Monero Mining Performance

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

Patrick

Administrator
Staff member
Dec 21, 2010
12,511
5,792
113
@MiniKnight they are moving slightly as I get better numbers/ re-test some of the nodes with L3 cache MB / 2.
 

Patrick

Administrator
Staff member
Dec 21, 2010
12,511
5,792
113
@gigatexal all I saw was user-submitted benchmarks. Those are good but you have very little idea what is going on for each one. I wanted to have the best dataset with a consistent testing methodology.
 
  • Like
Reactions: gigatexal

Patrick

Administrator
Staff member
Dec 21, 2010
12,511
5,792
113
Something like that is the goal. This thread is largely my notes while building that piece. Well, and folks like @fractal @cafcwest and @Kal G telling me what I am doing wrong!

I would run an ultra low fee STH pool but I think our combined mining power would be around 30KH/s so too low to effectively mine in a small pool.
 

MiniKnight

Well-Known Member
Mar 30, 2012
3,072
973
113
NYC
I feel like I'm rolling in cash now. $12.20! I've almost got 2.0 XMR

Now I've just gotta find out how to buy Sam Adams with it.
 

cafcwest

Member
Feb 15, 2013
136
14
18
Richmond, VA
Just a quick update:

Still doing the Minergate thing, as I have not really had the time to devote to doing anything else, nor have I made a decision or taken any steps on putting together some dedicated miners from used enterprise servers (i.e. - getting "serious" about mining)

I had re purposed my old lab C6100 machines for this - they had been sitting idle for the last year or more. I have (7) C6100 nodes with dual E5620's (8th node is running a few VMs for a pet project). With 8 physical/16 logical cores, I found the sweet spot for Minergate was utilizing 12 of those cores - that has me up to a little over 200H/s per node.

In addition, I threw Minergate on my home utility/gaming desktop. It is a Dell OptiPlex desktop with an i7-6700 and an old GTX 750ti card. Running both the CPU and GPU, the processor is spitting out 160H/s, with the GPU doing another 120H/s.

All total, spitting out 1.7-1.8 kH/s with minimal effort. Only paying for the extra electricity at home - my "co-lo" is free. So at current pricing, I'm making something like $80/month. I am going to do some more research regarding the market predictions for Monero and decide whether to get "serious" about this. If this is a fad - cool, I'll make some beer money.
 

Patrick

Administrator
Staff member
Dec 21, 2010
12,511
5,792
113
@cafcwest great results! so you are using 12 of 16 threads per node? That is a great result. Which version are you using?

Also, how close is your KH/s in the console to your estimated earnings? I think I have aggregate 23KH/s but am seeing returns of more like 17KH/s. I am not seeing lost shares in the logs/ console. The moneropool.com UI seems to be 23KH/s as well. Returns are low enough I was thinking of just setting up a pool myself to see if that would help.
 

Patrick

Administrator
Staff member
Dec 21, 2010
12,511
5,792
113
@jim any tips on getting that setup on Ubuntu? I will try to do it in Docker this weekend and see how it performs on larger machines.
 

jim

New Member
Jul 22, 2016
7
5
3
32
@jim any tips on getting that setup on Ubuntu? I will try to do it in Docker this weekend and see how it performs on larger machines.
I'm running it on gentoo (with NUMA enabled), but should work similarly for ubuntu....

It does need some initial setup.... Here are the steps I did following the author's README.
First of all I changed some configs as follows:
Code:
diff --git a/CMakeLists.txt b/CMakeLists.txt
index b377c22..464cb63 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -19,7 +19,7 @@ if("${CMAKE_BUILD_TYPE}" STREQUAL "")
     set(CMAKE_BUILD_TYPE RELEASE)
 endif()

-set(CMAKE_C_FLAGS "-DNDEBUG -march=westmere -O3 -m64 -s")
+set(CMAKE_C_FLAGS "-DNDEBUG -march=native -mtune=native -Ofast -funroll-loops -m64 -s")
 set(CMAKE_CXX_FLAGS "${CMAKE_C_FLAGS} -std=c++11")

 set(CMAKE_EXE_LINKER_FLAGS_RELSEASE "")
diff --git a/config.txt b/config.txt
index 23f5dc3..e29c37d 100644
--- a/config.txt
+++ b/config.txt
@@ -3,7 +3,7 @@
  * here is the size of your L3 cache divided by 2. Intel mid-to-high end desktop processors have 2MB of L3
  * cache per physical core. Low end cpus can have 1.5 or 1 MB while Xeons can have 2, 2.5 or 3MB per core.
  */
-"cpu_thread_num" : 2,
+"cpu_thread_num" : 20,

