My rig freezes when mining eth

[Zucca]

I have a mining rig with:

• MSI Z270 Gaming Plus
• EVGA Supernova G2 1300W
• 6 Rx 570 8G
• 4Gb RAM

With 4gpu it works completely stable, after adding the fifth one, no matter which one, it freezes after 2/3 min.
At first it threw Video TDR blue screen failures before restarting, then started to just freeze. A couple of times, my MB bios CMO got cleared.
I have above 4G enabled, pcie latency 96, gen 2/3.
Latest Adrenaline drivers, and gpu stock BIOS (never oc)
I am using Claymore v15.0 with the suggested env vars set
Windows 10 PRO updated version

Ive tried:

• Replacing PCIE risers
• Switching pcie gen from 3 to 2
• Adding 8gb more ram
• Removing straps from Claymores configuration
• Switching from Claymore to Phoenix miner
• Upgrading Adrenaline driver to latest beta version
• Upgrading Windows virtual memory from 16Gb to 32Gb

Any ideas?

[waywoos]

virtual memory?
clock to high?
power supply is maxed?
you are only mining with a single 4 gig ram id would increase the virtual memory in windows

[Zucca]

virtual memory?

32000 mb

clock to high?

Gpus untouched, mining at stock

power supply is maxed?

Don't know what that means

you are only mining with a single 4 gig ram id would increase the virtual memory in windows

Done it and tried more ram

[P00P135]

Add way more virtual memory.  What do you need drive space for anyway.

[Zucca]

Add way more virtual memory.  What do you need drive space for anyway.

More than 32Gb??? How much? 64Gb? I don't know if there is space in disk disk for that

[WhyMe]

You need at minimum 6*8GB of virtual memory ...
And mining @stock is just a joke, stock voltage is just burning your rig ...

[philipma1957]

i would suggest you get a 15-20 dollar kwatt meter and see what you are pulling at the meter.

If you are over 1100 watts and have been using that psu for a long time at say 900 watts the psu may not be up to the task. so pull a gpu to see how it runs.

[Zucca]

You need at minimum 6*8GB of virtual memory ...

Thanks that was the issue. I remember using 16gb for my old 6x 4gb gpu rig so I though 32gb would work

And mining @stock is just a joke, stock voltage is just burning your rig ...

I tried MSI afterburner, it just simply does not work, I will try claymore auto oc or MOD the bios

i would suggest you get a 15-20 dollar kwatt meter and see what you are pulling at the meter.

If you are over 1100 watts and have been using that psu for a long time at say 900 watts the psu may not be up to the task. so pull a gpu to see how it runs.

I have a Kill-a-watt, 910w total with 6 gpu on

[Zucca]

Okay, it started to fail again, with 50Gb virtual memory, it started to throw Video TDR failures.
I also changed the PSU for another similar one with no luck.
It was an improvement however, I can mine with 5 gpu for some hours without getting my rig freeze/restarted
With 6 GPU it instantly gets frozen.

I tried to modify straps with Claymore automatic feature, however it throws Video TDR failure when starting.
I will MOD bios to change straps and reduce voltage with AMD drivers, it my final guess.

[WhyMe]

No, first thing is reducing voltage. No need to edit bios for that.

[Zucca]

No, first thing is reducing voltage. No need to edit bios for that.

Too late
I flashed 1 gpu using ubermix 3.1 straps (it is a samsung memory)
After that MB won't boot with HDMI connected to it, however it is still being recongnized in windows and capable to mine.
I flashed back its original BIOS, however the display was not fixed.

It is still on warranty so I will return the card.

EDIT:
Apparently the backed up ROM extracted from gpu-z of that card was invalid, and the strap modification I did it on that ROM. Good thing is, I did a backup of all ROM so I could reestablish it and its working

[deadsix]

No, first thing is reducing voltage. No need to edit bios for that.

Too late
I flashed 1 gpu using ubermix 3.1 straps (it is a samsung memory)
After that MB won't boot with HDMI connected to it, however it is still being recongnized in windows and capable to mine.
I flashed back its original BIOS, however the display was not fixed.

It is still on warranty so I will return the card.

EDIT:
Apparently the backed up ROM extracted from gpu-z of that card was invalid, and the strap modification I did it on that ROM. Good thing is, I did a backup of all ROM so I could reestablish it and its working

As a rule of thumb never use GPUZ to save roms. Use the bios flashing utility itself.

