Topic: Rig freezes with 5 or 6 GPU attached (Read 7264 times)

Bumping! Same issue I am facing now with RX570 cards

Patch installed works with 6x rx290x.

Sorry for bump in after long time of inactivity of this thread ...
But I almost have the same problem
I setup a new rig and have problem mining with 6 cards, all 6 cards installed and detected by system
When I enable the 6th one after 10-15 minutes of mining cpu usage goes high and reach 90%. no program with high cpu in task manager but "system"
Tried with MSI afterburner running to speed up fans, and without it 

All 6 GPU are detected and working fine, tried to run with 5 and cycle the 6th one to check if problem is from risers or power cables ( Disabled pcie1 , run with 5, disabled pcie2 run with 5 .... )
All ok with 5 GPU no matter which one, as soon as I enable 6th the cpu goes high.
system is :
Windows 10, 1703 , all drivers updated to latest with "Driver Booster"
Mobo: MSI Z170A XPower Gaming Titanium Edition
PSU: Thermaltake Toughpower XT Gold 1275W
GPU: 6x 1070 Asus Dual 8OG
Ram: ADATA XPG Z1 DDR4 8GB 2800MHz CL17 Dual Channel
CPU: Pentium G4400 3.3GHz LGA 1151 Skylake CPU
HDD: SAMSUNG 750 EVO 120GB SATAIII Solid State Drive


PEG0, PEG1, PEG2 gen 1 (tried with auto too)
4G enabled
IGD disabled (tried with enable too)


Can any one help me please.

2 risers max per line, its better.  Might be your PSU, but try lowering your core clocks, upping your voltage on your gpu's.  Sometimes the primary gpu needs more voltage, and if it crashes, it will crash your rig.  It happens.  Set pci gen to 1 or 2 in your motherboard bios....hmmm other than that you should be good.

Please let me know whether that fixes the issue. I made the same mistake a while back 

Now that I think about it, I might have had 2 attached to a cable, but added the 6th onto the line to make it 3. I didn't think about that. This could be an issue as well.

Thank you.

Windows 10

I'm going to first think it could be drivers, then I'll go down the PSU rabbit hole.

Thank you for your input

The thought has crossed my mind that I may not have enough RAM for 5 or 6 cards. However with this rig I have had 5 cards before and it was running perfectly fine. I had to take a GPU off because it killed itself and for about 3 or 4 months, it was hashing with 4 cards.  I add the 5th, I see some issues occurring. I add the 6th, and the GPU drivers begin to crash. 

My PSU should be ok. Haven't had any issues with it yet.

Your third point is not a bad theory. I can believe that.

I'll keep this in mind.

How many risers are you powering with each molex/sata PSU lead?  Don't power more than two.

your missing out on the most important clue and usually the fault.

Are you running on windows 7 or windows 10? If its 7 then its the capability that you need to ammend the windows register and other hacks before it will be able to read more than 5 cards. If its windows 10 then likely its a driver or Power Supply issue.

JMHO

i have noticed a similar issue with multiple of the anniversary boards. i build ETH and ZEC rigs for a friend (270x's) and with multiple rigs we have noticed an issue with adding card 6.

What ive decided so far:
1)could be not enough ram to run each card, i read somewhere that ZEC/ETH mining requires 1-2GB of ram per card (no idea if true)
2) could be your PSU cant handle all cards properly
3)My personal opinion, the anniversary board was selling far cheaper then the 'ProBTC'... I personally think there is something with the board model itself, as each of the rigs i have this issue with ... do it in the same slot (slot 3... the first after the 16x)

No custom roms, however I did have the rig undervolted -84 for ETH and -96 for XMR.  With 4 GPUs and those underclock settings the rig was fine. Once the 5th GPU came in, that's when the problems began. 

The rig is about 5 months old with 4 RX 480's I bought a 5th GPU less than a week ago and the 6th this past weekend. The rig did run 24/7 while it had 4 cards.

The PSU is an evga 1600w gold

I never had any issues until I added the 5th GPU and it was intermittently freezing. Like, it would run for a few hours then freeze. I didn't test what would happen with 6 because I reinstalled windows and all the other stuff.

Do you use custom rom for those 480?
any undervolting?

How old is it? what brand is it? Have you been running it 24/7 under full load?

not all 1600w psu are created equal.

I have a 1600w psu
Default clock settings

Maybe the dust or riser.  I'll double check.

Could be a poor power supply or a faulty riser or simply dust causing weak connectivity to a card/cards.

Or too high overclock.

Hello everyone, I was hoping someone might have some insight for me about why my rig may be freezing.

I have
H97 anniversary mobo.
4gig RAM.
Cheapo CPU .
6 MSI RX 480 Gaming X 8GB - Stock settings and -86/-96 undervolt for ETH and XMR respectively when the rig was stable.
No custom GPU BIOS.
Windows 10.
Virtual Memory set to 16GB min -> 32GB max.

The rig will freeze if I have 5 or 6 GPUs attached. The driver will constant fail, freeze, and sometimes reset the pc. The rigs will freeze mining Monero and Ethereum.

I have done a full system reset. Installed only the necessary drivers plus GPU drivers, set the virtual memory to 16gb min and 32gb max. What else may I be missing.

I appreciate any help.  Thank you.

*EDITED for additional info.*

List of possible problems:

• Mobo BIOS may need updating - Have not tested yet.
• Change mobo to Gen 1/2/3 - I had it on auto with the continuous crashing. Currently on Gen 2, and still crashing but not as often. Need to test Gen 1 and Gen 3.
  
• Not enough RAM - I don't think this is the problem. My other 480 rig has 4GB of RAM and runs ETH and XMR mining flawlessly
• PSU - Tested, good.
• 3 risers on 1 SATA power line - None of my risers were 3 to a line
• Drivers - Tested different drivers versions. Each version crashes approximately around the same time. I don't think the driver is the problem.

I would also like to add that when I completed the reset of Windows 10, installed necessary drivers, and waiting for Windows 10 to complete updating, I now have 4 of the GPUs hashing with no freeze problems. Which was my setup previously before I added GPU 5 and 6.  I would need to double check and see if now that the system may seem stable, I can try GPU 5 and/or 6 to see how the rig functions.
This part just seemed weird to me. I can't explain it.

Thank you everyone. Will keep updating.

I'll be working on the miner in a few days. I'll keep this post updated for reference in case someone decides to search for themselves.