Author

Topic: [OS] nvOC easy-to-use Linux Nvidia Mining - page 415. (Read 418251 times)

legendary
Activity: 1260
Merit: 1009
I'm using a Gigabyte 990FXA-UD3 R5.

To get ethernet working on 16.04 server I just had to enable ioummu in the bios, and just set dhcp in /etc/network/interfaces.  

Have you tried disabling and then enabling the ethernet?

Quote
press f12 to open the guake terminal

type:

ifconfig -a

then press enter

This should list all your network interfaces.  If you are currently using wifi there are probably 3 listed.  Your ethernet, your wifi, and your loopback (lo).

Note the interface name that corresponds to your ethernet: lets call it "ethernet" for the following commands; just swap "ethernet" with this interface name.

type:

sudo ifconfig ethernet down

then enter:

 miner1

when prompted for the password

type:

sudo ifconfig ethernet up

then enter:

 miner1

when prompted for the password

if ethernet is still not working at this point try:

sudo ifup ethernet

then enter:

 miner1

when prompted for the password


I had a few of these boards years ago; although maybe they were an earlier revision.  Used them to mine BTC with 7970s.   Smiley

newbie
Activity: 8
Merit: 0
What motherboard are you using?

I'm using a Gigabyte 990FXA-UD3 R5.

To get ethernet working on 16.04 server I just had to enable ioummu in the bios, and just set dhcp in /etc/network/interfaces.  
legendary
Activity: 1260
Merit: 1009
Newegg currently has the BIOSTAR TB85 motherboard in stock 1 per customer:

https://www.newegg.com/Product/Product.aspx?Item=N82E16813138436

I have rigs using this motherboard; it is fully compatible with nvOC.

Rarely in stock.  Get one while you can.

These are back in stock again.  Smiley
legendary
Activity: 1260
Merit: 1009
I'm running a couple of 1080 ti cards, but need it to work without a monitor

Remote access is via the built in IPMI on the Supermicro motherboard
https://www.servethehome.com/supermicro-ipmiview-review-remote-server-monitoring-management-ipmi-20-kvm-over-ip/

I might do some XMR mining as it's a dual E5 Xeon board Smiley

The same symptoms occur if I take a vanilla ubuntu install and then load the nvidia drivers.
Login prompt takes the password, but then just reloads to the login again

If vanilla ubuntu has the same problem; might be a compatibility problem with the nvidia driver and IPMI.

Maybe the CUDA driver will work with a vanilla ubuntu install?

I am not familiar with IPMI at all; but I found this:  http://openipmi.sourceforge.net/

Also; assuming you already configured oneBash with your info: when you boot the server with nvOC does it start mining?
sr. member
Activity: 349
Merit: 250
Doesn't help Sad

I think the issue is with the fact I'm running headless over iKVM on a SuperMicro X9DR7-LN4F that is using the Matrox GPU built into the IPMI chipset.


That's probably it.

Might work with a monitor direct to the primary GPU.

What cards are you running on this?

What program or application layer protocol are you using to remote access?

Do you mine XMR or other coin with the cpus?

I'm running a couple of 1080 ti cards, but need it to work without a monitor

Remote access is via the built in IPMI on the Supermicro motherboard
https://www.servethehome.com/supermicro-ipmiview-review-remote-server-monitoring-management-ipmi-20-kvm-over-ip/

I might do some XMR mining as it's a dual E5 Xeon board Smiley

The same symptoms occur if I take a vanilla ubuntu install and then load the nvidia drivers.
Login prompt takes the password, but then just reloads to the login again
legendary
Activity: 1260
Merit: 1009
Doesn't help Sad

I think the issue is with the fact I'm running headless over iKVM on a SuperMicro X9DR7-LN4F that is using the Matrox GPU built into the IPMI chipset.


That's probably it.

Might work with a monitor direct to the primary GPU.

What cards are you running on this?

What program or application layer protocol are you using to remote access?

