Pages:
Author

Topic: [OS] nvOC easy-to-use Linux Nvidia Mining - page 8. (Read 418313 times)

jr. member
Activity: 128
Merit: 1
It's supported for sure, it's the first thing I pushed into nvOC.

Tweaking per-vendor or model is a no go because of the silicon lottery, so use the algo specific with offset and you will be happy.

edit: it is everywhere mentioned in 3main, just search for "offset"
https://github.com/papampi/nvOC_by_fullzero_Community_Release/blob/3.0-stable/3main
member
Activity: 224
Merit: 13
That line about "pro settings" just refers to more complex policies for OC levels, like algo specific with gpu offset. They are quite complex to understand at a first glance but will be clear what they do once someone realize he needs such useful policies.
That type of OC is just what I want as I have 2 hosts with GPUs that are the same model (e.g., 1080 Ti) but from different vendors so they require different power and clocks. Ideally, I would like to be able to OC per model/vendor and again for algo. Currently, I do this with different 1bash files named 1bash_$coin. I suppose that is the least painful way to do it. My point was that I see this one mention of it in 1bash but no support for it in 3main. Which branch has it?
jr. member
Activity: 128
Merit: 1
Open a quick issue on github for the powerlimit reporting bug, such that we will be able to remember to fix this the next time attaching this to another more urgent PR.

That line about "pro settings" just refers to more complex policies for OC levels, like algo specific with gpu offset. They are quite complex to understand at a first glance but will be clear what they do once someone realize he needs such useful policies.
member
Activity: 224
Merit: 13
Hey Devs:

I found a bug in the nvOC script. I noted that it was not reporting the max power limit for GPUs correctly so I dug in. On line 196 of the 3.0-stable release, you have this:
Code:
   POWERMAX=$(echo "$PWRLIMIT/1" | bc)
but it should be
Code:
   POWERMAX=$(echo "$POWERMAX/1" | bc)

Also, where can I find more info on this (from 1bash of 3.0-stable):
Code:
#        - Pro Settings are not for the faint-hearted. Dont use if you dont know what they are.

I scoured 1bash and 3main for this release and don't see anything else about it. Perhaps it is in a different release branch?

Thanks.
member
Activity: 224
Merit: 13
@stubo every time you open a gnome-terminal to its default profile nvOC runs again from scratch, it is like this from fullzero nvOC autostart method design. If you just need a shell open guake or lauch gnome-terminale with another profile. There is an alternative to gnome-terminal autostart based on systemd. Dig further in the code, the initial setup install may suffer some rough edges but in the end works like a charm, and the restart/stop commands are much more reliable.

The option in firstboot to use a different path works, but you simply can't use your user profile home as the root of a git repo. You must specify an empty folder where the m1 user has write access. Regarding the target branch name, it has to be an existing branch of the git repo. Go to github, whatever you see as git branch name is a valid string to put there, release (latest stable release, currently fed up from 3.0-stable, will upgrade to 4.0 whenever released), 19-2.1 (old dev branch), 3.0-stable (the current stable branch), 3.1-dev (the current dev branch) or any other feature branch of whatever is not merged yet, including your own branches if you have push rights to the git repo.

If you would like to write some good doc about firstboot.json options would be just awesome.



Thanks for the tips. I don't have much time to devote to this unlike this time last year. I will say again what a great job you guys have done here and I realize how hard it is to be perfect, but you have to try, right?

So, I have it up and running on my test rig now and I just noticed that "nvoc start" now works in REMOTE mode. This is AWESOME as it gives one the ability to do lots of things like change OC, COIN, etc. on the fly without a reboot. Start never did work in 19-2.0 when in REMOTE but it worked fine for LOCAL. I am very happy to see you guys got it working fully.
newbie
Activity: 2
Merit: 0
Hey there. I've gotta a problem while mining XMR. It shows : Result rejected by the pool.
Im running nvOC_v0019-1.4

Anyone had this?
jr. member
Activity: 128
Merit: 1
@stubo every time you open a gnome-terminal to its default profile nvOC runs again from scratch, it is like this from fullzero nvOC autostart method design. If you just need a shell open guake or lauch gnome-terminale with another profile. There is an alternative to gnome-terminal autostart based on systemd. Dig further in the code, the initial setup install may suffer some rough edges but in the end works like a charm, and the restart/stop commands are much more reliable.

