Topic: Building Cheap Miners : My "Secret" - page 2. (Read 60219 times)

Hello Again - How many threads and cores are you using to get 6866.8 H/s?

I now have Dell R815 Servers. Is it possible to overclock them?

Also I am trying to replace some 6176 CPUs with 6376 CPUs.   I naively thought they would work as they are the same socket. They do not.  I am using DL385 from HP.   Do you have some ideas on what I need to do to make the CPU work please? When I boot the machine I just get a red flashing light at the front.  No boot sequence. When I reinstall the 6176 everything works fine.

Dell R815 Ubuntu 16.04 Desktop Installation Steps:

Go here to get Ubuntu 16.04 LTS
http://releases.ubuntu.com/16.04

Download the 64-bit PC (AMD64) desktop image
http://releases.ubuntu.com/16.04/ubuntu-16.04.6-desktop-amd64.iso

After downloading has completed: install it onto a USB stick so that you can boot from it

I use the free Rufus program: https://rufus.ie
https://github.com/pbatard/rufus/releases/download/v3.8/rufus-3.8.exe

Now put the Ubuntu USB stick into the USB port on the server front.

Now power up the server and when prompted press F11 to boot from the USB stick that has Ubuntu 16.04 LTS on it.

In the F11 boot menu select "Hard Drive" and then select the Front USB drive.

--------------------------------

You should now see Ubuntu booting.

When it has completed booting click on "Install Ubuntu"

On the next screen check mark the "download updates while installing Ubuntu" line.

On the "Installation Type" screen select "Something Else"

Select/Highlight "/dev/sdb" and click "New Partition Table"

Now Select "Free Space" and click the "+": Enter 975MB and "Swap Area"
Now again Select remaining "Free Space" and click the "+": "EXT4 Journaling File System" and "Mount Point" is "/" then continue

You will see that two partitions will be created: One is the Swap the Other is "EXT4" where everything else is installed to.
If you see more partitions go back and correct it. If you don't then sometime in the future you will run out of disk space.

Now continue with the installation.

Select your correct Time Zone and Keyboard Layout.

On the setup screen I do the following:

I leave "Your Name" blank

"Computer Name": R815-1
For each additional R815 I configure I change this to R815-2, R815-3, etc

Username: miner

I use logon name of: miner
If you use a different name then other commands will need "miner" changed to whatever you name it to.

Password: xxxx

Password is whatever you want and can remember: I use "xxxx" for simplicity.

I let the system autologon when booting

When Installation completes: reboot
Be sure to remove the installation Ubuntu USB stick from the server front USB port.

Wait for the initial boot of Ubuntu from the internal USB drive to complete.

Now do all Updates.

Either wait for the Update Icon to appear on the left side of the screen (in about 5 minutes) or force updates with the following commands:

T to open a terminal window then do these commands:

Code:

   sudo apt update
    sudo apt upgrade

When Updates have completed then do these command form a terminal window

Code:

   sudo apt install shellinabox
    sudo apt install mc

Shell in a box: https://www.tecmint.com/shell-in-a-box-a-web-based-ssh-terminal-to-access-remote-linux-servers
Usage: https://blog.homeatcloud.com/manuals/the-shellinabox-tool

Midnight Commander (MC): https://www.linode.com/docs/tools-reference/tools/how-to-install-midnight-commander
How to use MC: http://linuxcommand.org/lc3_adv_mc.php

Now Get the Server IP address with
    

Code:

   hostname -I

Now do the following from a from a remote browser (FireFox, Chrome. Internet Explorer) on your local network https://:4200

This bring up a Shellinabox session.
Using Shellinabox simplifies setup as you can copy and paste from this guide using a right mouse click in the Shellinabox browser window.

-----------

Install R384 Nvidia Driver - see if it already installed

Code:

   nvidia-smi

Remove older Nvidia driver

If your graphic is supported, you can go ahead and remove all previously installed Nvidia drivers on your system. Enter the following command

Code:

   sudo apt purge nvidia*

Add the graphics drivers PPA

Let us go ahead and add the graphics-driver PPA -

Code:

   sudo add-apt-repository ppa:graphics-drivers

And update

Code:

   sudo apt update

Install (and activate) the latest Nvidia graphics drivers.
Enter the following command to install the version of Nvidia graphics supported by your graphics card

// Check here for Linux Nvidia Drivers

https://launchpad.net/~graphics-drivers/+archive/ubuntu/ppa


// Check launchpadstatus here

https://twitter.com/launchpadstatus?lang=en


// Quadro 600

Code:

   sudo apt install nvidia-384

Reboot your computer for the new driver to kick-in.