Do you mine XMR or other coin with the cpus?
sr. member
Activity: 349
Merit: 250
Doesn't help Sad

I think the issue is with the fact I'm running headless over iKVM on a SuperMicro X9DR7-LN4F that is using the Matrox GPU built into the IPMI chipset.

legendary
Activity: 1260
Merit: 1009
Pretty sure this is something really simple...but I can't get a login to the desktop Sad
I get a prompt with either m1 or Guest session and the password miner1 doesn't work for me at all.
It seems to take it but just jumps back to the login screen again

Can you log in in console mode?

press:   ctrl + alt + f1

to enter console mode

login is:  m1

password: miner1

If you can login in console mode; type:

reboot

and press enter

the miner should restart and auto login with full X.
sr. member
Activity: 349
Merit: 250
Pretty sure this is something really simple...but I can't get a login to the desktop Sad
I get a prompt with either m1 or Guest session and the password miner1 doesn't work for me at all.
It seems to take it but just jumps back to the login screen again


legendary
Activity: 1260
Merit: 1009
Thank you for this!  I was pulling my hair out trying to OC on 16.04.

As a token of my appreciation, I'll be mining ZEC with the default oneBash until I get a faster usbstick.

Thanks for the hashes.  Grin

Quote
I'm having trouble getting my ethernet to work, but I'm sure I'll figure it out. 16.04 server also didn't connect via ethernet until I tweaked a few things, but unfortunately the same tweaks aren't working for nvOC, and my usbstick is  too painfully slow to want to keep troubleshooting. Wifi works no problem though.

What motherboard are you using?
newbie
Activity: 8
Merit: 0
Thank you for this!  I was pulling my hair out trying to OC on 16.04.

I'm having trouble getting my ethernet to work, but I'm sure I'll figure it out. 16.04 server also didn't connect via ethernet until I tweaked a few things, but unfortunately the same tweaks aren't working for nvOC, and my usbstick is  too painfully slow to want to keep troubleshooting.  Wifi works no problem though.


As a token of my appreciation, I'll be mining ZEC with the default oneBash until I get a faster usbstick.
legendary
Activity: 1260
Merit: 1009
Newegg currently has the BIOSTAR TB85 motherboard in stock 1 per customer:

https://www.newegg.com/Product/Product.aspx?Item=N82E16813138436

I have rigs using this motherboard; it is fully compatible with nvOC.

Rarely in stock.  Get one while you can.
legendary
Activity: 1260
Merit: 1009
released v0014:  OP is updated

v0014 changes:
  NOTE: this version is for USB keys only
  added individual card powerlimits
  added manual fan control
  added Guake terminal (press f12 for a standard terminal)
  fixed ccminer libcudart.so.8.0 error
  removed build error logs causing system error messages
  oneBash changes:
    moved oneBash to the Windows partition (you can now configure oneBash on the computer you image the USB key with)
     If you don't have a text editor with syntax highlighting: open it with wordpad.

I recommend downloading atom if you don't have a text editor with syntax highlighting.

Image, Configure, and Mine.

legendary
Activity: 1260
Merit: 1009
Hello, good OC. But could you integrate fan control in the system.

This can be done with a modification to oneBash.  I didn't include fan speed control because once a card is set to manual fan control; even if the fan speed is not set or set to 0: the fan will not turn on no matter how hot the card gets.  Thus an incomplete configuration could damage cards permanently.  

There are 2 commands involved with fan speed:

the one which toggles between automatic and manual fan modes:

Code:
nvidia-settings -a [gpu:0]/GPUFanControlState=1

and the one that sets the actual fan speed:

Code:
nvidia-settings -a [fan:0]/GPUTargetFanSpeed=75

Note these are per card

You MUST ALWAYS set the fan speed when you set the fan to manual ( I would keep it above 50% if on manual; preferably higher if your ambient temperature is high).