The option in firstboot to use a different path works, but you simply can't use your user profile home as the root of a git repo. You must specify an empty folder where the m1 user has write access. Regarding the target branch name, it has to be an existing branch of the git repo. Go to github, whatever you see as git branch name is a valid string to put there, release (latest stable release, currently fed up from 3.0-stable, will upgrade to 4.0 whenever released), 19-2.1 (old dev branch), 3.0-stable (the current stable branch), 3.1-dev (the current dev branch) or any other feature branch of whatever is not merged yet, including your own branches if you have push rights to the git repo.

If you would like to write some good doc about firstboot.json options would be just awesome.

newbie
Activity: 40
Merit: 0
How X16R start with T_Rex ?

In 1bash set x16r miner to T_Rex

Code:
X16R_MINER="T_Rex"

I did so, but mining did not start then. And what do you recommend for X16r ? ZENEMYminer or T_Rex ?
member
Activity: 224
Merit: 13
"release" branch is always same as latest stable branch which is now "3.0-stable"
You can set nvoc_branch to "3.0-stable" for 3.0 in firstboot.json
or "release" which will get updated when new stable branch get released.

Try to run a firstboot with its defaults and see how it goes

Edit:
These problems occur when you leave the team for too long  Wink


Yep, it has been a while. I used the defaults and that worked fine. But when I try to use firstboot to set nvoc_path to just /home/m1, it goes craptastic (same errors as post above) and I am left with a base Ubuntu install. Why give people the option to set it if it doesn't work? The recompile miners and expand disk options did work fine, however.

The other bug that I wanted to point out is that if while running REMOTE, on the console, every time I open a terminal it tries to re-run nvOC. I didn't dig into the details but whatever launches that should check to see if it is already running. All that I wanted to do was open a terminal to troubleshoot miner startup issues right there on the console without having to ssh in from another machine.

Enough with the bad, what is really cool is the reorganization of 1bash and the total rewrite/streamlining of 0miner. It took me a few minutes to figure out what is going on with 0miner as it is such a deviation from the very repetitive old version. Kudos on that. The downside is that adding a new miner that doesn't fit the mold of an existing algo is quite cumbersome. However, I don't figure most people would ever need to do that.

But overall, great job to the team. I am pleased to see how much progress you guys have made.
newbie
Activity: 8
Merit: 0
hello
I have a problem with the temperature control on the 18.04 lts version miscellaneous 410 updated
when I change currency the 6tempcontrol no longer increases the speed of the fans
my solution for now is to restart nvoc ( nvoc reboot) to have this control
Thanks for the help.


We are aware the problem in ubuntu 18.04 with tempcontrol and have not found the solution yet.


Ok
I am also looking for a solution but without success for the moment
jr. member
Activity: 112
Merit: 3
Linux Forever... Resistance is futile!!!
hello
I have a problem with the temperature control on the 18.04 lts version miscellaneous 410 updated
when I change currency the 6tempcontrol no longer increases the speed of the fans
my solution for now is to restart nvoc ( nvoc reboot) to have this control
Thanks for the help.


We are aware the problem in ubuntu 18.04 with tempcontrol and have not found the solution yet.
I will create a 16.04 Dual Cuda image as soon as I get some free time until we found the solution for 18.04.
newbie
Activity: 8
Merit: 0
hello
I have a problem with the temperature control on the 18.04 lts version miscellaneous 410 updated
when I change currency the 6tempcontrol no longer increases the speed of the fans
my solution for now is to restart nvoc ( nvoc reboot) to have this control
Thanks for the help.
jr. member
Activity: 112
Merit: 3
Linux Forever... Resistance is futile!!!
So in the new version what does the ALGO_SPECIFIC mean ? Does it automatically set the overclock ?

Its same as before, just set your algo desired oc values in "ALGO_SPECIFIC Power Limit and Clock settings" section
Set the OC and PL to ALGO_SPECIFIC
And when you switch coins oc will change with your coin.
jr. member
Activity: 112
Merit: 3
Linux Forever... Resistance is futile!!!
Hey Pap and All:

It has been a long time and it would appear that lots of work has been done.

