I have set up a 9 gpu system with nvOC 16 and here's my experience.
nvOS is stellar. Fullzero has put something really special together here. Hats off to you. I have used windows builds with Claymore as well as other linux os such as PIMP. This is by far the best experience I have has so far.
nvOS is as easy as setting up in windows but with the advantage of linux as well as being able to control everything including over clocking with a bash script file that's as easy as filling in a form.
Currently this 9 gpu rig is hashing ETH and DCR using the suprvova pools. I have found v16 with claymore 9.5 to be far superior to the previous version and i think it will be our goto for larger deployment for the time being.
Right now it hashes at around 183 Mh/s. Here's the build:
1x Motherboard Asus z270 K
http://amzn. to/2tIqN86 x1
1x Celeron CPU
http://amzn. to/2u2GVkk x1
1x Memory x1
http://amzn. to/2rPPjme x1
2x PSU
http://amzn. to/2rPKeKQ x2
1x USB Drive
http://amzn. to/2s9HB5W x1
2x m2 pcie adapter
http://amzn. to/2rPKqty x2
3x pcie riser
http://amzn. to/2tIwPFA x3
9x GPUs
http://amzn. to/2rPX1Ni x9
1x power switch
http://amzn. to/2tIomCo x1
1x Wooden frame
http://amzn. to/2s4ZsQn x1
around $3300 with tax
Here is a pic.
Here are some nvOC wallpapers for your rigs.
https://drive.google.com/drive/folders/0B0lx4O21EyI3eEFHcG8tWEhIU0k?usp=sharingDuring setup i encountered numerous errors, all of which were remedied. I went with 2 psus, not sure if its better in teh long run but splitting 1100 watts over two 850s seem to be decent for efficiency. I have built 7 of these now and a few with the z270A version using 8 GPUS. Both are good. I have created this system as the model for our small mining farm in California, so i will continue to feedback on these builds as I go.
Bios settings were very important. As was bios version.
Damaged risers and a bum gpu were also headaches during set up so be on the lookout for that. Once you have the bio settings right and the hardware actually works it really is as simple as flashing the drive, plugging it in and turning it on. If anyone wants the bios settings PM me but its here in the thread and is mostly about setting all pcie settings to gen2, using bios 0801 and not flashing the newer 0906. Which was the opposite to PIMP which did not seem to work for me on the older bios.
One last thing, with these gpus I set the OC to
__CORE_OVERCLOCK=150
MEMORY_OVERCLOCK=825
and it seems to be running stable with no ETH rejections and about 1 in 200 DCR rejections. temps at 80 fans at 65% -dcri 50 and pulling about 1050 watts.
I built two of these for a friend and two for my sister.
Both are pointing at nice hash. On is here
https://new.nicehash.com/miner/1GCSdasYQ3wWp8zAHHvHyao83aoyA6AFVM - currently its waiting on pcie m2 adapters to get all 9 cards going. They will recoup their investment in 3-4 months and their minds have been opened to the new world of block chains, contracts and decentralized crypto currency.
Thanks fullzero this is a brilliant project.
I'm glad you like nvOC
I like your frame:
the labeling + super cost effective: all while maintaining functionality
I have made similar frames in the past. The best part about this type of frame is the materials should be locally available.
I am not a fan of shorted weblinks: other than when using twitter: I see no reason to use them. Please use normal links in the future.
I will add a link to your wallpapers to the OP.
Hi guys, I have a 7x card 1070 rig which has been rock solid for 99% of 2 weeks.
On two occasions it seems to have lost internet connection ( it is hard wired to the router / not wifi )
Killing the session and restarting it does not solve the issue however turning off the rig and resetting does fix the issue.
Has anyone else experienced a similar issue or seen this with certain settings?
When the console says that internet connection has been lost it seems to be correct as opening a fresh web page says the same thing.
Just weird that resetting the rig through power off/on fixes the issue, cheers!!
Ubuntu ubdates may be causing this: you can disable them by:
clicking the top left ubuntu button:
typing: u
then selecting:
Software and Updates
clicking the Updates tab
then unselecting everything but security updates (these are always small; and a good idea to keep on)
So I had the rig working with one MSI gtx 1080 connected to the x16 slot, when I connect a riser to the x16 slots (any of them) the card fans are working but the card won't be recognised by the operating system, replaced riser and still, doesn't work. The card is working of I put it in the x16 slot.
