when u launch miner again, GPU cant assign correct value cause if this "irq" zombie, just do "sudo pkill -e irq" and u can launch it

Are you using the newest watchdog; the one from the purple download link on the OP?

Nop actually i use the old one, i prefer to wait version 0019 to get everything up to date. I can be patient and is actually work fine like this

[BigSmurf]

https://i.imgur.com/fVp8VRy.jpg

Some proof things work. still need a good Tutorial on 1bash file meanings for nvOC so ppl don't wast time on trial and error.

[Thagodeus]

Your photo link doesn't work.

Whoops sorry about that, here is the link! http://i999.photobucket.com/albums/af112/Ben_klink/E411BAAB-938C-42A3-BDE6-0B6B5610CF6A_zpskweq0qgp.jpg

Thanks!

[fred070]

When I want to enter commands I'm using Guake terminal. Sometimes it doesn't show uo even I hit f12.

Why is it unstable?

[damNmad]

When I want to enter commands I'm using Guake terminal. Sometimes it doesn't show uo even I hit f12.

Why is it unstable?

Not sure how you mean, but from my experience I can say, Guake doesn't work straight away after boot, the 1bash need to run first and you will see notification about Guake(press F12).

Then you can press F12, after 1st run of 1Bash after reboot only Guake works normal with F12.

Hope it makes sense and helps!

[fred070]

When I want to enter commands I'm using Guake terminal. Sometimes it doesn't show uo even I hit f12.

Why is it unstable?

Not sure how you mean, but from my experience I can say, Guake doesn't work straight away after boot, the 1bash need to run first and you will see notification about Guake(press F12).

Then you can press F12, after 1st run of 1Bash after reboot only Guake works normal with F12.

Hope it makes sense and helps!

Here's the scenario what happened earlier.

I wait 3 minutes for the system to fully load its daemon and programs and then I open up the guake (hit f12). After that I opened up the firefox and searched something (I often hit alt+tab to monitor if the miner still works).

When I'm about to check the watchdog, I can't access it anymore. So I hit f12 but no good. I clicked the launcher but it gives mo a "loading" mouse.

I made a workaround for now, I went to guake properties and put the window "always on top".

[ivoldemar]

When will you make a fix for P106

[BigSmurf]

When will you make a fix for P106

Fullzero said as soon as he gets a hold of 2 P106-100 cards.

Read pages 125 up to now and you can make it work

[damNmad]

When I want to enter commands I'm using Guake terminal. Sometimes it doesn't show uo even I hit f12.

Why is it unstable?

Not sure how you mean, but from my experience I can say, Guake doesn't work straight away after boot, the 1bash need to run first and you will see notification about Guake(press F12).

Then you can press F12, after 1st run of 1Bash after reboot only Guake works normal with F12.

Hope it makes sense and helps!

Here's the scenario what happened earlier.

I wait 3 minutes for the system to fully load its daemon and programs and then I open up the guake (hit f12). After that I opened up the firefox and searched something (I often hit alt+tab to monitor if the miner still works).

When I'm about to check the watchdog, I can't access it anymore. So I hit f12 but no good. I clicked the launcher but it gives mo a "loading" mouse.

I made a workaround for now, I went to guake properties and put the window "always on top".

That's something i haven't faced yet! probably your temp fix might help someone else's problem.

Thanks

[lards]

Hey guys,

Does anyone know how to set up the autostart 1bash , temp control and watchdog bash files when you use SSH ? Is it possible to set up in a way so when I turn my rig on using SSH remote control , the 1bash will autostart and mining goes without me using ' bash /home/m1/1bash ' ?

I've tried to add these processes into the autostart but with no luck, I always have to open 1bash myself through the console.

Thanks for you help.

By the way, since in my set up I do not have any normal cards with video output(only using P106-100 cards) , I had problems with integrated graphics processor , when I installed the fresh OS on the rig without attached monitor to mobo and used SSH(for editing 1bash , etc.) it seemed to me that my usual "NVLM error 15 " or just freezing of the system doesn't come up as often as when I was using a monitor and local controling of the rig. I think this is because now in xorg files there's no mention of my "intel graphics". So everyone who has the same problem try to install the OS this way and remote control through SSH. I'm still to try and work around the overclocking these cards , hopefully fullzero will connect sometime soon and help me with that.

[car1999]

Hey guys,

Does anyone know how to set up the autostart 1bash , temp control and watchdog bash files when you use SSH ? Is it possible to set up in a way so when I turn my rig on using SSH remote control , the 1bash will autostart and mining goes without me using ' bash /home/m1/1bash ' ?

I've tried to add these processes into the autostart but with no luck, I always have to open 1bash myself through the console.

Thanks for you help.

By the way, since in my set up I do not have any normal cards with video output(only using P106-100 cards) , I had problems with integrated graphics processor , when I installed the fresh OS on the rig without attached monitor to mobo and used SSH(for editing 1bash , etc.) it seemed to me that my usual "NVLM error 15 " or just freezing of the system doesn't come up as often as when I was using a monitor and local controling of the rig. I think this is because now in xorg files there's no mention of my "intel graphics". So everyone who has the same problem try to install the OS this way and remote control through SSH. I'm still to try and work around the overclocking these cards , hopefully fullzero will connect sometime soon and help me with that.

1bash is autostarted by default.
By the way, it's a good idea to buy a gtx1060 or 1050ti as primary GPU.

