xmrig Monero Dwarfpool Image Testing

[Patrick]

Are you running it in -d mode?

 

[RBE]

Only if the -d stands for dunce mode. I have modified my Portainer service so that it only passes in the 'username' environment variable. Not actually sure how you go about including the '-itd' runtime options via Portainer. As a result, any time I want to see what is going on with a particular node, I simply ssh to the node and type:

docker attach --sig-proxy=false <container-id>

where the 'sig-proxy' bit is to enable me to use [CTRL+C] to detach from the container without stopping it.

Edit: Found the problem. When reading the introductory post I failed to realise that the email environment variable is NOT optional. I simply assumed it was, based on my experience with the image that uses Wolf's miner . I have fixed this and now we are off to the races. I have added the hard and soft memory limits as per the xmr-stak-cpu readme file, and will see what the hashrate is like overnight.

 

[Algeroth]

I

GRUB_CMDLINE_LINUX_DEFAULT="transparent_hugepage=always"

then run "update_grub". After a reboot, running the command:

$ cat /sys/kernel/mm/transparent_hugepage/enabled

returns

[always] madvise never

in the basics transparent_hugepages != hugepages
transparent hugepages had in the past massive negative impact on performance degradation....

to enable huge pages all You have to do in modern OS (linux) is just put
sysctl -w vm.nr_hugepages=X
into wahtsoever starting script
or in /etc/sysctl.conf...

regards
A

 

[RBE]

@Algeroth Thank you for pointing out the difference. Some further digging turned up this helpful primer, courtesy of RedHat. The question I have now is, what is an appropriate value for X? I note that xmr-stak-cpu recommends allocating 128 huge pages in README.md, but there is no similar recommendation in the xmrig documentation, so for now I will go with 128.

 

[RBE]

Having figured out the whole huge pages thing, I set up a new Portainer service using numthreads=20 and startport=8100, and here are the results from the first 16 BSMC nodes:



Not bad I guess - a slight improvement on Wolf's miner for sure. It will be interesting to see how the hashrate reported by Dwarfpool changes over the next few days. There are always daily/hourly variations in the reported hashrate, but hopefully a clear trend will emerge over time. Thanks to all those who pointed me in the right direction!

 

[Patrick]

2x AMD EPYC 7301 = 2954H/s 311W

 

[Patrick]

2x AMD EPYC 7351 = 3074H/s

 

[spoutman]

Dear
I have AMD Ryzen 7 1700X ( Ubuntu 16.04 minimul ) Server and install Docker.
wget -qO- https://get.docker.com/ | sh
I am using root account.

1. docker run -itd -e username=$wally -e email=x servethehome/monero_dwarfpool:zen
[2018-01-27 22:25:16] accepted: 1/1 (100.00%), 527.06 H/s at diff 100002 (yay!!!)

Speed is OK around 520 H/s


2.docker run -it -e username=$wally -e email=x -e numthreads=32 -e --cpuset-cpus="0-16" servethehome/monero_dwarfpool:xmrig
* VERSIONS: XMRig/2.3.1 libuv/1.9.1 gcc/6.3.0
* HUGE PAGES: available, enabled
* CPU: AMD Ryzen 7 1700X Eight-Core Processor (1) x64 AES-NI
* CPU L2/L3: 4.0 MB/16.0 MB
* THREADS: 16, cryptonight, av=1, donate=0%
* POOL #1: xmr-usa.dwarfpool.com:8050
* COMMANDS: hashrate, pause, resume
[2018-01-27 22:02:13] speed 2.5s/60s/15m 71.1 189.6 n/a H/s max: 189.1 H/s
[2018-01-27 22:03:13] speed 2.5s/60s/15m 142.2 189.4 n/a H/s max: 189.2 H/s

Speed very slow....around 190 H/S

lscpu
=================
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Byte Order: Little Endian
CPU(s): 16
On-line CPU(s) list: 0-15
Thread(s) per core: 2
Core(s) per socket: 8
Socket(s): 1
NUMA node(s): 1
Vendor ID: AuthenticAMD
CPU family: 23
Model: 1
Model name: AMD Ryzen 7 1700X Eight-Core Processor
Stepping: 1
CPU MHz: 2200.000
CPU max MHz: 3400.0000
CPU min MHz: 2200.0000
BogoMIPS: 6799.86
Virtualization: AMD-V
L1d cache: 32K
L1i cache: 64K
L2 cache: 512K
L3 cache: 8192K
NUMA node0 CPU(s): 0-15
=======================

xmrig speed too much slow.
Is there any way to make it fast ?