If you set the fan speed too low and damage your cards that is your own fault.  If you are aware of the potential to damage your cards due to incomplete configuration and still want manual control; here is a version of oneBash that has fan control:

(note nvOC has a link to this thread in firefox)

Open oneBash with gedit,
delete all the code in oneBash,
then copy and paste the following code,
modify addresses, ect.. as you normally would when configuring oneBash and remember to save your changes


if [ $POWERLIMIT == "YES" ]
then
sleep 6
#change powerlimit by changing the number after -pl to the desired wattage
expect -c 'spawn sudo nvidia-smi -pl 125
expect "*password*:"
send "miner1\r" '
sleep 6
fi

if [ $INDIVIDUAL_POWERLIMIT == "YES" ]
then
sleep 6
#change GPU 0 powerlimit by changing the number after -pl to the desired wattage
expect -c 'spawn sudo nvidia-smi -i 0 -pl 125
expect "*password*:"
send "miner1\r" '
sleep 6

#change GPU 1 powerlimit by changing the number after -pl to the desired wattage
expect -c 'spawn sudo nvidia-smi -i 1 -pl 125
expect "*password*:"
send "miner1\r" '
sleep 6

#change GPU 2 powerlimit by changing the number after -pl to the desired wattage
expect -c 'spawn sudo nvidia-smi -i 2 -pl 125
expect "*password*:"
send "miner1\r" '
sleep 6

#change GPU 3 powerlimit by changing the number after -pl to the desired wattage
expect -c 'spawn sudo nvidia-smi -i 3 -pl 125
expect "*password*:"
send "miner1\r" '
sleep 6

#change GPU 4 powerlimit by changing the number after -pl to the desired wattage
expect -c 'spawn sudo nvidia-smi -i 4 -pl 125
expect "*password*:"
send "miner1\r" '
sleep 6

#change GPU 5 powerlimit by changing the number after -pl to the desired wattage
expect -c 'spawn sudo nvidia-smi -i 5 -pl 125
expect "*password*:"
send "miner1\r" '
sleep 6
fi

#  settings above
#########################################################################
#########################################################################
#########################################################################
#########################################################################
#  implementation below

NVD=nvidia-settings
COOL="sudo nvidia-xconfig --cool-bits=24"

expect -c "spawn $COOL
expect '*password*:'
send 'miner1\r'"

if [ $INDIVIDUAL_CLOCKS == "NO" ]
then
__CORE_OVERCLOCK_0=$__CORE_OVERCLOCK
MEMORY_OVERCLOCK_0=$MEMORY_OVERCLOCK
__CORE_OVERCLOCK_1=$__CORE_OVERCLOCK
MEMORY_OVERCLOCK_1=$MEMORY_OVERCLOCK
__CORE_OVERCLOCK_2=$__CORE_OVERCLOCK
MEMORY_OVERCLOCK_2=$MEMORY_OVERCLOCK
__CORE_OVERCLOCK_3=$__CORE_OVERCLOCK
MEMORY_OVERCLOCK_3=$MEMORY_OVERCLOCK
__CORE_OVERCLOCK_4=$__CORE_OVERCLOCK
MEMORY_OVERCLOCK_4=$MEMORY_OVERCLOCK
__CORE_OVERCLOCK_5=$__CORE_OVERCLOCK
MEMORY_OVERCLOCK_5=$MEMORY_OVERCLOCK
fi

TI="3"