 /*
  * Thread configuration for each thread. Make sure it matches the number above.
@@ -22,8 +22,26 @@
  *
  */
 "cpu_threads_conf" : [
-       { "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 0 },
-       { "low_power_mode" : false, "no_prefetch" : false, "affine_to_cpu" : 1 },
+       { "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 0 },
+       { "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 1 },
+       { "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 2 },
+       { "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 3 },
+       { "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 4 },
+       { "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 5 },
+       { "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 6 },
+       { "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 7 },
+       { "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 8 },
+       { "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 9 },
+       { "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 10 },
+       { "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 11 },
+       { "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 12 },
+       { "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 13 },
+       { "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 14 },
+       { "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 15 },
+       { "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 16 },
+       { "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 17 },
+       { "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 24 },
+       { "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 25 },
 ],
After cloning the git repo and apply the above changes,
Code:
sudo sysctl -w vm.nr_hugepages=128
This can be made persistent into /etc/sysctl.conf or /etc/sysctl.d/XXX.conf
Then edit /etc/security/limits.conf and add the following lines
Code:
YourUsername soft memlock 262144
YourUsername hard memlock 262144
Then install libmicrohttpd (or libmicrohttpd-dev), compile and run
Code:
cmake .
make
while true; do ./bin/xmr-stak-cpu; done;
libmicrohttpd is not required if you use an earlier version.
 

gigatexal

I'm here to learn
Nov 25, 2012
2,913
607
113
Portland, Oregon
alexandarnarayan.com
Found this interesting about this miner:

Default dev donation

By default the miner will donate 1% of the hashpower (1 minute in 100 minutes) to my pool. If you want to change that, edit donate-level.h before you build the binaries.

If you want to donate directly to support further development, here is my wallet

4581HhZkQHgZrZjKeCfCJxZff9E3xCgHGF25zABZz7oR71TnbbgiS7sK9jveE6Dx6uMs2LwszDuvQJgRZQotdpHt1fTdDhk
Also it's based on C++11 so that is probably one reason it's fast.
 

cafcwest

Member
Feb 15, 2013
136
14
18
Richmond, VA
@cafcwest great results! so you are using 12 of 16 threads per node? That is a great result. Which version are you using?

Also, how close is your KH/s in the console to your estimated earnings? I think I have aggregate 23KH/s but am seeing returns of more like 17KH/s. I am not seeing lost shares in the logs/ console. The moneropool.com UI seems to be 23KH/s as well. Returns are low enough I was thinking of just setting up a pool myself to see if that would help.
Yes, 12 of 16 threads. Anything more or less drops below 200H/s.

The console widely varies from refresh to refresh. But based on the 'off the hip' averages of each of the devices, I came up with my 1800H/s number. I took a 24 hour period and reversed the number from the difference in total mined from the dashboard and it came out pretty close. As an FYI, my bad shares comes out to about 0.007% of total shares.

If you are willing to put in the work to set up a pool, I will certainly onboard. Might need a high-level tutorial on getting one of these other applications working, as I'm mostly a MS button clicker type.
 

BigDaddy

Member
Aug 8, 2016
38
20
8
I am interested in the potential pool as well. My dual E5-2686 v3 running 2x 16 thread Mint VMs gets me ~ 1.2kh/s w/ minergate. Would love to have a docker image and STH pool. I am guessing that if I set it up better I should be able to get around atleast 1.5 to 1.6.
 

Bilbo

New Member
Feb 19, 2016
19
3
3
37
I have been demo-ing a couple of HP DL380s with dual E5-2665s. I am running 2x virutallized Ubuntu VMs with the Monero GUI beta 1 per host. I have 16 threads to work with and gave each VM 7 to mine with. Each VM seems to get about 325 h/s. If I just configure one VM with all 16 threads and let it mine with 14 threads... I only get about 575 h/s for some reason. After a week of mining I have received no monero.

I have also played with minergate on my home PC and found that compared to the monero CLI there seems to be a small performance hit. It also pounds the CPU a lot harder when given the same amount of threads to work with as the CLI. However, based on my lack of results in the lab with the DL380s, I am starting to think a pool is really the way to go.

I have 16x C6220s all with dual E5-2665s. If STH wanted to start a pool, I would be a lot less hesitant about joining a pool and would be happy to commit 50% of my resources (should be about 4Kh/s).
 
  • Like
Reactions: gigatexal

Patrick

Administrator
Staff member
Dec 21, 2010
12,511
5,792
113
Will see if I can get a pool up and running with enough of the optimizations to make it worthwhile.

After trying minergate my general idea is that it seems to have a lower H/s on the machine I am using, but given that, it seems to be accurate in payouts.
 

Marsh

Moderator
May 12, 2013
2,642
1,496
113
Patrick, please get a pool going.
Here is my experience of Minergate (only 14 hours )

Hardware: MSI X99 , 2 x 8gb ram , Intel M2 SSD, no case or case fan.
Using Windows 8.1 as OS and Windows version of Minergate software.
E5-2650L v3 production CPU , L3 is 30MB , mining software thread count is set to 14
322 H/S ( low )

14hours running time, used 1.2KW @0.18KW/h ( tier1 , lowest cost tier ) = $0.216
Generated 0.02 XMR = $0.24
I think I made 3 cents in 14 hours. Is the math right?
 

Patrick

Administrator
Staff member
Dec 21, 2010
12,511
5,792
113
@Marsh it is on my list to do over the next two weeks of travel. It is going to be "chunkier" XMR earnings since it will be a smaller pool.

I am also doing things like I will have my FS thread items payable in XMR.
 
  • Like
Reactions: gigatexal

zxt

New Member
Jan 22, 2017
1
0
1
41
Hello, I'm wondering how you can use Xeon Phi to mining Monero. Can you give some detail?