[Support] lnxd - XMRig


83 posts in this topic Last Reply

Recommended Posts

12 hours ago, lnxd said:

Hey there,

 

Sorry for the delay, I missed your question. The script should be run on your Unraid server. (random) People (on the internet) are reporting around 1900H/s with your CPU for mining with RandomX. You can also try adding --asm intel to the Additional XMRig Arguments field like so:

307755228_ScreenShot2021-04-07at11_56_52am.thumb.png.e0835820f63d8481b888a1364691d566.png

With the user script, privileged mode and duplicating these arguments my rate went from 700 H/s to 7K H/s on a dual E5-2667 v2 rig, using :latest tag. At 700, I was seriously considering pulling the plug, just not worth it.

 

Tuning is key. 🤣

Link to post
  • Replies 82
  • Created
  • Last Reply

Top Posters In This Topic

Top Posters In This Topic

Popular Posts

Hey @horphi, this is quite difficult to set up but possible. Ravencoin uses the kawpow algorithm which is a GPU algorithm. To use XMRig you will need to go on the latest-root tag and re-install it fro

Overview: Support thread for lnxd/XMRig in CA. Application: XMRig - https://github.com/xmrig/xmrig Docker Hub: https://hub.docker.com/r/lnxd/xmrig GitHub: https://github.com/lnxd/docker

Thanks for the heads up @Squid! A very good call to do this after the recent wave of compromised servers with XMRig running. I'll update OP.    And @jonathanm that recommendation made me l

Posted Images

7 hours ago, jonathanm said:

With the user script, privileged mode and duplicating these arguments my rate went from 700 H/s to 7K H/s on a dual E5-2667 v2 rig, using :latest tag. At 700, I was seriously considering pulling the plug, just not worth it.

 

Tuning is key. 🤣

😂 tuning is definitely key. 

Link to post

What am I doing wrong... In log

* HUGE PAGES supported
* 1GB PAGES supported

 

randomx failed to allocate RandomX dataset using 1GB pages

 

Settings

 

--no-color --asm intel --randomx-1gb-pages

 

Running privileged

 

 

 

 

 

 

 

Link to post
Posted (edited)
13 minutes ago, G Speed said:

What am I doing wrong... In log


Near the top of the log it should show a line that looks like this:

 

* MEMORY 12.1/31.1 GB (38%)


What does yours say? The container likely reports that you have a higher RAM usage than you actually do.


I'm currently researching a similar issue for @SPOautos.  If you don't mind posting your diagnostics.zip it might help me find the cause a bit quicker 🙃

Edited by lnxd
Link to post
9 hours ago, lnxd said:


Near the top of the log it should show a line that looks like this:

 



* MEMORY 12.1/31.1 GB (38%)


What does yours say? The container likely reports that you have a higher RAM usage than you actually do.


I'm currently researching a similar issue for @SPOautos.  If you don't mind posting your diagnostics.zip it might help me find the cause a bit quicker 🙃

 

* ABOUT XMRig/6.10.0 gcc/9.3.0
* LIBS libuv/1.34.2 OpenSSL/1.1.1f hwloc/2.1.0
* HUGE PAGES supported
* 1GB PAGES supported
* CPU Intel(R) Core(TM) i3-8100 CPU @ 3.60GHz (1) 64-bit AES
L2:1.0 MB L3:6.0 MB 4C/4T NUMA:1
* MEMORY 11.2/15.5 GB (72%)

 

Without XMrig running, i'm only at 15% ram utilization
So yeah.. something wrong

 

Edited by G Speed
Link to post
Posted (edited)
19 hours ago, G Speed said:

Without XMrig running, i'm only at 15% ram utilization
So yeah.. something wrong

Can you please stop xmrig and post the output of:

free -m


And also show me a screenshot of your RAM usage from the dashboard:
688459108_ScreenShot2021-04-09at8_18_40am.thumb.png.095528375574ca311980c4aedf5fa8e5.png

Also, please start the container again and post the output of this if there's nothing in there you don't want to share on the forum:

top -n 1 -o %MEM

 

What you just posted shows that the container can see 1 NUMA node, so the RAM requirement for 1GB Huge Pages on your host would be 3GB * 1 = 3GB, and you don't have enough RAM available to the container when it is launching. So there's something else that's unexpectedly being allocated RAM and preventing XMrig from being able to access it, I'm guessing it's QEMU but I'll await your response 🙂

If you think XMRig is using more RAM than it should, you can add eg. --memory=4g to Advanced > Extra Parameters on the Edit screen to limit it to eg. 4gb like so:
1539471591_ScreenShot2021-04-09at10_41_29am.thumb.png.97fc90f9d5c6e81bccf8ee6c7b07578d.png

 

17 hours ago, horphi said:

@lnxd do you plan to activate cuda-plugin for using gpu?

https://github.com/xmrig/xmrig-cuda

Probably not, it's much more profitable to mine Ethereum, it's kind of a waste GPU mining Monero. For GPU, check out PhoenixMiner in my signature, it works with both AMD and Nvidia cards. 

