Bitcoin Forum

So i have a problem that i can't seem to be able to fix for awhile now

I have a rig with 6 x rx 580 nitro+ 8gb, dual core celeron, 4 gb ram, PSU EVGA 1300 w and a h110 d3a MB.This rig can only function with 5 gpus atm because as soon as i connect the 6th gpu , 2 other gpus along with the last connected gpu mine with 0, so i have 3 gpus that mine with 0.Soon after this happens i get blue screen, and restart.

I operate on windows 10, GPUS are all undervolted with -70 mv, core and memory 1090 and 2090.

I tried switching riser 2 times, switching gpu power cable , switching pci slot order, riser power cable and taking each gpu one at a time.

What i found out is the following.Gpu is fine and can mine ok when it is alone, all gpus are working actually so no problem from here.Risers are ok aswell(slot,cable and pci jack), power cable also.
Everything is ok when i add 2nd,3rd and 4th GPU.

Problem comes up when i add the 5th gpu this time from the opposite part of the rig, so that as soon as mining starts after dag is finished i can see that 2 or 3 cards mine with 0 , and soon after , error screen and restart.
What is this?I never encountered this problem before, i have 3 other rigs exactly like this and they all work fine and stable.
PSU problem?MB problem?

Here are some screenshots

https://postimg.cc/gallery/2aqkgyupe/

I have a feeling a PSU issue, Since you have still 3 working rigs identical I would suggest

- Take one of the working RIG's PSU and swap it to the faulty one

Troubleshooting if its truly a PSU prob.

thank you , will try that in the following days and will let you know !

It seem to be PSU 's problem
That PSU doesn't have enough power to operate all 6 cards
You should test with other PSU to make decision

It's seem like your PSU problem, because as i know power consumption for each RX 580 is 40-80 w. So maybe your PSU something happen with it, even actually with that PSU is enough for 5-6 your GPU. And also about MB, you have tell us if has try every slot of PCIe, and everything is fine. Please try to replace it with other PSU.

So i managed to switch the PSU this weekend, but the problem is not from here.Any other suggestions?

1st if GPU is OC then revise to default value and run for few hours to check, If still the same problem then change check Sata Cable from PSU is okay not burned etc. then change riser sata cable then usb cable then riser.
Issue will be from above one of them. I believe

Get better Air circulation around the rig. I had to shutdown the rigs due to too much heat and bad air circulation. I was getting similar issue

I had experienced this stuff when i do add up a card and troubleshooting up first on the PSU but i didnt resolve the issue. The problem is on sata cable is burned down.
OP, have you checked out those connections? Be sure not to miss a single one and be keen on burned parts.

to use 6+ GPUs on Windows you should -
 
1) turn on Above 4G Decoding in UEFI settings
2) your hard disk drive should use GPT partition scheme
3) and you should boot Windows in UEFI mode (BIOS/CSM/Legacy mode won't work and WILL cause issues exactly like yours when mining with more than 5 cards).

more info about converting your Windows installation to GPT/UEFI boot mode here -
https://www.windowscentral.com/how-convert-mbr-disk-gpt-move-bios-uefi-windows-10

From the looks of it and troubleshootings done on these two suspected parts


you might probably have a faulty board

just wonder if a certain PCI-e slot filled then crash, then taking it empty will be ok

try it out

everything can be possible:
1. the problem with PSU (cables)
2. The powering for/to USB Riser
3. USB Riser/ loose installation.
4. Windows is corrupt
5. a dummy output.

add me on telegram tommyzjh , i can diagnose through teamviewer for a fee :).. easy

be careful OP, do not give home keys to strangers especially if your rig has sensitive data and personal data.

I'll probably give my keys to a trusted member than a newbie ::)

even this forum warned us dealing with them, its always been a precautionary measure

yes, mate (@leonix007). I advise prevention because I see in Meta has many account members hacked in many ways.

agree

who puts sensitive data on their rig? I have configured and successfully diagnosed for more than 100 rigs, welcome to join telegram group chat @Cryptosgminer

each person.
You can't make sure everyone puts/doesn't place sensitive data

that's not a problem you've solved 100 rigs and tried to introduce who you are with a telegram but it's about trust (https://bitcointalk.org/index.php?action=trust;u=1955415) for everyone giving access to their rigs.

In this forum, a trust rating system is an assessment barometer because almost all members don't know their attitude in the real life.

---

why should you give a different telegram to everyone?

I did switch to standard bios on GPUS and i have the same problem.I will change the riser sata cable, i didn't try that yet.Thank you !

It is not a heating problem, i have 10 cm between gpus and rigs are in an air conditioned room.