 

[spoutman]

Dear
I have AMD EPYC 7401 ( Ubuntu 17.10 . minimal ) Server and install Docker.
wget -qO- https://get.docker.com/ | sh
I am using root account.

1. docker run -itd -e username=$wally -e email=x servethehome/monero_dwarfpool:zen
Result ==>
[2018-01-27 21:37:00] accepted: 1/1 (100.00%), 1193.44 H/s at diff 100002 (yay!!!)
[2018-01-27 21:41:13] Stratum detected new block
[2018-01-27 21:41:18] accepted: 2/2 (100.00%), 1189.19 H/s at diff 100002 (yay!!!)
Very slow around 1190 H/s

2. docker run -it -e username=$wally -e email=x -e numthreads=96 -e startport=8100 --cpuset-cpus="0-47" servethehome/monero_dwarfpool:xmrig
* VERSIONS: XMRig/2.3.1 libuv/1.9.1 gcc/6.3.0
* HUGE PAGES: available, enabled
* CPU: AMD EPYC 7401P 24-Core Processor (1) x64 AES-NI
* CPU L2/L3: 12.0 MB/64.0 MB
* THREADS: 96, cryptonight, av=1, donate=0%
* POOL #1: xmr-usa.dwarfpool.com:8100
* COMMANDS: hashrate, pause, resume
[2018-01-27 22:15:15] use pool xmr-usa.dwarfpool.com:8100 66.70.234.210
[2018-01-27 22:15:15] new job from xmr-usa.dwarfpool.com:8100 diff 100001
[2018-01-27 22:16:18] speed 2.5s/60s/15m 115.9 562.9 n/a H/s max: 222.9 H/s
[2018-01-27 22:17:01] accepted (1/0) diff 100001 (125 ms)
[2018-01-27 22:17:18] speed 2.5s/60s/15m 61.4 554.8 n/a H/s max: 222.9 H/s
[2018-01-27 22:18:18] speed 2.5s/60s/15m 47.6 526.9 n/a H/s max: 222.9 H/s
[2018-01-27 22:19:18] speed 2.5s/60s/15m 151.4 538.6 n/a H/s max: 222.9 H/s

xmrig Too much slow......


lscpu
==================
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Byte Order: Little Endian
CPU(s): 48
On-line CPU(s) list: 0-47
Thread(s) per core: 2
Core(s) per socket: 24
Socket(s): 1
NUMA node(s): 4
Vendor ID: AuthenticAMD
CPU family: 23
Model: 1
Model name: AMD EPYC 7401P 24-Core Processor
Stepping: 2
CPU MHz: 1200.000
CPU max MHz: 2000.0000
CPU min MHz: 1200.0000
BogoMIPS: 3992.28
Virtualization: AMD-V
L1d cache: 32K
L1i cache: 64K
L2 cache: 512K
L3 cache: 8192K
NUMA node0 CPU(s): 0-5,24-29
NUMA node1 CPU(s): 6-11,30-35
NUMA node2 CPU(s): 12-17,36-41
NUMA node3 CPU(s): 18-23,42-47
=======================

xmrig is too much slow and other also slow.
Would you give me any good Idea to fix this problem ?