[john1010]

That is also my problem right now, one of my rig has 12 rx 570 4gb and I am currently using a mining expert motherboard, I powerup my rig with 1000w corsair and 850w fsp,  7gpu for my corsair 1000w and 5 for the psf 850w, I noticed that in the first run of my mining apps, the hashrate was good but after a few minutes, my hash power decrease and some of my gpu turn to zero hashrate, and today everytime i run the miner it was frozen. I did almost of all of the troubleshooting method, changing the riser, the usb cable, etc. Now, my question is, is my gpu enough to power my rig? I saw also in the miner that my rig only consuming 1098w.

[Zucca]

No, first thing is reducing voltage. No need to edit bios for that.

Thanks man!, I undervolted the gpus core form 1150mV to 950mV, now it is running stable at 700w according to my kill-a-watt

As a rule of thumb never use GPUZ to save roms. Use the bios flashing utility itself.

Good to know, thanks.

[philipma1957]

No, first thing is reducing voltage. No need to edit bios for that.

Thanks man!, I undervolted the gpus core form 1150mV to 950mV, now it is running stable at 700w according to my kill-a-watt

As a rule of thumb never use GPUZ to save roms. Use the bios flashing utility itself.

Good to know, thanks.

And psu can now handle the load.

as a rule of thumb 60% on an atx works well 24/7/365. and 1300 x .60 = 680 watts.

[philipma1957]

That is also my problem right now, one of my rig has 12 rx 570 4gb and I am currently using a mining expert motherboard, I powerup my rig with 1000w corsair and 850w fsp,  7gpu for my corsair 1000w and 5 for the psf 850w, I noticed that in the first run of my mining apps, the hashrate was good but after a few minutes, my hash power decrease and some of my gpu turn to zero hashrate, and today everytime i run the miner it was frozen. I did almost of all of the troubleshooting method, changing the riser, the usb cable, etc. Now, my question is, is my gpu enough to power my rig? I saw also in the miner that my rig only consuming 1098w.

no try

five on the 1000 watt

four on the 850 watt

see if they run overnight.

a lot of atx psu age. a year or two into operation they cant do it.

if 5 and 4 work over night

try 6 and 5

remember to downvolt.

[adaseb]

The only thing I can think of is that your motherboard is pretty much sharing some PCIe lanes and why you are getting this issue when you add the 5th GPU. I had some motherboard like this in the past, and wasted so much time trying to figure out what was wrong but basically according to the owners manual there were some shared lanes on the PCIe slots.

Basically you can have 2 PCIe slots but you might only be able to use 1 at a time even if you are using risers and running it at 1x. Go to your motherboards manufacteure website and download the owners manual and see if there are any shared lanes.

[john1010]

That is also my problem right now, one of my rig has 12 rx 570 4gb and I am currently using a mining expert motherboard, I powerup my rig with 1000w corsair and 850w fsp,  7gpu for my corsair 1000w and 5 for the psf 850w, I noticed that in the first run of my mining apps, the hashrate was good but after a few minutes, my hash power decrease and some of my gpu turn to zero hashrate, and today everytime i run the miner it was frozen. I did almost of all of the troubleshooting method, changing the riser, the usb cable, etc. Now, my question is, is my gpu enough to power my rig? I saw also in the miner that my rig only consuming 1098w.

no try

five on the 1000 watt

four on the 850 watt

see if they run overnight.

a lot of atx psu age. a year or two into operation they cant do it.

if 5 and 4 work over night

try 6 and 5

remember to downvolt.

My rig brings me a headache, sometimes the zeroed hash gpu will randomly rotated, after I've change the riser, usb cable and run the miner it will took around 3-4 hours then it will give the zeroed hashrate again in one of my gpu, then my rig started to freeze.

[Zucca]

And psu can now handle the load.

as a rule of thumb 60% on an atx works well 24/7/365. and 1300 x .60 = 680 watts.

Just realized that 1300 x.6 is 780 watts

Anyway, I kept changing the voltage down to 875mV for both core and memory and it is still stable.
How much is it safe to downvolt?, I don't dare to put any lower value.

[adaseb]

And psu can now handle the load.

as a rule of thumb 60% on an atx works well 24/7/365. and 1300 x .60 = 680 watts.

Just realized that 1300 x.6 is 780 watts

Anyway, I kept changing the voltage down to 875mV for both core and memory and it is still stable.
How much is it safe to downvolt?, I don't dare to put any lower value.

I don't think its your PSU. If you want to eliminate your PSU as the culprit just run the miner with reduced intensity, basically run it at 50% and see if it runs then. It will use less power and you can see if its your issue or not.

Try what I suggested and get your motherboard manual and find out about the available lanes on your PCIe slots.