Edited by lnxd
Link to post
8 hours ago, G Speed said:

 

* ABOUT XMRig/6.10.0 gcc/9.3.0
* LIBS libuv/1.34.2 OpenSSL/1.1.1f hwloc/2.1.0
* HUGE PAGES supported
* 1GB PAGES supported
* CPU Intel(R) Core(TM) i3-8100 CPU @ 3.60GHz (1) 64-bit AES
L2:1.0 MB L3:6.0 MB 4C/4T NUMA:1
* MEMORY 11.2/15.5 GB (72%)

 

Without XMrig running, i'm only at 15% ram utilization
So yeah.. something wrong

 

 

My XMrig log was also showing real high ram usage....like 26GB out of 32GB but it was related to me having Plex transcoding in ram and I had a large chunk of my ram allocated for that purpose so XMrig was seeing that and considering it as being used eventhough my actual usage was only like 7GB. 

 

HOWEVER, that said, even after I made some changes and XMrig only sees my usage as like 7GB out of 32GB, I still cant use the 1GB pages. For some reason RandomX 1GB pages fails. Not sure why.

 

Anyway, are you doing Plex transcoding in ram? If so, pull up the script, click edit, see how much ram you have allocated to it.....I had way more than I needed allocated so I changed it to like 4GB then restarted my array so it would change (just running the script didnt seem to work but stopping then starting the array did change how much was allocated). 

Link to post
Posted (edited)
1 hour ago, SPOautos said:

 

My XMrig log was also showing real high ram usage....like 26GB out of 32GB but it was related to me having Plex transcoding in ram and I had a large chunk of my ram allocated for that purpose so XMrig was seeing that and considering it as being used eventhough my actual usage was only like 7GB. 

 

HOWEVER, that said, even after I made some changes and XMrig only sees my usage as like 7GB out of 32GB, I still cant use the 1GB pages. For some reason RandomX 1GB pages fails. Not sure why.

 

Anyway, are you doing Plex transcoding in ram? If so, pull up the script, click edit, see how much ram you have allocated to it.....I had way more than I needed allocated so I changed it to like 4GB then restarted my array so it would change (just running the script didnt seem to work but stopping then starting the array did change how much was allocated). 

@G Speed In addition to what @SPOautos mentioned, another likely cause is kernel caching. Might also be worth dropping your caches by running the following in terminal:
 

echo "Before:" && free -h && sync; echo 3 > /proc/sys/vm/drop_caches && sleep 10 && echo "After:" && free -h

 

There's a real chance you'd be fine starting XMRig with 1gb Large Pages soon after a reboot. But the Kernel caches regularly accessed things to RAM for the purposes of speed & efficiency. XMRig just asks for the RAM with a lower priority and its request gets denied. The command above forces the server to drop these caches, so XMRig has less competition when it tries to launch with 1gb Huge Pages. The kernel will then rebuild them as necessary.

On any other OS this would cause a (probably unnoticeable on a modern PC) drop in performance in the short term, but keep in mind that on Unraid the Kernel is run from memory anyway so it shouldn't impact much at all. You'll also likely notice that free RAM visible from terminal (free -h) is different to free RAM visible on the Dashboard, the Dashboard seems to ignore cached data in RAM. I'm assuming this choice was made because the system will automatically drop it if you ask it to. Eg. if you try to use that RAM to start a VM it just takes that space in your memory, so it isn't useful information for most users.


I'm going to make some updates to OP as well.