Yes i def. need to change riser sata cable, i didn't do that yet.If its from the cable it would be an easy fix.Other connections seem fine otherwise.

1.Changed psu, not from here
2.Maybe -  need to try
3.Nope
4.If nothing on the hardware part works, i will need to try a fresh reinstall.
5.How so?I only use onboard vga when i connect the rig to a monitor.Other than that i connect to it with remmote desktop.

I use a h110 d3a MB with a minig mode.As soon as i enable mining mode, 4G is automatically enabled, but i will check bios again to see if everything is ok.
I have GPT partition.
I need to check point 3.

Thanks!

Adjusted page file?

So after a crazy amount of:

Switching to a new PSU, new Motherboard,changed driver version for chipset, switched ram and CPU,windows uefi mode, Reinstalled different version of windows 10, applied a registry blocker,Adjusted page file to 50 GB, Switched other risers, and power cable,Unistalled Afterburner, Made settings in wattman,Reinstalled afterburner, set GPU bios to stock for 1,2,3,4,5,6 Gpus in this order, than back to mining bios,switched riser power cables again, changed again using molex instead of sata and other stuff -  i still did not find the problem.

What i can tell you is what i have observed so far.Sometimes miner starts with 6 gpus with full hash and runs ok,sometimes it doesn't.Usually when it doesn't it is when i pulled the power cable out after shut down and let miner like that for 10-15 minutes(no ideea why).When it doesn't it can give 0 hash for 1-3 GPUS like in the image below.


VERY WEIRD AND I GUESS HERE LIES THE SOLUTION FOR WHO UNDERSTANDS WHY IT IS HAPPENING

When miner starts with 0 hash for some of the GPUs, i have observed that said GPUS with 0 hashrate do not generate theyr DAG.So if i will have 0 hash on gpu 4 5 6 they didn't generate theyr dag at the beginning of the mining process and after a while they "will not respond" while GPUs 1,2,3 generated DAG succesfully at the beginning of mining process and will not give that error.So after the miner works like this for a couple of minutes , it gives the error "Threads not responding, restarting" for all the GPUS that had 0 Hashrate and basically stops mining process for the whole system, this keeps repeating in a loop, with the gpus having 0 hashrate beeing always the same.

I added a Screenshot with this error.

https://imgur.com/a/plF8HvL

Sometimes all 6 gpus complete theyr dags succesfully and miner mines ok, but most of the time 1-3 GPUS do not generate DAGS and soon the respective error follows.

What is this?Does anybody know?It is making me crazy !
Please help !

Did you check GPU memory error from HWinfo ??. if there are no memory errors reported, it means your GPU seems fine with current clocks setting.   

or you can try to add -eres 0 or -eres 1 on the bat file and you can add -lidag 1 or -lidag 2 at the same time.

Hi thanks for answer

I checked with HWinfo and i have no errors.

What are eres and lidag options for?

I got some screenshots here showing the randomness of this weird problem.
In the first Screenshot i am on stock bios and Only 2 gpus generate dag and hash, while on the next 2 screenshots i am on modded bios for mining, here all 6 gpus generate dag and hash.
Thing is it is random, it can happen in reverse aswell, or other times only 3 gpus don't hash or only 1.It has no specfic rule.It is random.

Here it is the first screenshot on stock bios with 4 gpus not generating DAG and hashing with 0.All 6 GPUS are starting up but only 2 generate DAG, so 4 GPUS will hash with 0.

https://imgur.com/a/YBVqiRC

Second pair of screenshots.This is after restart and switching to moded bios.All 6 Gpus start up and all generate dag successfully.

https://imgur.com/a/fUIUHfv

add these function to claymore/phoenix start.bat file or config file. If you want to know about the details, you can found it on readme.txt, under claymore or phoenix folder.

Ok will do that.

Do you know anything about the errors in the screenshots though?

L.E: I added the options you mentioned but still the problem persists

It seems you used phoenix-miner, have you tried using Claymore or ETH-miner ?

Or maybe you can try to mine a different algorithm to figure out the problem.

maybe one of the GPU is playing joke with you
if it was me , i probably try to found this gpu and let rig run without it
you said - now you have exra part to build small rig, just connect only this GPU  and play with it (like new drivers , not OC to much )
try mine differernt algos
who knows maybe it helps

Good bro

Hi so i have found finally the real problem.

One of the GPUS in the rig is not "broken" because it can work fine in gaming and other applications but it cannot sustain the constant heat/pressure of the mining process so after a while it crashes and brings the whole rig down.I suspect that this GPU has a lower quality than the other 5 gpus in the rig.

I changed this GPU and everything started to work fine.

Problem SolveD!