Code:

   sudo reboot

You can check your installation status with the following command

Code:

   lsmod | grep nvidia

If there is no output, then your installation has probably failed.
It is also possible that the driver is not available in your system's driver database.
You can run the following command to check if your system is running on the open source driver nouveau. If the output is negative (blank) for nouveau, then all is well with your installation.

Code:

   lsmod | grep nouveau

Prevent automatic updates that might break the drivers by blocking minor version updates. Enter the following command

Code:

   sudo apt-mark hold nvidia-384

    sudo usermod -a -G video miner

----

Disable graphical interface startup. I like to see the boot messages rather than a screen with a solid color and no indication it is doing anything for a long time.

I use Midnight Commander (mc) when I say edit. Use sudo mc if editing system files


// (Ubuntu 16.04) disable the boot splash


    edit /etc/default/grub


remove 'quiet' and 'splash' from GRUB_CMDLINE_LINUX_DEFAULT line and add nomodeset

so, assuming you have no other boot parameters, it would look like

    GRUB_CMDLINE_LINUX_DEFAULT="nomodeset"

save that, then run

Code:

   sudo update-grub

// Reboot System

Code:

   sudo reboot

This section is for Dell R815 Hardware configuration and BIOS Settings:

I install an inexpensive ($10) Quadro 600 graphics card in each server because the built-in Server GPU has a limit of 1024x768 resolution which is painful to my eyes. Some of the instructions that follows are necessary for using the Quadro 600. If you use another GPU then you will need to change those steps to reflect the GPU you choose.

Initial Dell R815 Server Hardware steps:

Open Server top and install a 16GB USB stick inside at the front left side where there is an internal USB port.
Install Quadro 600 graphics card.

I remove any RAID card and SAS cables as they are not needed and burn unnecessary power while mining.
Also remove any other PCIe cards as they are also unnecessary.

Initial setup for the Dell R815 Server:

Hook up display to the Embedded Video Controller port as the Quadro 600 is not set to display the BIOS screen until a setting in the BIOS is changed.

Boot server and press F2 to enter BIOS when the screen stating "Press F2 to Enter BIOS" is seen.
You should see that line change to "Entering BIOS" when F2 has been pressed.

Note the BIOS version. All my Dell R815 Servers are on Version 3.2.2
If you have a newer version that should also be fine.

Bios Version 3.2.2 is located here:
https://www.dell.com/support/home/us/en/04/drivers/driversdetails?driverid=jf8yh

BIOS settings should be:

--------------------------------------

Under Memory Settings Screen

(On each reboot Memory is tested) When you are satisified that memory has been tested enough then change to Disabled


--------------------------------------

Under Processor Settings Screen


--------------------------------------

Under Integrated Devices Screen


Note on Reboot the the Embedded Video Controller is disabled so switch the monitor cable to the Quadro 600 (or other GPU) you installed.

--------------------------------------

Under Power Management Screen


--------------------------------------

Now Press to exit the System Setup program
and select Save Changes and Exit


Don't forget that on Reboot the the Embedded Video Controller is disabled so switch the monitor cable to the Quadro 600 (or other GPU) you installed.


On each reboot Memory is tested. When you are satisified that memory has been tested enough then change:
System Memory Testing: Disabled

--------------------------------------

I use Ubuntu 16.04 Desktop which is FREE.

I will follow up this post with detailed instructions for Ubuntu 16.04 Desktop installation.

On the Dell R815's I put in a 16GB USB stick installed inside at the front left side where there is an internal USB port.
On the HP DL580's I use a cheap 60GB 2.5" SSD installed inside.

I also install an inexpensive ($10) Quadro 600 graphics card in each server because the built-in Server GPU has a limit of 1024x768 resolution which is painful to my eyes.