[lards]

Hey guys,

Does anyone know how to set up the autostart 1bash , temp control and watchdog bash files when you use SSH ? Is it possible to set up in a way so when I turn my rig on using SSH remote control , the 1bash will autostart and mining goes without me using ' bash /home/m1/1bash ' ?

I've tried to add these processes into the autostart but with no luck, I always have to open 1bash myself through the console.

Thanks for you help.

By the way, since in my set up I do not have any normal cards with video output(only using P106-100 cards) , I had problems with integrated graphics processor , when I installed the fresh OS on the rig without attached monitor to mobo and used SSH(for editing 1bash , etc.) it seemed to me that my usual "NVLM error 15 " or just freezing of the system doesn't come up as often as when I was using a monitor and local controling of the rig. I think this is because now in xorg files there's no mention of my "intel graphics". So everyone who has the same problem try to install the OS this way and remote control through SSH. I'm still to try and work around the overclocking these cards , hopefully fullzero will connect sometime soon and help me with that.

1bash is autostarted by default.
By the way, it's a good idea to buy a gtx1060 or 1050ti as primary GPU.

I think it's not the case here since the mining process doesn't start by itself or any other bash files like temp control and Watchdog,also checked through 'nvidia-smi' and 'screen -r miner' , I have to use "bash /home/m1/1bash" in my SSH terminal to start mining.

I was thinking about that but since it's not as easy for me to obtain new hardware quickly I have to adjust to what I have at the moment. I will probably end up obtaining one 1050ti card just to be safe since I still have one slot free anyway.

Did you have any luck with OC in your set up with 1 normal card and 12 P106-100 ?

[fred070]

Hi Fullzero,

Do we need to add a swap space for v0019 release?
In claymore README for Windows 10, it's recommended to increase the pagefile to 16,000MB. I'm wondering if we also need to setup an extra "memory" for nvOC.

[fred070]

Hey guys,

Does anyone know how to set up the autostart 1bash , temp control and watchdog bash files when you use SSH ? Is it possible to set up in a way so when I turn my rig on using SSH remote control , the 1bash will autostart and mining goes without me using ' bash /home/m1/1bash ' ?

I've tried to add these processes into the autostart but with no luck, I always have to open 1bash myself through the console.

Thanks for you help.

By the way, since in my set up I do not have any normal cards with video output(only using P106-100 cards) , I had problems with integrated graphics processor , when I installed the fresh OS on the rig without attached monitor to mobo and used SSH(for editing 1bash , etc.) it seemed to me that my usual "NVLM error 15 " or just freezing of the system doesn't come up as often as when I was using a monitor and local controling of the rig. I think this is because now in xorg files there's no mention of my "intel graphics". So everyone who has the same problem try to install the OS this way and remote control through SSH. I'm still to try and work around the overclocking these cards , hopefully fullzero will connect sometime soon and help me with that. thanks

Not sure if it's possible to use the integrated graphics in your mobo. Fullzero has a note in first page.



NOTE: If you are connecting a monitor directly to the motherboard (don't do this);
only connect a monitor to the primary GPU (the one in the 16x slot closest to the CPU)

[lards]

Hey guys,

Does anyone know how to set up the autostart 1bash , temp control and watchdog bash files when you use SSH ? Is it possible to set up in a way so when I turn my rig on using SSH remote control , the 1bash will autostart and mining goes without me using ' bash /home/m1/1bash ' ?

I've tried to add these processes into the autostart but with no luck, I always have to open 1bash myself through the console.

Thanks for you help.

By the way, since in my set up I do not have any normal cards with video output(only using P106-100 cards) , I had problems with integrated graphics processor , when I installed the fresh OS on the rig without attached monitor to mobo and used SSH(for editing 1bash , etc.) it seemed to me that my usual "NVLM error 15 " or just freezing of the system doesn't come up as often as when I was using a monitor and local controling of the rig. I think this is because now in xorg files there's no mention of my "intel graphics". So everyone who has the same problem try to install the OS this way and remote control through SSH. I'm still to try and work around the overclocking these cards , hopefully fullzero will connect sometime soon and help me with that. thanks

Not sure if it's possible to use the integrated graphics in your mobo. Fullzero has a note in first page.



NOTE: If you are connecting a monitor directly to the motherboard (don't do this);
only connect a monitor to the primary GPU (the one in the 16x slot closest to the CPU)

I know you are just trying to help but if you read my post accurately I specified that I don't have any cards with video output so I can't connect a monitor to the primary GPU and therefore I explained what I've done to make it usable with my setup.

So does anyone know why the 1bash doesn't autostart when I use the remote connection through SSH(I have to use 'bash /home/m1/1bash' )? I am gonna try to work around that problem today but would be grateful if somebody can point me in the right direction of how to make it work.

As I stated before I have first booted without the monitor using just P106-100 cards and then changed all the bash files to new ones in the first post through SSH control.

[dbolivar]

Do we need to add a swap space for v0019 release?
In claymore README for Windows 10, it's recommended to increase the pagefile to 16,000MB. I'm wondering if we also need to setup an extra "memory" for nvOC.

I never needed to add swap to mine in Linux, using Claymore or whatever other miner. In nvOC, for instance, the RAM usage is very low, less than 2 GB, so if you have 4 GB or more of RAM (the minimum recommended, by the way), you shouldn't need a swap file or partition.