At a complete loss...

Joined
Aug 30, 2017
Messages
6
I just completed an 8 GPU mining rig. I used Asus Rog Strix GTX 1080 Ti OC, Asrock z270 killer mobo, 8 GB GSkill Ram and 2 platinum rated 1200 watt psus. I'm running windows 10 but nothing I do has been able to get the rig running stable, even at stock settings. I've changed risers, reloaded my os, re-downloaded the drivers, I even have my virtual memory set to 64 GB. I'm just so exhausted because I have no idea what's wrong or what else to try.
 
Please elaborate on what it is doing versus what it should be doing.
 
The longest uptime I've had is about 10 minutes and then I get a cascade of gpu failures. There is no similarity on which one fails first, it watchdogs and restarts but at some point I get a hard freeze and have to hold the power button to restart the computer.
 
Win10, right? Anything in the event viewer?
I only have enough knowledge to make me dangerous so I'm not sure what you mean by event viewer but windows gives me no alerts whatsoever. The only information I get is gpu failure with an unspecified launch failure.
 
rip out all the gpus and reinstall just one. make sure its running correctly in just a basic config. then start adding things back until the problem reoccurs. go from there.
 
It appears like the m.2 ports may be the problem. Is there another option to get 8 GPUs connected without using m.2?
 
Ooookay? So replace the adapter.

My response on the risers was to your question on how to do this without M.2.
 
connecting GPUs to m.2 sockets is a reallly bad idea. if you want to run more than 6 gpus you really need to get a board made for that biostar, asrock and asus all make boards that can support more than 6 gpus for mining.

https://pcmrneeds.help/img/brutusfriedgpu.jpg

that was the result when someone tried to do that with an rx480 elsewhere for example... they don't have the same ability to supply power as normal PCI-E slots.
 
Last edited:
Did you ever get this figured out OP? Just curious as I am assembling a 6gpu rig as I type this
 
I don't know about bad ideal. I have a board running 9 cards and 2 of the cards are use powered m.2 adapters. I have another board running 12 cards.

71hGWR992%2BL._SL1500_.jpg


connecting GPUs to m.2 sockets is a reallly bad idea. if you want to run more than 6 gpus you really need to get a board made for that biostar, asrock and asus all make boards that can support more than 6 gpus for mining.

https://pcmrneeds.help/img/brutusfriedgpu.jpg

that was the result when someone tried to do that with an rx480 elsewhere for example... they don't have the same ability to supply power as normal PCI-E slots.
 
Look at the gpu failures as they cascade. The first gpu is typically the one that caused the failure.

Set your TDP to %75 and stock clocks and see what happens. Underclock the one that fails first
 
Those 3-in-1 risers are very dangerous. I hope you have home Insurance :)
 
On my 2 rigs I have use them for the last 2 months around the clock and no problems. But I also have talk to other miners that use them. They say to use just 2 of the 3 1x slots.


Those 3-in-1 risers are very dangerous. I hope you have home Insurance :)
 
anyone got a link to the 1 to 3 pcie 1x adapter? (amazon preferred)

Might want to give one of them a spin in a motherboard of mine that has 2 PCIe connectors (each with a GPU in it) but the power supply is capable of handing another 2 GPU's easy. Instead of buying a new motherboard/cpu combo, just expand one of the lanes to accept some more GPUs.
 
Back
Top