 

[mumford]

reduce # of threads. Try 1 thread per 2M of L3 cache.

numthreads=32 for 64M L3

 

[Patrick]

@spoutman try this for a single socket EPYC 7401

#!/bin/bash
wally=<your wallet>
emal=<your email>
ythreads=8
docker run -itd --privileged -e username=$wally.$(hostname)-1 -e email=$emal -e numthreads=$ythreads --cpuset-cpus="0-5,24,27" servethehome/monero_dwarfpool:xmrig
docker run -itd --privileged -e username=$wally.$(hostname)-2 -e email=$emal -e numthreads=$ythreads --cpuset-cpus="6-11,30,33" servethehome/monero_dwarfpool:xmrig
docker run -itd --privileged -e username=$wally.$(hostname)-3 -e email=$emal -e numthreads=$ythreads --cpuset-cpus="12-17,36,39" servethehome/monero_dwarfpool:xmrig
docker run -itd --privileged -e username=$wally.$(hostname)-4 -e email=$emal -e numthreads=$ythreads --cpuset-cpus="18-23,42,45" servethehome/monero_dwarfpool:xmrig

You want 4x instances, one for each NUMA node. Each instance you want to use L3 cache / 2 so 16MB/2 = 8. You then want to prioritize physical cores.

 

[spoutman]

It seems good working.
VERSIONS: XMRig/2.3.1 libuv/1.9.1 gcc/6.3.0
* HUGE PAGES: available, enabled
* CPU: AMD EPYC 7401P 24-Core Processor (1) x64 AES-NI
* CPU L2/L3: 12.0 MB/64.0 MB
* THREADS: 8, cryptonight, av=1, donate=0%
* POOL #1: xmr-usa.dwarfpool.com:8050
* COMMANDS: hashrate, pause, resume
[2018-01-28 08:20:34] use pool xmr-usa.dwarfpool.com:8050 144.217.61.241
[2018-01-28 08:20:34] new job from xmr-usa.dwarfpool.com:8050 diff 50000
[2018-01-28 08:20:40] accepted (1/0) diff 50000 (135 ms)
[2018-01-28 08:21:37] speed 2.5s/60s/15m 413.5 412.4 n/a H/s max: 413.0 H/s
[2018-01-28 08:21:51] accepted (2/0) diff 50000 (125 ms)
[2018-01-28 08:22:23] new job from xmr-usa.dwarfpool.com:8050 diff 50000
[2018-01-28 08:22:37] speed 2.5s/60s/15m 413.9 412.1 n/a H/s max: 414.0 H/s

4 dockers around 415 H/s

Thanks to help.

@spoutman try this for a single socket EPYC 7401

#!/bin/bash
wally=<your wallet>
emal=<your email>
ythreads=8
docker run -itd --privileged -e username=$wally.$(hostname)-1 -e email=$emal -e numthreads=$ythreads --cpuset-cpus="0-5,24,27" servethehome/monero_dwarfpool:xmrig
docker run -itd --privileged -e username=$wally.$(hostname)-2 -e email=$emal -e numthreads=$ythreads --cpuset-cpus="6-11,30,33" servethehome/monero_dwarfpool:xmrig
docker run -itd --privileged -e username=$wally.$(hostname)-3 -e email=$emal -e numthreads=$ythreads --cpuset-cpus="12-17,36,39" servethehome/monero_dwarfpool:xmrig
docker run -itd --privileged -e username=$wally.$(hostname)-4 -e email=$emal -e numthreads=$ythreads --cpuset-cpus="18-23,42,45" servethehome/monero_dwarfpool:xmrig

You want 4x instances, one for each NUMA node. Each instance you want to use L3 cache / 2 so 16MB/2 = 8. You then want to prioritize physical cores.

 

[dexvx]

Just got this setup and running yesterday.

Generic dwarfpool question, am I correct in reading that for Monero the tx transaction fee is 0.008 XMR per payout?

 

[Patrick]

Just got this setup and running yesterday.

Generic dwarfpool question, am I correct in reading that for Monero the tx transaction fee is 0.008 XMR per payout?

It has changed quite a bit over time. May still be that way.

BTW - working on testing different compilers with xmrig for the next docker iteration.