Edited by lnxd
Link to post
14 hours ago, lnxd said:

Can you please stop xmrig and post the output of:


free -m


And also show me a screenshot of your RAM usage from the dashboard:
688459108_ScreenShot2021-04-09at8_18_40am.thumb.png.095528375574ca311980c4aedf5fa8e5.png

Also, please start the container again and post the output of this if there's nothing in there you don't want to share on the forum:


top -n 1 -o %MEM

 

What you just posted shows that your CPU has one NUMA node, so the RAM requirement for 1GB Huge Pages on your host would be 3GB * 1 = 3GB, and you don't have enough RAM available to the container when it is launching. So there's something else that's unexpectedly being allocated RAM and preventing XMrig from being able to access it, I'm guessing it's QEMU but I'll await your response 🙂 If you think XMRig is using more RAM than it should, you can add eg. --memory=4g to Advanced > Extra Parameters on the Edit screen to limit it to 4gb like so:
1539471591_ScreenShot2021-04-09at10_41_29am.thumb.png.97fc90f9d5c6e81bccf8ee6c7b07578d.png

 

Probably not, it's much more profitable to mine Ethereum, it's kind of a waste GPU mining Monero. For GPU, check out PhoenixMiner in my signature, it works with both AMD and Nvidia cards. 

 

At work right now, I will do when I get home, but to answer some of your question
Correct, I have it limited to one numa mode

With xmrig off that screen shot would show 15% usage
In the docker page, when xmrig is running it's using approx 2.Xgb



 

Link to post
Posted (edited)
52 minutes ago, G Speed said:

 

At work right now, I will do when I get home, but to answer some of your question
Correct, I have it limited to one numa mode

With xmrig off that screen shot would show 15% usage
In the docker page, when xmrig is running it's using approx 2.Xgb

No worries, and between 2-3gb sounds about right. It's like I said, the dash doesn't report allocated RAM just used RAM, whereas XMRig will report allocated RAM + used RAM. So if the dash shows around 15% before XMRig launches, and XMRig shows 72% when it launches (and is only sitting at around 300mb overhead), there's something unaccounted for that at least one of the above steps should solve.

I'm thinking once you force your host to drop caches XMRig should have a fighting chance. As you noted the other day though, even with your huge pages set to 1gb, we're only talking a 1-3% increase in efficiency / hash rate. You said you have an Intel CPU and you limited it to 1 NUMA node. I'm guessing you have a Xeon/s then? If that's the case, your hash rate will likely benefit more from changing to the latest-root tag so that XMRig can see what hardware you really have and optimise things accordingly.  

XMRig reports how much RAM it's using a few lines down from what you shared in the logs, Eg.:


1gb Huge Pages Enabled with 1 NUMA node:

randomx allocated 3072 MB (2080+256) huge pages 100% 3/3 +JIT (144 ms)


1gb Huge Pages Disabled with 1 NUMA node:

randomx allocated 2336 MB (2080+256) huge pages 100% 1168/1168 +JIT (241 ms)

 

The container will use what it says there + around 300mb.

Edited by lnxd
Link to post

actually it's just an i3 on one of my rigs

 

xmrig off

              total        used        free      shared  buff/cache   available
Mem:          15884        1906        2546         738       11431       12931
Swap:             0           0           0

Xmrig on

 

              total        used        free      shared  buff/cache   available
Mem:          15884        4268         197         738       11418       10571
Swap:             0           0           0

 

Edited by G Speed
Link to post
3 hours ago, G Speed said:

actually it's just an i3 on one of my rigs

 

xmrig off


              total        used        free      shared  buff/cache   available
Mem:          15884        1906        2546         738       11431       12931
Swap:             0           0           0

Xmrig on

 


              total        used        free      shared  buff/cache   available
Mem:          15884        4268         197         738       11418       10571
Swap:             0           0           0

 

Haha yeah sorry I was a bit tired when I asked, dumb question because it says it's an i3 in your logs. But yeah drop your caches and then restart the container:
 

18 hours ago, lnxd said:

 


echo "Before:" && free -h && sync; echo 3 > /proc/sys/vm/drop_caches && sleep 10 && echo "After:" && free -h

 

 

Link to post

For those who have dual Intel Xeon E5 2650-3 (or maybe others Xeon CPU too) I remove the first 6 cores and their threads and now I have 5000-5500H/s instead of ~3500H/s or less on anything else I tried:

What I tried = 1 more core/threads plus or minus, all of them, only the first core/thread removed like the OP said, and more.

p.s. I also use the OP script and this as additional arguments:

--no-color --asm intel --randomx-1gb-pages

 

If that can help anyone 😎

 

Thanks for putting this docker image in CA 😃, it's really appreciated.

I saw you tell people that it's not worth it to put GPU support for it, but ethereum have gone beyond the 4gb mark,lots of people still have 4gb GPU it may be nice to have the possibility to mine monero since we can't do ethereum anymore (maybe with a hack, but probably not recommended). It may not be that profitable, but hey, were doing that for fun mostly.