See here for cheap reliable and fast 16GB USB sticks:
https://bitcointalksearch.org/topic/m.51847827

Thanks for the update. Lets hope RandomX is as profitable as expected.

I have two HP DL580 G7 Quad E7-8837 and have 5 R815 coming on Friday.

What I did not expect is that Windows 10 has a limit of 2 CPU!  So do CPU go unused.

How do you get around this problem? Which OS are you using?

Obviously Windows Server is expensive.

Thanks for your support.

17 days until Monero forks to RandomX.

I am pumped.

I am currently in the process of wiring another 20 AMP 240 VAC circuit to attach my third 16 AMP PDU.
This is needed for my third set of four Dell R815 Servers.

My garage mining setup will contain twelve Dell R815 Servers with a mixture of quad 6348 & 6378 Opterons.
Also a single SuperMicro Server with dual 6376 Opterons overclocked to 3 GHz.

This is what I have measured for four Dell R815 Servers on one PDU.
Two with quad 6348 and two with quad 6378 Opterons.

On the current CN/R algorithm:

6866.8 H/s
2880 watts (12 Amps * 240 VAC) avg of 720 watts per server measured at the panel.
9.2c per KWH

This setup would be mining for a loss of $4.18 a day which is why these have been powered off and will stay off until the November 30th fork.

https://www.cryptocompare.com/mining/calculator/xmr?HashingPower=6866.8&HashingUnit=H%2Fs&PowerConsumption=2880&CostPerkWh=0.092&MiningPoolFee=1.0

The same same four servers produce 43890 H/s for RandomX.

I will post the RandomX numbers, Hash Rate, Power, and hopefully Profit for this setup after the fork.

Good luck everyone.

FYI: I am using XMRig for my RandomX miner. You can set it to 1% donation or 0% if you compile from the source

https://github.com/xmrig/xmrig/releases/tag/v5.0.0

XMRig 5.0.0, stable release, unified 3 in 1 miner, CPU+OpenCL+CUDA
https://www.reddit.com/r/MoneroMining/comments/dvv637/xmrig_500_stable_release_unified_3_in_1_miner

I see two issues:


The issue is that the reported and usable L3 Cache is 48.0 MB whereas it should be 64.0 MB.

Each AMD Opteron 6378 has 16MB (2x 8MB) of L3 Cache so four of them has 64MB of L3 Cache:
http://www.cpu-world.com/CPUs/Bulldozer/AMD-Opteron%206378%20-%20OS6378WKTGGHK.html

Look for an option in the BIOS (probably in the CPU/Processor section) that says something like "HT Assist" or "Probe Filter Lookup". If that is enabled then it steals 2MB L3 Cache from each NUMA node or 4MB L3 from each 6378 Opteron which reduces the L3 Cache per 6378 Opteron from 16MB to 12MB and 12MB time four is the 48MB you are seeing.

So find that "HT Assist"/"Probe Filter Lookup" option, disable it, reboot and retest.

If you don't see that setting make sure you are on the latest BIOS version for your motherboard.


-------------------------------------------------------------------------------------------------------------------




By selecting the coin to be monero you are effectively mining the current algorithm CryptonightR and not RandomX.
You need to reset coin back to null in the config.json file to test RandomX.

Hi,

Hopefully I'm not doing something I'm not supposed to by posting my questions here.

Information on this post has been really useful, thank you.
I've been GPU mining for a number of years, but this is my first foray into server / CPU mining.  I have a Supermicro H8QGi-F motherboard with 4 x Opteron 6378's.  Have 8Gig memory per Opteron. Running xmrig 4.5.0-beta (prebuilt, I did not compile from source) on Ubuntu 18.04.3 LTS.

I have edited my config.json file as per MinerRus, but I must be doing something fundamentally wrong, as my Hash Rate reported by xmrig is a fraction of what it should be.  I'm starting xmrig by using ./xmrig with no command line arguments.  Am I supposed to be using
start /node 0 xmrig.exe --threads=11
start /node 1 xmrig.exe --threads=11
start /node 2 xmrig.exe --threads=11
start /node 3 xmrig.exe --threads=11
or similar to initiate xmrig?