if [ $___1050_or_1050ti == "YES" ]
then
    TI="2 3"
fi

for j in $TI
  do
    ${NVD} -a [gpu:0]/GPUGraphicsClockOffset[${j}]=${__CORE_OVERCLOCK_0}
    ${NVD} -a [gpu:0]/GPUMemoryTransferRateOffset[${j}]=${MEMORY_OVERCLOCK_0}
    ${NVD} -a [gpu:1]/GPUGraphicsClockOffset[${j}]=${__CORE_OVERCLOCK_1}
    ${NVD} -a [gpu:1]/GPUMemoryTransferRateOffset[${j}]=${MEMORY_OVERCLOCK_1}
    ${NVD} -a [gpu:2]/GPUGraphicsClockOffset[${j}]=${__CORE_OVERCLOCK_2}
    ${NVD} -a [gpu:2]/GPUMemoryTransferRateOffset[${j}]=${MEMORY_OVERCLOCK_2}
    ${NVD} -a [gpu:3]/GPUGraphicsClockOffset[${j}]=${__CORE_OVERCLOCK_3}
    ${NVD} -a [gpu:3]/GPUMemoryTransferRateOffset[${j}]=${MEMORY_OVERCLOCK_3}
    ${NVD} -a [gpu:4]/GPUGraphicsClockOffset[${j}]=${__CORE_OVERCLOCK_4}
    ${NVD} -a [gpu:4]/GPUMemoryTransferRateOffset[${j}]=${MEMORY_OVERCLOCK_4}
    ${NVD} -a [gpu:5]/GPUGraphicsClockOffset[${j}]=${__CORE_OVERCLOCK_5}
    ${NVD} -a [gpu:5]/GPUMemoryTransferRateOffset[${j}]=${MEMORY_OVERCLOCK_5}
done

if [ $MANUAL_FAN == "YES" ]
then
    ${NVD} -a [gpu:0]/GPUFanControlState=1
    ${NVD} -a [fan:0]/GPUTargetFanSpeed=${FAN_SPEED}
    ${NVD} -a [gpu:1]/GPUFanControlState=1
    ${NVD} -a [fan:1]/GPUTargetFanSpeed=${FAN_SPEED}
    ${NVD} -a [gpu:2]/GPUFanControlState=1
    ${NVD} -a [fan:2]/GPUTargetFanSpeed=${FAN_SPEED}
    ${NVD} -a [gpu:3]/GPUFanControlState=1
    ${NVD} -a [fan:3]/GPUTargetFanSpeed=${FAN_SPEED}
    ${NVD} -a [gpu:4]/GPUFanControlState=1
    ${NVD} -a [fan:4]/GPUTargetFanSpeed=${FAN_SPEED}
    ${NVD} -a [gpu:5]/GPUFanControlState=1
    ${NVD} -a [fan:5]/GPUTargetFanSpeed=${FAN_SPEED}
fi

if [ $MANUAL_FAN == "NO" ]
then
    ${NVD} -a [gpu:0]/GPUFanControlState=0
    ${NVD} -a [gpu:1]/GPUFanControlState=0
    ${NVD} -a [gpu:2]/GPUFanControlState=0
    ${NVD} -a [gpu:3]/GPUFanControlState=0
    ${NVD} -a [gpu:4]/GPUFanControlState=0
    ${NVD} -a [gpu:5]/GPUFanControlState=0
fi

if [ $COIN == "ZEC" ]
then
HCD='/home/m1/zec/miner'
ZECADDR="$ZEC_ADDRESS.$ZEC_WORKER"
until $HCD --eexit 3 --fee $EWBF_PERCENT --pec --server $ZEC_POOL --user $ZECADDR --pass z --port $ZEC_PORT --log 2 --logfile ewbf.log;
   do
   echo "FAILURE; reinit in 5" >&2
   sleep 5
done
fi

if [ $COIN == "LBC" ]
then
HCD='/home/m1/SPccminer/ccminer'
ADDR="$LBC_ADDRESS.$LBC_WORKER"

until $HCD -a lbry -o $LBC_POOL -u $ADDR -p x
   do
   echo "FAILURE; reinit in 5" >&2
   sleep 5
done
fi

if [ $COIN == "ETC" ]
then
HCD='/home/m1/9.0/ethdcrminer64'
ETCADDR="$ETC_ADDRESS/$ETC_WORKER"
until $HCD -epool $ETC_POOL -ewal $ETCADDR -epsw x -mode 1
   do
   echo "FAILURE; reinit in 5" >&2
   sleep 5
done
fi