What is the best way to get started from nothing and get up to the latest version? My rigs are all on 19.2.0 and I just want to start over and hopefully get the latest Ubuntu and Cuda toolkit. I want to play around with mining RVN and the latest Z-enemy miner seems to require at least Cuda 9 and I am stuck at Cuda 8 with my older versions.

Thanks in advance.

Hi mate
Long time...

Have a look at nvOC GitHub readme if you want to make a new image or check This Post to just get the latest nvOC
Both 16.04 and 18.04 images download and set latest release point at first boot.

There is some issue on ubuntu 18.04 with tempcontrol which fans wont update and we gets multiple instances of it on some rigs after running for a while and we cant figure out why it happens.
My test rig with 3 cards shows no problem, but rigs with 6+ cards does.

If you have time to install 18.04 and see if you can pinpoint the problem would be a big help

I followed the directions as per the README.md, specifically "For insiders: Pre-built images targeting beta/pre-release branches". I couldn't find good documentation for the firstboot.json mods. I had a question about the release naming and could not find it anywhere. So I ended up with a bunch of failures:
Code:
Thu Oct 25 03:07:35 EDT 2018 - nvOC FirstBoot start

 + Looking for the small fat partition
/media/m1/12D3-A869 is not a mountpoint
  ++ mounting fat partition

 + Parsing firstboot.json
  ++ selected branch: '3.0'
  ++ nvOC will install to: '/home/m1'

 + Wait for internet connection to continue nvOC setup
Connection to google.com 443 port [tcp/https] succeeded!

 + Cloning '3.0' nvOC branch into /home/m1
  ++ Checking if selected branch actually exists...
   +++ Selected branch not found, falling back to 'release'
  ++ /home/m1 already exists, cleaning up
rm: cannot remove '/home/m1/.config/enchant': Permission denied
rm: cannot remove '/home/m1/firstboot/tmplogs': Device or resource busy
fatal: destination path '/home/m1' already exists and is not an empty directory.

 + Updating miners submodule
  ++ No cache found in /home/m1/.miners-cache
fatal: not a git repository (or any of the parent directories): .git

 + Checking free space
  ++ Available free space in root partition: 100773288 KB

 + Installing miners (recompile: true)
/home/m1/firstboot/firstboot.sh: line 118: pushd: /home/m1/miners: No such file or directory
bash: nvOC_miner_update.sh: No such file or directory
/home/m1/firstboot/firstboot.sh: line 129: popd: directory stack empty

 + Looking for your customized 1bash
dos2unix: converting file /home/m1/1bash to Unix format...

 + Setting 2unix as custom-command for gnome-terminal 'mining' profile
bash: /home/m1/firstboot/profile-manager.sh: No such file or directory

 + Determining if firstboot can be disabled
  ++ FAILURE: keeping firstboot as default gnome-terminal profile
  ++ This script will run again on the next reboot.
  ++ Check your fat partition contents or internet connectivity.

  + Saving firstboot_90ccfd.log to small fat partition and /home/m1/firstboot

Thu Oct 25 03:07:36 EDT 2018 - Done.
cp: cannot stat '/home/m1/firstboot/tmplogs/firstboot.log': No such file or directory
cp: cannot stat '/home/m1/firstboot/tmplogs/firstboot.log': No such file or directory

I used value "3.0" and home of "/home/m1" but that was a guess as it could have been lots and lots of things - "nvOC-19.3.0", nvoc-19_3.0" and on and on. It also looks like the script defaulted to "release" which was promising but still failed.

What is the secret release to use for this field? Also, is it problem to use home "/home/m1"? I don't like having superfluous extra subdirectories if not absolutely necessary.




"release" branch is always same as latest stable branch which is now "3.0-stable"
You can set nvoc_branch to "3.0-stable" for 3.0 in firstboot.json
or "release" which will get updated when new stable branch get released.

Try to run a firstboot with its defaults and see how it goes

Edit:
These problems occur when you leave the team for too long  Wink
jr. member
Activity: 112
Merit: 3
Linux Forever... Resistance is futile!!!
How X16R start with T_Rex ?

In 1bash set x16r miner to T_Rex

Code:
X16R_MINER="T_Rex"
member
Activity: 224
Merit: 13
Hey Pap and All:

It has been a long time and it would appear that lots of work has been done.