See below my output from xmrig.

Any help / insights would be much appreciated.


/xmrig-4.5.0-beta$ ./xmrig
 * ABOUT        XMRig/4.5.0-beta gcc/5.4.0
 * LIBS         libuv/1.31.0 OpenSSL/1.1.1c hwloc/2.0.4
 * CPU          AMD Opteron(tm) Processor 6378 (4) x64 AES
                L2:64.0 MB L3:48.0 MB 64C/64T NUMA:8
 * DONATE       1%
 * ASSEMBLY     auto:bulldozer
 * POOL #1      rx.minexmr.com:4444 coin monero
 * COMMANDS     hashrate, pause, resume
 * OPENCL       disabled
 * CUDA         disabled
[2019-11-09 00:39:38.859]  net  use pool rx.minexmr.com:4444  138.201.21.239
[2019-11-09 00:39:38.859]  net  new job from rx.minexmr.com:4444 diff 50000 algo rx/0 height 1338190
[2019-11-09 00:39:38.860]  rx   init datasets algo rx/0 (64 threads) seed a6d2136a753660af...
[2019-11-09 00:39:39.673]  rx   #0 allocated 2080 MB huge pages 100% (812 ms)
[2019-11-09 00:39:39.673]  rx   #1 allocated 2080 MB huge pages   0% (813 ms)
[2019-11-09 00:39:39.673]  rx   #2 allocated 2080 MB huge pages   0% (813 ms)
[2019-11-09 00:39:39.674]  rx   #3 allocated 2080 MB huge pages   0% (0 ms)
[2019-11-09 00:39:39.674]  rx   #5 allocated 2080 MB huge pages   0% (1 ms)
[2019-11-09 00:39:39.674]  rx   #6 allocated 2080 MB huge pages   0% (0 ms)
[2019-11-09 00:39:39.674]  rx   #4 allocated 2080 MB huge pages   0% (1 ms)
[2019-11-09 00:39:39.674]  rx   #7 allocated 2080 MB huge pages   0% (0 ms)
[2019-11-09 00:39:39.820]  rx   #0 allocated  256 MB huge pages 100% +JIT (146 ms)
[2019-11-09 00:39:39.820]  rx   -- allocated 16896 MB huge pages  14% 1168/8448 (960 ms)
[2019-11-09 00:39:42.787]  rx   #0 dataset ready (2967 ms)
[2019-11-09 00:39:45.416]  rx   #4 dataset ready (2629 ms)
[2019-11-09 00:39:45.423]  rx   #6 dataset ready (2635 ms)
[2019-11-09 00:39:45.467]  rx   #7 dataset ready (2679 ms)
[2019-11-09 00:39:45.467]  rx   #2 dataset ready (2680 ms)
[2019-11-09 00:39:45.568]  rx   #3 dataset ready (2780 ms)
[2019-11-09 00:39:45.568]  rx   #1 dataset ready (2781 ms)
[2019-11-09 00:39:45.572]  rx   #5 dataset ready (2785 ms)
[2019-11-09 00:39:45.572]  cpu  use profile  rx  (56 threads) scratchpad 2048 KB
[2019-11-09 00:39:45.780]  cpu  READY threads 56/56 (56) huge pages 100% 56/56 memory 114688 KB (208 ms)
[2019-11-09 00:39:58.509]  net  new job from rx.minexmr.com:4444 diff 50000 algo rx/0 height 1338191
[2019-11-09 00:40:45.677] speed 10s/60s/15m 872.9 43.2 n/a H/s max 882.6 H/s
[2019-11-09 00:41:45.765] speed 10s/60s/15m 873.3 883.4 n/a H/s max 882.6 H/s
[2019-11-09 00:42:45.861] speed 10s/60s/15m 874.2 883.3 n/a H/s max 882.6 H/s
[2019-11-09 00:43:45.953] speed 10s/60s/15m 874.6 883.7 n/a H/s max 882.6 H/s
[2019-11-09 00:44:46.031] speed 10s/60s/15m 869.8 882.2 n/a H/s max 882.6 H/s
[2019-11-09 00:44:58.533]  net  new job from rx.minexmr.com:4444 diff 25000 algo rx/0 height 1338191
[2019-11-09 00:45:09.023]  cpu  accepted (1/0) diff 25000 (351 ms)

RandomX has specific limits in that it needs 256KB L2 cache and 2MB L3 cache for each mining thread.

http://www.cpu-world.com/CPUs/Xeon/Intel-Xeon%20E5-2620.html

Each Xeon E5-2620 processor has: 6 real cores and 6 HT cores and can run twelve threads.