The decision is yours to take, that was just my 2 cents ;) 

Edited by Steace
typo
Link to post
Posted (edited)
12 minutes ago, Steace said:

Thanks for putting thit docker image in CA 😃, it's really appreciated.

I saw you tell people that it's not worth it to put GPU support for it, but ethereum have gone beyond the 4gb mark,lots of people still have 4gb GPU it may be nice to have the possibility to mine monero since we can't do ethereum anymore (maybe with a hack, but probably not recommended). It may not be that profitable, but hey, were doing that for fun mostly.

The decision is yours to take, that was just my 2 cents ;) 

Thanks for all the suggestions in your post! Super valuable information for people with Xeons.

If I'm honest about GPU Mining Monero, the environmentalist inside me kicks in and says it's a waste of electricity 😅 My hash rate with my RX580 with XMRig/RandomX is around 900H/s which is around $2.56AUD/month at current rates, but the cost at 80w is around $12.85AUD/month worth of electricity.

I might write up a tag for people who are left out (and are lucky to have free electricity). It just requires one extra XMRig build so that it works with Nvidia cards + adding the AMD drivers which I already have a base for.

Edited by lnxd
Link to post
4 minutes ago, lnxd said:

Thanks for all the suggestions in your post! Super valuable information for people with Xeons.

If I'm honest about GPU Mining Monero, the environmentalist inside me kicks in and says it's a waste of electricity 😅 My hash rate with my RX580 with XMRig/RandomX is around 900H/s which is around $2.56AUD/month at current rates, but the cost at 80w is around $12.85AUD/month worth of electricity.

I might write up a tag for people who are left out (and are lucky to have free electricity). It just requires one extra XMRig build so that it works with Nvidia cards + adding the AMD drivers which I already have a base for.

Ouch... I just googled electricity cost in Australia, 20 to 33.c/kWh depending on state. I pay 0.07c/kWh. That's probably why you care and I don't :P 

Link to post
Posted (edited)
24 minutes ago, Steace said:

Ouch... I just googled electricity cost in Australia, 20 to 33.c/kWh depending on state. I pay 0.07c/kWh. That's probably why you care and I don't :P 

Haha yeah it's super expensive here, I calculated that at a rate of $0.21AUD/kWh. I have flat rate electricity (included in my rent) so I can play around a bit, but it would have to cost someone less than $0.035AUD/kWh to be profitable with an RX 580.

But hey, it's max 14c (57f) here today, and mining heats two rooms of my house 😂

Edited by lnxd
Link to post
30 minutes ago, lnxd said:

But hey, it's max 14c (57f) here today, and mining heats two rooms of my house 😂

Which means it's simply like running resistive heat strips, or a space heater. If you need the heat anyway, it's not really a waste, except that a heat pump is a little more efficient. When summer comes around however, you'd be paying twice, once to mine and once to move the heat outside with the A/C.

 

So rest your environmental head, as long as it's cold outside and you would normally run a heater, mine away. At the very least it mines more coins than a baseboard heater.

Link to post
15 hours ago, Steace said:

For those who have dual Intel Xeon E5 2650-3 (or maybe others Xeon CPU too) I remove the first 6 cores and their threads and now I have 5000-5500H/s instead of ~3500H/s or less on anything else I tried:

What I tried = 1 more core/threads plus or minus, all of them, only the first core/thread removed like the OP said, and more.

p.s. I also use the OP script and this as additional arguments:


--no-color --asm intel --randomx-1gb-pages

 

If that can help anyone 😎

 

Thanks for putting this docker image in CA 😃, it's really appreciated.

I saw you tell people that it's not worth it to put GPU support for it, but ethereum have gone beyond the 4gb mark,lots of people still have 4gb GPU it may be nice to have the possibility to mine monero since we can't do ethereum anymore (maybe with a hack, but probably not recommended). It may not be that profitable, but hey, were doing that for fun mostly.

The decision is yours to take, that was just my 2 cents ;) 

 

I have a xeon e5-2690v3 and had --asm intel and --randomx-1gb-pages but I could never seem to get the 1gb pages to work. When I removed those two arguements my H/s actually went UP about 200 H/s. I have no idea why.

 

Have you done before/after to see what difference they are making? Also, can you tell in your container log that your actually getting 1gb pages?