What is the best way to get started from nothing and get up to the latest version? My rigs are all on 19.2.0 and I just want to start over and hopefully get the latest Ubuntu and Cuda toolkit. I want to play around with mining RVN and the latest Z-enemy miner seems to require at least Cuda 9 and I am stuck at Cuda 8 with my older versions.

Thanks in advance.

Hi mate
Long time...

Have a look at nvOC GitHub readme if you want to make a new image or check This Post to just get the latest nvOC
Both 16.04 and 18.04 images download and set latest release point at first boot.

There is some issue on ubuntu 18.04 with tempcontrol which fans wont update and we gets multiple instances of it on some rigs after running for a while and we cant figure out why it happens.
My test rig with 3 cards shows no problem, but rigs with 6+ cards does.

If you have time to install 18.04 and see if you can pinpoint the problem would be a big help

I followed the directions as per the README.md, specifically "For insiders: Pre-built images targeting beta/pre-release branches". I couldn't find good documentation for the firstboot.json mods. I had a question about the release naming and could not find it anywhere. So I ended up with a bunch of failures:
Code:
Thu Oct 25 03:07:35 EDT 2018 - nvOC FirstBoot start

 + Looking for the small fat partition
/media/m1/12D3-A869 is not a mountpoint
  ++ mounting fat partition

 + Parsing firstboot.json
  ++ selected branch: '3.0'
  ++ nvOC will install to: '/home/m1'

 + Wait for internet connection to continue nvOC setup
Connection to google.com 443 port [tcp/https] succeeded!

 + Cloning '3.0' nvOC branch into /home/m1
  ++ Checking if selected branch actually exists...
   +++ Selected branch not found, falling back to 'release'
  ++ /home/m1 already exists, cleaning up
rm: cannot remove '/home/m1/.config/enchant': Permission denied
rm: cannot remove '/home/m1/firstboot/tmplogs': Device or resource busy
fatal: destination path '/home/m1' already exists and is not an empty directory.

 + Updating miners submodule
  ++ No cache found in /home/m1/.miners-cache
fatal: not a git repository (or any of the parent directories): .git

 + Checking free space
  ++ Available free space in root partition: 100773288 KB

 + Installing miners (recompile: true)
/home/m1/firstboot/firstboot.sh: line 118: pushd: /home/m1/miners: No such file or directory
bash: nvOC_miner_update.sh: No such file or directory
/home/m1/firstboot/firstboot.sh: line 129: popd: directory stack empty

 + Looking for your customized 1bash
dos2unix: converting file /home/m1/1bash to Unix format...

 + Setting 2unix as custom-command for gnome-terminal 'mining' profile
bash: /home/m1/firstboot/profile-manager.sh: No such file or directory

 + Determining if firstboot can be disabled
  ++ FAILURE: keeping firstboot as default gnome-terminal profile
  ++ This script will run again on the next reboot.
  ++ Check your fat partition contents or internet connectivity.

  + Saving firstboot_90ccfd.log to small fat partition and /home/m1/firstboot

Thu Oct 25 03:07:36 EDT 2018 - Done.
cp: cannot stat '/home/m1/firstboot/tmplogs/firstboot.log': No such file or directory
cp: cannot stat '/home/m1/firstboot/tmplogs/firstboot.log': No such file or directory

I used value "3.0" and home of "/home/m1" but that was a guess as it could have been lots and lots of things - "nvOC-19.3.0", nvoc-19_3.0" and on and on. It also looks like the script defaulted to "release" which was promising but still failed.

What is the secret release to use for this field? Also, is it problem to use home "/home/m1"? I don't like having superfluous extra subdirectories if not absolutely necessary.


newbie
Activity: 96
Merit: 0
So in the new version what does the ALGO_SPECIFIC mean ? Does it automatically set the overclock ?
newbie
Activity: 40
Merit: 0
How X16R start with T_Rex ?
jr. member
Activity: 112
Merit: 3
Linux Forever... Resistance is futile!!!
Is there a reasons why i should change my Hive.Os and simplemining for this?

Just that nvOC is free and open source.
You can check all the codes and scripts in its GitHub.
I have no idea about Hive OS and simplemining.
jr. member
Activity: 112
Merit: 1
Look ARROUND!
Is there a reasons why i should change my Hive.Os and simplemining for this?
Pages:
Jump to: