Topic: [OS] nvOC easy-to-use Linux Nvidia Mining - page 394. (Read 418253 times)

Is anyone heaving issues with ZEC miner? I have 4 cards in one of my rigs: 1050ti, 1060 6G and 2x 1070 and it's running at full speed and after few minutes the mining speed drops to like 40% and after 2-3 minutes it's back to full speed and it's doing that over and over again...
I have all cards with around 70% tdp +140 clock and memory at stock.

At first I thought the v0015 image had a problem with this motherboard.  I tested almost every possible configuration of cards; this is what I found.  The nvOC_v0015 image will OC even 8x GPUs on the Asus Z270-P when using 2x m2 adapters.  It will OC 6x GPUs if using 6 GPUs; the only time I had a problem is when I was using only one GPU.  Whenever I tried to use only one GPU; a blank xorg.conf is generated and the normal xorg.conf is moved to a backup file with the current date.

So the takeaway is this:  Use more than one GPU, with one of those GPU's being attached to the primary pcie slot, and v0015 will work as intended and OC all the cards.  If you use only one card; the system will create a blank xorg.conf and the card will not OC.  Note once you have booted the system with only one card you will either need to reimage or perform the following process before the USB will be able to OC multiple cards normally again.

If you want to use only one GPU with nvOC_v0015 you can; for now you will have to do the following before the solo GPU will be properly recognized:

look up the current date then think of it in MMDDYYYY format  that is 2 digit month, 2 digit day, 4 digit year

press f12 to open the guake terminal, if cpuminer is running press ctrl + c to close it or open a new tab and enter:

replace the MMDDYYYY in the next line with the date; for example today would be; 06082017


and enter the password: miner1 when prompted.

logout and login

Your solo GPU should now be properly recognized.  


So when I tested this earlier I entered:


and entered the password: miner1 when prompted.

logout and login, and the GPU is properly recognized.  

I will fix this problem, but as the easy fix is to simply use more than one card or enter this one liner into the guake terminal one time.  This should be ok for the time being.


Edit:  Note if the date you are looking for is the date you first booted with one GPU.  If you are unsure of the date you can do this:

Open the guake terminal:


then:


look for a file that looks like: '/etc/X11/xorg.conf.06082017'  use the date from this file in the command above.

Your right; was late when I came up with this.  I was thinking 1300 G2 with 2x molex cables from the PSU; to be used together to power the 7th  1060.  With this PSU (the 1000GQ); the best solution would be to power 2x 1060 off of one pcie 2x 8pin cable, one at the 1st node and the other with an extender from the 2nd node to the 7th GPU.

$9 pcie extender:

https://www.amazon.com/Cable-Matters-2-Pack-6-Pin-Extension/dp/B01DV1Z4EQ

technically you could have the cards very close together and not use an extender.

Also with 2x 1060s on one cable you don't want the power draw to go above 210 watts total; so you would NEED to ensure that a powerlimit of 105 or lower is set.

The safer solution would be to get a second PSU to power the 7th (and potentially 8th and 9th if you can find 2x m2 adapters) and use a

$22 PSU joiner:

https://www.amazon.com/Add2PSU-Multiple-Power-Supply-Adapter/dp/B009P98Q8U

Glad you figured out some linux. 

I just setup this mobo and booted v0015.  For some reason this mobo uses very abnormal pcie addressing.  This is also causing the xorg.conf to revert to a blank xorg and placing the actual xorg.cong into the backup. 

I will make the v0015 image compatible with this addressing scheme and upload a new image that will support this mobo. Later tonight.

This sounds like a TPM or Secureboot problem ( has to be changed in the bios).  If there is an option for secureboot; disable it, likewise if there is an option to disable TPM do that as well.

Your rig is somehow not assigning a GPU 0.  Are all the pcie slots populated starting closest to the mobo CPU and then moving to the right (open slots if any on the right side)?

If not, powerdown your rig and rearrange the pcie slots and reboot.

Yeah I am only using one, Thanks for the info!

If you are only using 1 dimm it really doesn't matter.  If you are using 2 dimms then you need to read the motherboard manual and see which 2 slots to use. 

When using only one dimm on a z270 chipset; I use the second slot counting from the CPU side.

Matters. You should read more about RAM memory channels.

noob question, but does it matter which slot the ram goes into on the motherboard?

Hello, i'm getting the following error: https://pastebin.com/raw/y71QELyk

Using Z97X MBO and GTX1060 that aren't plugged in yet but getting some system errors about nvidia-smi command doesn't exist and nvidia-ml library doesn't exist. Moved oneBash to desktop and changed 2unix to correct paths. The system is run from SSD. Tried reimaging the disk a couple of times but same errors appear. Currently using onboard graphics for video over VGA and waiting for GPU delivery in the morning.

Reading the complete topic now to see if the error has been resolved somewhere.

Thanks

About to get started with a build. Wish me luck!
https://i.imgur.com/3UBjM7N.jpg

Well, I put 5 1070's (6 wouldn't work?) on an ASRock z97x mobo and cranked up the miner, but I get a grub error at startup:

--> cannot find boot/grub/i386/pc/normal.mod

Seems like its a partition configuration issue, but since that is out of my control, maybe its the motherboard or the bios? Is the z97x a compatible board?

Thanks!

Hello,
I found solution for the overclocking.
My driver version is NVIDIA 381.22

Use these two commands reboot and OC is not a myth anymore

command1:
nvidia-xconfig -s -a --allow-empty-initial-configuration \
--cool-bits=12 --registry-dwords="PerfLevelSrc=0x2222" \
--no-sli --connected-monitor="DFP-0" -o /etc/X11/xorg.conf

command2:
sed -i '/Driver/a Option "Interactive" "False"' /etc/X11/xorg.conf

Why the $6 molex  adapter?

On rig G3930 , but i'm doing some settings of the image on another PC so there i have i3 2120 no luck there also.

I just updated to latest bios on TB-250 and also to the latest nvOC , now my rig won't boot.

Anyone know what causes this error?

I have the mobo in miner mode and set to 3.5gb.

Should have done one thing at a time so I know what caused it.


Edit: I got it to boot by typing :
help
fsck /dev/sda2
 
Then pressing Y for every question.

I got more errors in the miner window once it booted up though so I am now doing a fresh image.

Edit 2 : Re-imaged USB and all good

Reimage didn't have any effect on resolving the problem.  My CPU is a G3930 celeron.

They are both good mobos; the main reason is that I can't find any m2 adapters in stock from any seller in the US.  Without an m2 adapter the ASUS PRIME H270-PLUS can only use 6x GPUs.

I always consider the $/hash of an entire rig.  Using more cards without needing to use a more expensive PSU lowers the $/hash of the rig.

The stock Intel heatsink should be good for the level of CPU mining you would be doing on the side with nvOC (this is nothing like full out CPU mining; and even then: the stock heatsink would probably be ok unless you are using 8+ threads at max intensity)

I have one of these motherboards, but it is boxed.  I will get it out tonight and try to replicate what is happening myself.  In order to replicate the situation as closely as possible: what CPU are each of you using?