Each Xeon E5-2620 has 6 x 256 KB 8-way set associative L2 Cache and 15 MB 20-way set associative shared L3 cache

The 6 x 256 KB 8-way set associative L2 cache limits you to mining on the 6 real cores and not the HT cores.

The screenshot you posted mining RandomX shows that all six cores of both Xeon E5-2620 processors are used for mining (12 real cores) but the 12 HT cores are idle. If you actually have no other tasks running that will show 50% being utilized and that is correct as 12/24 equals 50%. Trying to use more cores for mining will actually decrease your hash rate.

https://drive.google.com/file/d/1M3u24W8GCo8yLrGZvFvFjmcr3T1SKBCP/view

The only way to improvement performance is to replace the E5-2620 Xeon's with something that has more real cores or is faster.

I have tested dual Sandy Bridge E5-2640 and E5-2670 processors in a Dell T5600 Workstation and these are the results:

Dual E5-2640, 12 threads, 4554 H/s, 310 watts: 14.7 H/s/W

Dual E5-2670, 16 threads, 6406 H/s, 357 watts: 17.9 H/s/W

Dual E5-2690, 16 threads, 7033 H/s, Power currently not measured, this also is a Dell T7600 Workstation

I recently picked up a pair of E5-2640 Xeon's for only $27.06 on eBay.

Also in the recent past I have picked up:

3x E5-2643 Xeon's for $31.31 or $20.88 for a pair
10x E5-2670 Xeon's for $387 or $77.40 for a pair
10x E5-2670 Xeon's for $511 or $102.20 for a pair

HELLO - Thanks for your feedback and concern about not being able to run other applications due to CPU UTILIZATION.

Actually I have the opposite problem.   CPUs are running at 58% and so I have plenty of spare capacity.  That figure includes running Google Remote Desktop without a problem.

So the CPU is under utilised. How can this be fixed please?

Task Manager show 58% CPU utilization: see  https://drive.google.com/file/d/13gGY6jBrF9qpCdjsxpRhlQPoyhpDsqS7/view?usp=sharing

Half the threads are unused. See https://drive.google.com/file/d/1d41Y5YaIscVGkg-kXfpSCYvZvuv_VAEu/view?usp=sharing


If I change the number of threads with '  -t  ' to say 13, 17, 24 or even 11 performance goes down and power consumption and CPU usage go up. There are 12 cores and 24 threads.

If I do not use a --cpu_priority with a value of 4 or greater the machine mines slower.

Is it possible to get more from these CPUs?

Thanks in advance.

Unless you have compiled XMRig from the source:


the "--donate-level 0" will default to Donate Level 5%. The lowest you can go is 1% so that would be:


Also I do not set


If the system is idle then all cpu cores that you have configured for mining will be going full out even without setting cpu-priority. If you do something like web browsing then that will steal some CPU cycles from mining. I worry if you set cpu-priority to 5 (highest) you may not be able to run other applications because they won't have enough cpu cycles to start or if they do start they may be sluggish.


If they are smart they will.

As I stated earlier I am at a loss in why anyone is currently mining LOKI because the numbers I plug in even with free electricity you will be losing money mining it (because of the pool fee).

Thanks. Did you see anything incorrect with the command lines I used for the Dual Xeons (see above)?

Also you do you think the Loki Hashrate is latent XMR hashrate? Ready to move across to XMR?

It looks like I was wrong as the only difference between Loki (RandomXL) and XMR (RandomX) is the RandomX program size (LOKI is larger).


https://www.reddit.com/r/LokiProject/comments/ceas1a/randomxl_mining

THANKS BUT my test server (Dell R610 with dual E5-2620) produced almost identical results for Loki and XMR.