if [ $COIN == "ETH" ]
then
HCD='/home/m1/9.0/ethdcrminer64'
ETHADDR="$ETH_ADDRESS/$ETH_WORKER"
until $HCD -epool $ETH_POOL -ewal $ETHADDR -epsw x -mode 1
   do
   echo "FAILURE; reinit in 5" >&2
   sleep 5
done
fi

if [ $COIN == "DCR" ]
then
HCD='/home/m1/pasc/sgminer'
ADDR="$DCR_ADDRESS.$DCR_WORKER"
until $HCD -k decred -o $DCR_POOL -u $ADDR -p x -p x -I 21 -w 64 -g2
   do
   echo "FAILURE; reinit in 5" >&2
   sleep 5
done
fi

if [ $COIN == "PASC" ]
then
HCD='/home/m1/pasc/sgminer'
ADDR="$PASC_ADDRESS..$PASC_WORKER"

until $HCD -k pascal -o $PASC_POOL -u $ADDR -p x -p x -I 21 -w 64 -g2
   do
   echo "FAILURE; reinit in 5" >&2
   sleep 5
done
fi

if [ $COIN == "DUAL_ETC_DCR" ]
then
HCD='/home/m1/9.0/ethdcrminer64'
ETCADDR="$ETC_ADDRESS/$ETC_WORKER"
DADDR="$DCR_ADDRESS.$DCR_WORKER"
until $HCD -epool $ETC_POOL -ewal $ETCADDR -epsw x -dpool $DCR_POOL -dwal $DADDR -dpsw x
   do
   echo "FAILURE; reinit in 5" >&2
   sleep 5
done
fi

if [ $COIN == "DUAL_ETC_PASC" ]
then
HCD='/home/m1/9.0/ethdcrminer64'
ETCADDR="$ETC_ADDRESS/$ETC_WORKER"
ADDR="$PASC_ADDRESS..$PASC_WORKER"
until $HCD -epool $ETC_POOL -ewal $ETCADDR -epsw x -dpool $PASC_POOL -dwal $ADDR -dpsw x -dcoin pasc
   do
   echo "FAILURE; reinit in 5" >&2
   sleep 5
done
fi

if [ $COIN == "DUAL_ETC_LBC" ]
then
HCD='/home/m1/9.0/ethdcrminer64'
ETCADDR="$ETC_ADDRESS/$ETC_WORKER"
ADDR="$LBC_ADDRESS.$LBC_WORKER"
until $HCD -epool $ETC_POOL -ewal $ETCADDR -epsw x -dpool $LBC_POOL -dwal $ADDR -dpsw x -dcoin lbc
   do
   echo "FAILURE; reinit in 5" >&2
   sleep 5
done
fi

if [ $COIN == "DUAL_ETH_DCR" ]
then
HCD='/home/m1/9.0/ethdcrminer64'
ETHADDR="$ETH_ADDRESS/$ETH_WORKER"
DADDR="$DCR_ADDRESS.$DCR_WORKER"
until $HCD -epool $ETH_POOL -ewal $ETHADDR -epsw x -dpool $DCR_POOL -dwal $DADDR -dpsw x
   do
   echo "FAILURE; reinit in 5" >&2
   sleep 5
done
fi

if [ $COIN == "DUAL_ETH_PASC" ]
then
HCD='/home/m1/9.0/ethdcrminer64'
ETHADDR="$ETH_ADDRESS/$ETH_WORKER"
ADDR="$PASC_ADDRESS..$PASC_WORKER"
until $HCD -epool $ETH_POOL -ewal $ETHADDR -epsw x -dpool $PASC_POOL -dwal $ADDR -dpsw x -dcoin pasc
   do
   echo "FAILURE; reinit in 5" >&2
   sleep 5
done
fi

if [ $COIN == "DUAL_ETH_LBC" ]
then
HCD='/home/m1/9.0/ethdcrminer64'
ETHADDR="$ETH_ADDRESS/$ETH_WORKER"
ADDR="$LBC_ADDRESS.$LBC_WORKER"
until $HCD -epool $ETH_POOL -ewal $ETHADDR -epsw x -dpool $LBC_POOL -dwal $ADDR -dpsw x -dcoin lbc
   do
   echo "FAILURE; reinit in 5" >&2
   sleep 5
done
fi
[/quote]