My mobo is an Asrock H81 pro btc r2.0 PSU is a Corsair Rx 750, in using a powered v3 riser, both molex on mobo are connected to same PSU as the riser and GPUs , on bios I have the pcie card as default, tried gen1 and gen2 everytime clearing CMOS after any plug, no way it is recognised by the pc. Anyone had same issue?
I think your risers might be the problem; I highly recommend using the newest PCI 6 pin powered risers (as they are almost perfect, almost all of them work without issue).
I understand it is very hard to find these at the moment; I'm sure production has already ramped up and there will be plenty again soon. I expect GPUs will soon be the hardest part to source; thanks to the z270 chipset working well for mining.
Hey,
Just wondering how many CPU threads you guys are using to mine monero without causing impact to GPU mining operations? Currently have mine set at 2 but was thinking it can probably do 3 using G4560.
Thanks
It depends on what Client you are using to GPU mine. EWBF is very light on the CPU so you could probably do 3 threads without noticing a difference in the GPU hashrate.
If you are using Claymore to mine a Ethash COIN; I wouldn't use more than 2: if dual mining with Claymore I wouldn't use more than 1.
With a weaker CPU I wouldn't cpu mine at all while dual mining with Claymore.
OK thanks for the advice, i'll apply as necessary.
One more thing, is it just me or do the 1070s refuse to draw more than 150W power even when the power limit is increased? Obviously it works upping the powerlimit in windows but it doesnt seem to work for me in linux, only on the reduction.
Cheers
You cannot exceed the top limit for each card. If you do nothing will be applied.
to find each cards limit you can enter:
in the guake terminal
no card should allow a 400 watt powerlimit; so smi should output the bottom and top limits for each card.
Pick the top limit and run
sudo nvidia-smi -pl toplimit
to set the cards to max
I would use caution when using max powerdraw; generally slightly above tdp power will make a difference; but max will make the card unstable and could potentially damage it if your ambient is high, your cables are loose, or your PSU is not high end.
Do any of you have idea why 6x Asus 1060 3GB make around 65MH/s on ETH? (65.529 MH/s, 25/0/0 (0.00%): 9.642+9.760+12.370+13.154+8.689+11.911 MH/s)
Gigabyte Z270 Gaming K3, 8GB, Pentium G4400
The same configuration on SMOS makes around 19MH per card.
Thanks
I don't know what settings your are using; but there is something wrong based on your hashrates.
I would try Tasbi's settings:
https://bitcointalk.org/index.php?topic=1854250.msg19703709#msg19703709
In version15 I added --log to the startup of EWBF and it would output the logfile. I tried it in version16 but cant seem to find the logs. Even a ps -ef | miner shows the following output snippet :
--pass d=256 --log --port 4048
Where can I find the log output of EWBF miner now? I tail the file in my ssh session to see of my rig is running healthy. Now I dont have means to see the status of my rig.
I removed the logging from the default oneBash as it was very slightly decreasing stability; you can add it by appending:
to the end of your workername
for example with the default workername I would change it from
to:
to add logging.
Let me know if you think it is slowing down the ZEC hashrate at all; in my tests it appeared to have no impact on simultaneous Equihash mining: but more data makes us better informed.
I now have 2 going.
a i5 6400t on a biostar z170 2 threads runs fine. better cpu so makes sense
a i3 6100t on a gigabyte b250m 2 threads crash one thread worked
Good to know. The t series have very low powerdraw overall.
now running 3 rigs
2x 1080ti gigabyte b250m mobo
2x 1080ti biostar z170 mobo
3x 1080ti biostar z170 mobo
I will now try a 4 card 1070 biostar z170 mobo
Let me know how it goes
Hello miners...
First things first, I would like to thank 'fullzero' (infact hero
) for this fantastic job, can't thank him enough.
I've built my first RIG very recently, I would like to find the best OC settings and pool settings to get the best out of it, with the help of miners expertise.
I was getting really good speed on windows but for some reason it doesn't allow me run more than 3 GPU's (it was a trial version), but it was doing 20 MH per card from stock with claymore dual mining.