LOKI 3600 H/S see https://drive.google.com/file/d/1XM3l6UzTeQmfUlCQ9DJXco6y4hN1oJUt/view?usp=sharing

XMR 3700 H/S see https://drive.google.com/file/d/1M3u24W8GCo8yLrGZvFvFjmcr3T1SKBCP/view?usp=sharing

So are you sure that Loki and XMR have different hashrate results?

Perhaps I am doing something wrong.  Here are the command lines used:

LOKI: xmrig --url pool.hashvault.pro:80 --user L7f9SMpq53wNERBxCngFyqMoJvkMax8xSitiJY335ra12ucbiEFzgqKfHJg1Q9y58RGMi3TyGvPC94e 6w3zYKK9i9PHmDDr --pass x --keepalive --donate-level 0 --tls  --cpu-priority 5

XMR: xmrig -o randomx-benchmark.xmrig.com:7777 -u 47wcnDjCDdjATivqH9GjC92jH9Vng7LCBMMxFmTV1Ybf5227MXhyD2gXynLUa9zrh5aPMAnu5npeQ2t Ly8Z4pH7461vk6uo  --cpu-priority 5 --algo = rx/0

Can you see an error?

Thanks

THANKS. Do you think perhaps the 52 MH/S of Loki hashrate is latent XMR hashrate? i.e. machines that are commissioned and ready to move to XMR? The benefit of doing this to know the machines work and to make tweaks and test airflow etc

LOKI is a very very very minor coin vs Monero (XMR).

LOKI price $0.315
Monero price: $63.555

https://minerstat.com/miner/xmrig

RandomXL produces hash rates double what RandomX will produce so 52M H/s of RandomXL is equivalent to 26M H/s of RandomX.

I for one do not believe you can extrapolate what you see mining LOKI to Monero (RandomX).

I am also at a loss in why anyone is currently mining LOKI because the numbers I plug in even with free electricity you will be losing money mining it (because of the pool fee).

Even if you had electricity at 1 cent per KWh you will lose 5 cents a day with an EPYC 7742 using 225 watts and producing 77464 H/s of RandomXL.

RANDOMX CPU BENCHMARKS
https://randomx.monerobenchmarks.info/index.php

Loki mining calculator
https://minerstat.com/coin/LOKI

Great chart showing all the coins that XMRig CPU miner can mine
https://minerstat.com/miner/xmrig

As for your question: Do you think you will make any money mining XMR RandomX with servers?

Yes I do for Monero (XMR) but if I am wrong I will be able to resell the servers I have bought for a profit so my risk is minimum.
 

No command line options or changes for the Dell R815 Server with the quad 6348 Opteron's.

As I stated on the Dell R815 Server with the quad 6378 Opteron's it is best to disable one core per NUMA node to gain an additional 1% in hash rate while very slightly lowering power consumption and processor heat. This is done by first running XMRig to that it fills in the CPU core options in the config.json file. Then exiting XMRig and now editing the RX section to remove one core from each of the eight NUMA nodes.

I reformat the very long RX CPU section [0, 1, 2, ... 61, 62, 63] to this:

[0, 1, 2, 3, 4, 5, 6, 7,
 8, 9, 10, 11, 12, 13, 14, 15,

...

48, 49, 50, 51, 52, 53, 54, 55,
56, 57, 58, 59, 60, 61, 62, 63]

that way I can easily see each NUMA node. I then delete the last entry on each line. Be sure to leave the trailing comma in the first seven lines and on the last line remove the ", 63" but leave the "]" bracket.

Now when you run XMRig it will use 56 cores when mining.

THANKS. Did you use anything else on the command line/config file or CPU set up? Or is it optimal to let XMRIG figure it out?

Thanks for the information. Do you think you will make any money mining XMR RandomX with servers? May be you would if you had very cheap power. RandomX is currently used on LOKI, a tiny coin with a daily exchange volume of $10k.   Loki's network Hashrate is 52 MHS which is equivalent to 30000 Xeon E5-2620 CPUs or 4700 of the new Ryzen 3900X CPU.     So that is a huge hashrate to get just $10k out per day.

64GB of PC3-10600R memory (1333 MHz), 8x 8GB sticks

Windows 10 Pro