Link to post
Posted (edited)
4 hours ago, Squid said:

As an FYI, users with this container installed will have FCP issue a warning.  Since you've got this container running with the explicit purpose to run xmrig, this warning is quite safe to ignore.

Thanks for the heads up @Squid! A very good call to do this after the recent wave of compromised servers with XMRig running. I'll update OP. 
 

Quote

xmrig is currently running in your server. If you are purposely running mining software then this warning is safe to ignore. If you are not then your system has been possibly compromised (or you have installed a random docker via CA's dockerHub search that contains mining software


And @jonathanm that recommendation made me laugh this morning, it was one of the first things I read when I woke up and it's very true 🤣 
 

6 hours ago, Steace said:

Hi, it's me again :) 

How and where do we put our email/password from nanopool?

I just found out that we need that to be configured to change the min/max payout

Hey @Steace!

 

Yup you sure do. Otherwise it'll be a looooonngg time to payout. Just make sure you note their transaction fee when you're reducing the value.

If you slip it into the worker field it should work, eg. Unraid/test@example.com. Then just let it mine for a while before trying to update the payout threshold on the Nanopool website.
worker.thumb.png.b8471713895b7173d8e3acc838e4facf.png

Edited by lnxd
Link to post
Posted (edited)

Due to overwhelming demand (3 people 😂), I've just released a new version works with both AMD and likely Nvidia GPUs as well. You'll need to re-add the container from CA and use the repository to lnxd/xmrig:latest-root to use the new features; it'll probably get picked up some time within the next 2 hours. I haven't got an Nvidia GPU to test it with so if someone gets it working with one, please let me know. But be patient, it takes around 10 mins to install Cuda for me before it gets to the mining stage, I'm gonna work on reducing that when I get a chance.

 

It now includes XMRig 6.11.1 binaries made as part of the docker build process rather than separately, and it's 26mb smaller than before even with the optional CUDA library layer, probably because I got better at writing Dockerfiles? Not sure 😅 In case anyone is worried about bloat, the container doesn't contain the drivers by default. If you want to use them you need to select a version and it will install them.


PS. Sorry, I hate to double post rather than just edit. But I thought people monitoring this thread might be interested to see without getting an annoying alert.

Except @Steace

Edited by lnxd
Link to post

Hello @lnxd,

 

is it also possible, to mine https://rvn.nanopool.org/?

 

The pre-settings are showing --coin=monero

 

Where can i change this value?

 

Running xmrig as uid=0(root) gid=0(root) groups=0(root) with the following flags:
--url=stratum.ravenminer.com:3838 --coin=monero --user=REvztdeD9c1Z3Nc2er5gB7Y9ZM4PUqoxUY.Unraid --randomx-wrmsr=-1 --randomx-no-rdmsr --no-color --asm intel --random-1gb-pages
./xmrig: unrecognized option '--random-1gb-pages'
* ABOUT XMRig/6.10.0 gcc/9.3.0
* LIBS libuv/1.34.2 OpenSSL/1.1.1f hwloc/2.1.0
* HUGE PAGES supported
* 1GB PAGES disabled
* CPU Intel(R) Xeon(R) CPU E7- 8870 @ 2.40GHz (4) 64-bit AES
L2:10.0 MB L3:120.0 MB 40C/80T NUMA:4
* MEMORY 14.5/125.9 GB (11%)
DIMM A1: 16 GB DDR3 @ 1066 MHz M393B2K70CM0-YF8
DIMM A2: 16 GB DDR3 @ 1066 MHz HMT42GR7CMR4A-G7
DIMM B1: 16 GB DDR3 @ 1066 MHz HMT42GR7CMR4A-G7
DIMM B2: 16 GB DDR3 @ 1066 MHz HMT42GR7CMR4A-G7
DIMM C1: 16 GB DDR3 @ 1066 MHz M393B2K70CM0-YF8
DIMM C2: 16 GB DDR3 @ 1066 MHz HMT42GR7CMR4A-G7
DIMM D1: 16 GB DDR3 @ 1066 MHz M393B2K70CM0-YF8
DIMM D2: 16 GB DDR3 @ 1066 MHz HMT42GR7CMR4A-G7
* MOTHERBOARD Dell Inc. - 0FJM8V
* DONATE 1%
* ASSEMBLY intel
* POOL #1 stratum.ravenminer.com:3838 coin monero
* COMMANDS 'h' hashrate, 'p' pause, 'r' resume, 's' results, 'c' connection
* OPENCL disabled
* CUDA disabled

 

BR

 

Phil

Link to post

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.