mph_switch/README.md

nvOC MiningPoolHub Switcher

nvOC makes it easy to squeeze more hashrate out of your GPUs. MiningPoolHub makes it easy to mine altcoins and get paid in Bitcoin. However, nvOC is a customized Ubuntu Linux and the management software MiningPoolHub recommends for profit switching only runs under Windows. Their approach to coin-switching with nVidia GPUs under Linux is to loop through a bunch of different miners, erroring out on all but the currently-profitable coin. I lost nearly four hours this morning when a miner didn't exit properly during a switch.

This set of scripts bridges the gap. With them, you can use the miners included in nvOC (and others that you might add) to mine whatever's most profitable at MiningPoolHub. For each supported algorithm, you can customize power, clockspeed, and fan speed settings as you consider appropriate.

Setup

1. ssh into your mining rig's m1 account.

2. Clone these files into the same directory as nvOC's oneBash script.

3. Disable oneBash by commenting out all lines in ~/2unix. oneBash should be invoked manually if you add/remove GPUs so it can reconfigure /etc/X11/xorg.conf, but it otherwise isn't used. To keep gnome-terminal from sucking down CPU cycles by constantly restarting the 2unix script, you might want to include something like "sleep 86400" at the end.

4. Edit mph_conf.json. You'll set your user and miner names here. You can also customize your overclocking settings for each algorithm here.

5. If your network is IPv6-enabled, note that nvOC (as of v17) has a problem connecting to IPv6 hosts. MiningPoolHub makes its API available over both IPv6 and IPv4, so to force it to use IPv4 in the meantime, edit /etc/hosts as follows:

   104.27.168.52 miningpoolhub.com

6. Set mph_switch.py to launch with a cronjob:

   0,10,20,30,40,50 * * * * (cd /media/m1/1263-A96E && python2.7 mph_switch.py mph_conf.json)

Every 10 minutes, the script will run and determine which miner needs to start up. If it's already running, it will leave it alone. If it needs to switch, it kills one miner and starts the other. If nothing's running (because the rig just rebooted or the miner crashed), it will start the miner.

current-profit is a text file that will be updated on every run of switch.py to show which is currently the most profitable. Profitability is determined by the MiningPoolHub API and is a unitless number; higher is better.

algo-log is a list of coins and algorithms the switcher has selected when run. If you start seeing repeats of the same algo every 10 minutes, that's a sign that something has gone wrong with your GPUs and you might want to consider a reboot.

Miners are started in Screen sessions (http://www.gnu.org/software/screen) that can be viewed over an ssh login session:

screen -dr miner

When the rig switches algorithms, the Screen session will be closed and a new one opened. Repeat the above to monitor it, or you could use something like this on a monitoring workstation:

while true; do screen -dr miner; done

Adding Algorithms

Some algorithms supported by MiningPoolHub can't be mined with the miners included in nvOC. If you add the appropriate software, though, you can mine these as well. Instructions to add these capabilities will follow.

Sia

To add the miner:

sudo apt install golang-go mkdir /home/m1/gocode GOPATH=/home/m1/gocode go get github.com/robvanmieghem/gominer

To enable mining, create a new object named "sia" within the "miners" list in mph_conf.json. Here's what I use:

 "miners": 
 {
   [existing entries deleted for brevity]
    "Sia":
    {
      "bin": "/home/m1/SPccminer/ccminer -a sia -o stratum+tcp://{HOST}:{PORT} -u {NAME}.{MINER} -p x",
      "power_limit": [115,115,95], "gpu_oc": [-100,-100,0], "mem_oc": 0, "fan": [70,75,70]
    }
 }

CryptoNight

To add the miner:

cd ~
git clone https://github.com/KlausT/ccminer-cryptonight KTccminer-cryptonight
cd KTccminer-cryptonight
TAB="$(printf '\t')"
cat <<EOF | patch
--- Makefile.am 2017-07-26 17:32:02.000000000 -0700
+++ Makefile.am 2017-09-21 12:33:54.412247785 -0700
@@ -33,7 +33,7 @@
 ccminer_LDADD${TAB}${TAB}= @LIBCURL@ @JANSSON_LIBS@ @PTHREAD_LIBS@ @WS2_LIBS@ @CUDA_LIBS@ @OPENMP_CFLAGS@ @LIBS@
 ccminer_CPPFLAGS${TAB}= \$(CUDA_CFLAGS) -msse2 @LIBCURL_CPPFLAGS@ @OPENMP_CFLAGS@ \$(PTHREAD_FLAGS) -fno-strict-aliasing \$(JANSSON_INCLUDES) -DSCRYPT_KECCAK512 -DSCRYPT_CHACHA -DSCRYPT_CHOOSE_COMPILETIME

-NVCC_GENCODE${TAB}${TAB}= -D_FORCE_INLINES \$(CUDA_CFLAGS) -gencode=arch=compute_20,code=sm_20 -gencode=arch=compute_20,code=sm_21 -gencode=arch=compute_30,code=sm_30 -gencode=arch=compute_35,code=sm_35 -gencode=arch=compute_37,code=sm_37 -gencode=arch=compute_50,code=sm_50  -gencode=arch=compute_52,code=sm_52 -gencode=arch=compute_60,code=sm_60 -gencode=arch=compute_61,code=sm_61
+NVCC_GENCODE${TAB}${TAB}= -D_FORCE_INLINES \$(CUDA_CFLAGS) -gencode=arch=compute_61,code=sm_61

 # we're now targeting all major compute architectures within one binary.
 .cu.o:
EOF
./autogen.sh
./configure --with-cuda=/usr/local/cuda-8.0
make

To enable mining, create a new object named "cryptonight" within the "miners" list in mph_conf.json. Here's what I use:

 "miners": 
 {
   [existing entries deleted for brevity]
    "Cryptonight":
    {
      "bin": "/home/m1/KTccminer-cryptonight/ccminer -a cryptonight -o stratum+tcp://{HOST}:{PORT} -u {NAME}.{MINER} -p x",
      "power_limit": [115,115,95], "gpu_oc": -200, "mem_oc": 1000, "fan": [70,75,70]
    }
 }