(I will add these modifications to the next release, but for now)

then for this version you will need to go thru these steps:

open a standard terminal: To open a standard terminal open the mining terminal, select the terminal, move the cursor to the top left, select file, open terminal, select Unnamed

in the standard terminal enter the following; then press enter:

Quote
sudo nvidia-xconfig

enter the password:
Quote
miner1
when prompted

THEN

in the standard terminal enter the following; then press enter:

Quote
sudo nvidia-xconfig --cool-bits=4

enter the password:
Quote
miner1
when prompted

now ensure you have saved oneBash, close the mining process if open and logout (top right select logout, then confirm).

when x changes are done and you see the blue login screen; login


let the standard mining process start: you should see that it implements manual fan control but no OC.


open a standard terminal: To open a standard terminal open the mining terminal, select the terminal, move the cursor to the top left, select file, open terminal, select Unnamed

in the standard terminal enter the following; then press enter:

Quote
sudo nvidia-xconfig

enter the password:
Quote
miner1
when prompted

THEN

in the standard terminal enter the following; then press enter:

Quote
sudo nvidia-xconfig --cool-bits=28

enter the password:
Quote
miner1
when prompted

close the mining process if open and logout (top right select logout, then confirm).

when x changes are done and you see the blue login screen; login


Now you should be able to both OC and manual fan control by changing:

MANUAL_FAN="YES"         # YES NO

FAN_SPEED=85

in oneBash

a Final warning: always use a high enough fan speed if set to manual.


newbie
Activity: 1
Merit: 0
Hello, good OC. But could you integrate fan control in the system. Something like this.
Code:
#!/bin/sh
# cool_gpu       This script will enable or disable fixed gpu fan speed
#
# chkconfig: 345 95 5
# description:  A script hack for GPU fan control on headless GPU nodes
#

# Copyright (c)2011, Axel Kohlmeyer

# locations of all the magic
dir=/opt/set-gpu-fans
smi=/usr/bin/nvidia-smi
set=/usr/bin/nvidia-settings

# if we have a previous GPU logger, terminate it
nvlpid=`pgrep -P 1 nvidia-smi`
if [ "x${nvlpid}" != "x" ]
then
    kill -TERM ${nvlpid}
fi

# determine major driver version
ver=`awk '/NVIDIA/ {print $8}' /proc/driver/nvidia/version | cut -d . -f 1`

# drivers from 285.x.y on allow persistence mode setting
# so we should not need the logger hack anymore
if [ ${ver} -ge 285 ]
then
  ${smi} -pm 1
else
  # initialize GPU logger printing status once per hour
  # to keep an active handle on the GPUs and we don't
  # "lose" the settings applied in the following section.
  nohup ${smi} -d -l -i 900 < /dev/null &> /dev/null &

  # it always takes some time to get all the GPUs initialized,
  # (about 5 seconds) so we give nvidia-smi a little while until
  # we launch any X servers to avoid unwanted race conditions.
  sleep 10
fi

# for multiple tesla devices, only one display is supported.
# thus we need to launch the X server once for each display
# making each of the PCI IDs the primary device in turn.