I had to switch back to Ubuntu but the trouble is I'm only getting 18 MH on an average per card, I'm using claymore dual (ETC + PASC) with CUDA 8.x nvidia driver - 375.66, pulling 1044 watts from wall.
Temperatures : 65 - 73 C
Fans on Auto : 39 - 43
My config -
ASUS DUAL GTX 1060 - 8 GPU's
ASUS Z270P
8 GB RAM
2 x EVGA G2 750 Watts PSU units
8 x V6 Risers
120 GB SSD
Fellow miners using same cards, please share your Core, memory OC + fan + power settings. I would also like to know how to add multiple pools to the nvOC oneBash file.
Would also welcome suggestions about best combination cryptos for better ROI with my equipment.
Thanks in advance.
Regards,
damNmad
Glad you like nvOC
Are these 3gb or 6gb 1060s?
If they are 3gb I would try these settings:
https://bitcointalk.org/index.php?topic=1854250.msg19703709#msg19703709
I just got around to test v16.
Same error messages like below.
Getting infinite loop.
LCD connected to mobo graphics (like always in v15 and BIOS set to on board graphics not PCI)
Will test on PCI graphics later thi PM - hopefully no more infinite loops - will post results later.
Much better boot-up speed, this time can use USB3 port fully compatible with build?
I just added 5 more 20A circuits in my warehouse - fun times begins!
I can't seem to get v16 to work on tb-250.
It boots up then terminal pops up and says:
dos2unix: converting file /media/m1/1263-A96E/onebash to unix format
xorg problem detected
restarting xorg
rebooting in 5
Have tried different USB's and also booting with only one card but still have the same problem.
Anyone know what the problem might be?
Have you waited for the second boot?
If you are using one GPU it is expected that oneBash will need to rebuild the xorg file and reboot before OC will work as intended.
It should work without rebooting on the second boot; unless you have connected the monitor directly to the motherboard (always connect a monitor to the primary GPU).
I have had one member report that their rig is caught in an infinite loop of reboots; so I would like to know if this is happening to you as well.
Full Zero was right. If you change your primary display to PCI and plug your monitor into your primary card it will boot.
The other great thing is my clocks are working with v16...my 1070's have gone from 390MH to 430 MH
Thanks full zero!!
Thought I would give an update. After adjusting the power settings and doing some specific troubleshooting I have gotten the rig stable for 24 hours.
I took fullzero and Maxximus007 suggestions on the power settings and give it a bit more with more modest OC. It seems to have resolved the issue, as well I have found that one of the cards was not liking bigger OC's while the other 5 worked fine.
This was the outcome.
ETH - Total Speed: 180.068 Mh/s, Total Shares: 2000, Rejected: 0, Time: 15:48
ETH: GPU0 30.198 Mh/s, GPU1 30.240 Mh/s, GPU2 30.186 Mh/s, GPU3 29.500 Mh/s, GPU4 30.086 Mh/s, GPU5 29.857 Mh/s
SC - Total Speed: 1200.451 Mh/s, Total Shares: 429, Rejected: 0
SC: GPU0 201.322 Mh/s, GPU1 201.599 Mh/s, GPU2 201.240 Mh/s, GPU3 196.667 Mh/s, GPU4 200.575 Mh/s, GPU5 199.048 Mh/s
GPU0 t=62C fan=75%, GPU1 t=48C fan=75%, GPU2 t=64C fan=75%, GPU3 t=61C fan=75%, GPU4 t=60C fan=75%, GPU5 t=49C fan=75%
As for the settings, I have the following settings:
5 cards:
Overclock
cc 100
mc 1275
Powerlimit: 125 watts
1 card:
cc -100
mc 1225
Powerlimit: 120 watts
(this card does not like OC had to tweak it some for it to not cause instability within the rig)
Total watts from the wall: 866/867
I am still taking fullzero advice and adding 4 GB more of RAM it will be here in the next few days.
Maxximus007 or fullzero if you have a BTC,ETH address I would like to send you a small thank you for the help on troubleshooting.
These results are very good: Your getting typical ETH (single mining hashrates) while dualmining.
Let me know if you think the additional ram makes a difference.