Topic: how to build a stable 5GPU mining rig ...? (Read 617 times)

Rig is consuming too much power or bad rise,
One of these.
Easy clock down the core of those gpus to around 1100Mhz.
Also consider bios mod to bring down power. Should see 600Watts max
Get rid of those unpowered risers use powered risers only when multi-gpu.

First off, you need to undervolt and underclock your GPU's, they are drawing too much power to hit 800+ Watts.
I am assuming you have already modded the bios for memory timings since you mentioned the memory type of the cards.
With 5 GPU's you should ideally be under 600W total system load under windows. Use a tool like Watttool to underclock and undervolt the GPU core each time windows boots, and to apply the memory overclocks that your card is stable at.

Second, MSI b150 PC Mate is known to struggle with more than 4 cards, so that is most liely the source of your troubles. Not all motherboards will do more than 4 cards.
The B250 Z270 motherboards are much better in that respect.
Having said that, I would recommend digging into the BIOS and spend some time disabling every Sata port and Every USB port, Audio interface etc that you are not using. ALso change all PCIE lanes to Gen 1.

Third, you could try no tusing the M.2 slot ssd and try with a regular ssd connected to a regular sata port to see if that helps, just in case the m.2 slot is taking up a PCIe lane.

Best of luck.

Thank's hardwarebros, i'll take look at this.

Ok modzer0, thank you very much for your time, i'll try to change something in my config to fix that !

That's probably the max I would want to be running 24/7 on a 1000 watt PSU. It should be ok, but you don't usually want to go much over 80% load continuously. If the PSU has any minor issues at all, it could show up here. It sounds like you're probably ok here, but still could be a possibility.


You probably want to get rid of those SATA power adapters. If you read around on these forums, you'll find issues with them are common. At best they may not be able to deliver enough power, and at worst they can melt or start a fire. I use 4-pin molex directly, but risers with 6-pin power are even better.


Thanks, I read through those and I'm still not convinced that SSH is really the problem in those cases. I strongly feel they all have some other issue with their system, which just happens to get triggered with SSH. When the rig is working solidly, SSH should be no problem at all.


Good luck. Let us know how it goes!

I suggest you follow a tutorial on YouTube step by step, hopefully that will solve your problem.

Hi modzer0, thx for reply !

Yes, i have a power meter and my whole system draws 810 watts.

For the temperatures, all my cards are between 58 - 64C (set fan speed to 80%)

I use 2 risers 16x not powered and 3 risers 1x powered with SATA power adapters. They are very hightly rated too, i don't think that's the problem but why not.

I did not look at the logs but I will do it tonight on my way home from work. I might know more.

Yes i have fex sources (https://forum.z.cash/t/step-by-step-ubuntu-16-04-amd-pro-driver-rock-solid-optiminer-1-4-0/13977 at the end of the post), (http://www.cryptobadger.com/2017/04/build-ethereum-mining-rig-linux/   in the responses of the post).
I know that's weard but using SSH made directly crash Claymore... Don't using keep it works few hours.

Maybe my motherboard is just too weak for 5 GPU...

Again, thank's for reply, i hope i will find something.

I am running a 5 GPU mining system with Ubuntu, although with all different hardware than you. A few questions -

Do you have the rig plugged into a power meter so you can see how much power it's drawing? ie. how close you are coming to the limits of the power supply. Insufficient power is a common cause of strange instabilities. 1000 watts seems like enough, but you don't know for sure unless you can measure. I have a 1000 watt power supply and my whole system draws 760 watts (I have set power limits on the GPUs to keep consumption down)

What are your temperatures? Heat is another common cause of strange instabilities. My rig is in a cool basement, and all my cards run under 60C (I have set fan speeds higher than default to keep temperatures down)

What risers are you using? Low quality or unpowered risers can cause instabilities, through bad power delivery or the PCIe bus. I'm using risers I bought on Amazon that were very highly rated (and not the cheapest ones I could find), they are all powered and I'm not using SATA power adapters.

Are there any errors in syslog/dmesg from the drivers when Claymore crashes? Logs might contain hints as to what is going wrong.

Do you have a source for what you read about SSH not being good for Claymore? I do everything over SSH with my system. I've been a Linux user and sysadmin for over 15 years, and in my experience I can't see any way that SSH would have a negative effect on one specific miner, unless it is somehow a side-effect of other problems.

I'm confident your issues can be resolved... it's just a matter of troubleshooting and finding the cause.

Strange I have the same thing going on.. in a even weirder way.

I have rigs that have been running with 5 gpu for weeks now and suddenly only work with 4. To make it even weirder, a week ago it was my rigs with MSI Armor OC (570;s)that dropped from 5 to 4 GPU's and 2 days ago it happened with my XFX 570 rigs. When I place the gpu's that stopped into a new rig they mine perfectly. I have no clue what is happening. Windows update is off, the hardware updates are off, I ran the pixel patcher, I ran the windows hide updates tool. It's leaving me clueless.

Hello guys !

It's been a few days now that I've been trying to make my RIG 5 GPUs run stably.
First of all I specify my material :

- Carte mère MSI B150 PC MATE
- Alim 1000W EVGA
- 4 Go de RAM
- SSD 60go M2
- Processeur Intel Celeron
- GPU (RX580 4Go Pulse, Hynix et Elpida)

I initially started with a simple Ubuntu Server. With 4 GPU, no problems, but with the 5th, network loss, Claymore instability, etc, etc ...

Then, I followed a tutorial of a person who runs 6GPU with Ubuntu Server (https://blockoperations.com/build-6-gpu-zcash-headless-mining-rig-ubuntu-16-04-using-claymore /). Same problem as at the beginning, i can't even reboot normally, or after 2-3 try ... Brief not stable at all.

Looking for a bit, I found on the motherboard bios the 4G decoding function and changing the PCIe to GEN1 or GEN2. I noticed an improvement but still nothing significant.

Finally I tried this tutorial (http://www.cryptobadger.com/2017/04/build-ethereum-mining-rig-linux/). It worked a bit, about ten minutes, crash of 1 GPU and claymore crash ... Can not restart otherwise than by pressing the power button.

Then I read that the SSH wasn't good for Claymore, so I launched the mining without connecting in SSH, and I saw a clear improvement. My rig turn half an hour before crash, I revived it then it turned 3h. But still a problem of stability since last night, it cut to 1:30 am. This morning the cards turned all the way but no miner in activity.

On Windows 10 do not talk about it, the OS detects 1 times out of 5 the 5 GPUs ...

So I confess I'm lost and I'm starting to lose hope. Maybe you can help me ?

My next tracks are: use an earlier version of Claymore, maybe more stable. Trying with Stock bios on my GPU (although I do not believe it too much). Otherwise I do not see

Please help me !!

thanks in advance

Ps : sorry for my bad english..