# command to set fan speed on primary GPU.
nvscmd="${set} -a [gpu:0]/GPUFanControlState=1 -a [fan:0]/GPUCurrentFanSpeed=85"

# go back to automatic, if called with stop argument
if [ "x$1" == "xstop" ]
then
  # with the newer driver, going back to default
  # is simple. we just turn off persistence mode
  # and trigger one reset by showing the GPU status
  if [ ${ver} -ge 285 ]
  then
    ${smi} -pm 0
    ${smi}
    exit
  else
    nvscmd="${set} -a [gpu:0]/GPUFanControlState=0"
  fi
fi

# get PCI bus ids of Nvidia cards and convert from hexadecimal to decimal. watch out for the falling toothpicks.
pciid=`lspci  | sed -n -e '/VGA compatib.*nVidia/s/^\(..\):\(..\).\(.\).*/printf "PCI:%d:%d:%d\\\\\\\\n" 0x\1 0x\2 0x\3;/p'`

for s in `eval ${pciid}`
do \
    cfg=`mktemp /tmp/xorg-XXXXXXXX.conf`
    sed -e s,@GPU_BUS_ID@,${s},    \
        -e s,@SET_GPU_DIR@,${dir}, \
        ${dir}/xorg.conf >> ${cfg}
    xinit ${nvscmd} --  :0 -once -config ${cfg}
    rm -f ${cfg}
done

# no need to keep the logger around
if [ "x$1" == "xstop" ]
then
    nvlpid=`pgrep -P 1 nvidia-smi`
    if [ "x${nvlpid}" != "x" ]
    then
        kill -TERM ${nvlpid}
    fi
fi
Just find another way, but after this overclock doesn't work/
Code:
sudo nvidia-xconfig
sudo nvidia-xconfig --cool-bits=4
Reboot your system and open NVIDIA X Server from Unity dash.
You will find an extra Thermal Settings entry has been added where you can adjust GPU fan speed using the slider.
legendary
Activity: 1260
Merit: 1009
What about a situation like mine though, where I have 2 of one GPU and 2 of another, so I need to be able to specify different OC and powerlimits for the different cards.

v0013 supports individual card OCs; but it doesn't (as is) support individual card powerlimits.  I will add this to the next version.

If you modify oneBash.sh in the following way it will support individual card powerlimits.  You may have to stop and restart the mining process a few times before it initializes as intended; if you have a slow usb this is likely. (success can be verified either with a kill-a-watt or by viewing the power readout from EWBF).

Note: in nvOC v0013 firefox has a bookmark to this page.

replace the following part of oneBash.sh:

Quote
if [ $POWERLIMIT == "YES" ]
then
sleep 6
#change powerlimit by changing the number after -pl to the desired wattage
expect -c 'spawn sudo nvidia-smi -pl 125
expect "*password*:"
send "miner1\r" '
sleep 6
fi

with this:

Quote
if [ $POWERLIMIT == "YES" ]
then
sleep 6
#change powerlimit by changing the number after -pl to the desired wattage
expect -c 'spawn sudo nvidia-smi -i 0 -pl 125
expect "*password*:"
send "miner1\r" '
sleep 6
fi

if [ $POWERLIMIT == "YES" ]
then
sleep 6
#change powerlimit by changing the number after -pl to the desired wattage
expect -c 'spawn sudo nvidia-smi -i 1 -pl 125
expect "*password*:"
send "miner1\r" '
sleep 6
fi

if [ $POWERLIMIT == "YES" ]
then
sleep 6
#change powerlimit by changing the number after -pl to the desired wattage
expect -c 'spawn sudo nvidia-smi -i 2 -pl 125
expect "*password*:"
send "miner1\r" '
sleep 6
fi

if [ $POWERLIMIT == "YES" ]
then
sleep 6
#change powerlimit by changing the number after -pl to the desired wattage
expect -c 'spawn sudo nvidia-smi -i 3 -pl 125
expect "*password*:"
send "miner1\r" '
sleep 6
fi

replacing the number after pl to what wattage is desired for each card (see how cards are enumerated by EWBF to know which setting applies to which card ( the number after -i should match the number EWBF refers to the card as).

This is for 4x cards; as you said that is what you are using.  If you try this let me know if have any problems.



sr. member
Activity: 372
Merit: 250
The road of excess leads to the palace of wisdom
This is all very interesting to me because I currently use Windows 8.1 for my Nvidia Rig but in the past I have mined with my AMD rigs using EthOS and found it much more stable and reliable, and using much lower resources, etc - so a Linux solution for an Nvidia rig sounds great. I was considering trying the PiMP Nvidia version but when I tried that for my AMD rigs it confused the hell out of me.

Anyway, I have one big question because it doesn't seem clear to me from the small amount of info in this thread whether it will suit my rig. Reason being, my rig is currently 2 x  Gigabyte  gtx1070 G1 Gaming cards and 2 x Galax gtx 1060 6GB cards.

It seems from what I read above that when you set cc & mc overclock and powerlimit - this is one setting to apply to all cards on the rig. This is obviously suitable for the typical situation where people commonly use  multiple instances of 1 type of card on a rig, which I understand - people choose a GPU, buy however many of them and populate the rig with them.

What about a situation like mine though, where I have 2 of one GPU and 2 of another, so I need to be able to specify different OC and powerlimits for the different cards.

Currently in Windows 8.1 I just use MSI afterburner and set each card individually, so for example I mine ZEC and the 1070s are powerlimited to around 68%, core OC to +70 and mem OC to +700, while the 1060s are powerlimited to about 75%, core OC to +50 and me OC to +500. Using EWBF this is giving me approximately 1400sol/s using 500W at the wall.

Since the powerlimit in Linux seems to be set in watts rather than percent, one setting in a percentage (70% would probably work) but I would need to be able to set the 1070s to around 125W and the 160s to around 85W. The CC could all conceivably be set to +60 and MC to +500 or +600 but is there a way to set powerlimits individually for each card?
legendary
Activity: 1260
Merit: 1009
Thanks for the update!
Suggestion for future version:
http://glances.readthedocs.io/en/stable/aoa/gpu.html
Currently the best monitoring tool for Linux, IMHO.

Glances looks good; I like the minimalism.  Python as well.  I'll look into it.

Also:

It has come to my attention you may encounter the following error when you select LBC or otherwise use ccminer:

error while loading shared libraries: libcudart.so.8.0:

to resolve this:

Quote
open a standard terminal: To open a standard terminal open the mining terminal, select the terminal, move the cursor to the top left, select file, open terminal, select Unnamed

and enter the following in the new terminal window then press enter:
Quote
sudo ldconfig /usr/local/cuda/lib64

password is:
  miner1

sr. member
Activity: 308
Merit: 250
Version v0013 is up with all member requested changes.  If you only mine ZEC; unless you want to use individual overclocks for each card this update is not necessary.

Change Log:
v0013 (current release)
  dummy plug is no longer required
  added ccminer (both tpruvot and sp-hash)
  added CUDA 8.0
  updated Claymore to 9.0
  installed Ubuntu updates
  oneBash changes:
    moved pool addresses and ports to the top section
    added individual card cc and mc OC
    added 1050 switch (use if you have 1050's in your rig)
    added LBC, DUAL_ETC_PASC  DUAL_ETC_LBC  DUAL_ETH_PASC  DUAL_ETH_LBC

Thanks for the update!
Suggestion for future version:
http://glances.readthedocs.io/en/stable/aoa/gpu.html
Currently the best monitoring tool for Linux, IMHO.
legendary
Activity: 1260
Merit: 1009
Version v0013 is up with all member requested changes.  If you only mine ZEC; unless you want to use individual overclocks for each card this update is not necessary.

Change Log:
v0013 (current release)
  dummy plug is no longer required
  added ccminer (both tpruvot and sp-hash)
  added CUDA 8.0
  updated Claymore to 9.0
  installed Ubuntu updates
  oneBash changes:
    moved pool addresses and ports to the top section
    added individual card cc and mc OC
    added 1050 switch (use if you have 1050's in your rig)
    added LBC, DUAL_ETC_PASC  DUAL_ETC_LBC  DUAL_ETH_PASC  DUAL_ETH_LBC
Jump to: