Help me please: boards turn off post start

Hi

I need your help.

My rig have a fail that can’t resolve.

I have searched for information during the last 20 days without success, none seem to have a similar problem

It had 6 boards Nvidia RTX 3070, 3 msi and 3 Gigabyte.

Since the update 220 of mid-January the 3 gigabytes are turned off.

Hiveos sees them, I corroborate it through the roxTerm

At boot, when the miner start (gminer or miniz that’s the same) there the boards are turned off.

Configuration:
Motherboard Asus Prime B365M-A
Intel Celeron G4930@3.20Ghz.
RAM: 4Gb
Disk: usb 15.4Gb
3 x nvidia gforce RTX MSI 3070 8Gb
3 x nvidia gforce RTX Gigabyte 3070 8Gb

Hiveos 0.6-220@230127
A20.40 (5.11.0701)
N525.85.05

what i have tried:
-swap powers units.
-swap rizers.
-swap usbs’s cable.
-check bios settings.
-Hiveos upgrade.
-I try a downgrade but get an error -100.
-Upgrade nvidia drivers to 525.85.05 version via Maintenance mode.
-OC to default values for core, mem and PL.

With each test the behavior is always the same, the msi work perfectly, the gigabyte not.

Any new idea / suggest ?

Thank you very much.

PS: I can’t post screenshots because I new in the neighborhood.

PS: sorry for my english, I speek spanish.

Try a fresh install with the latest stable image on a known good drive (ssd preferably) you’re on an old image so there’s no sense in even spending time troubleshooting it before flashing the latest image.

ok I understand.
I hoped there was another way.
Thank you very much for your suggestion and your time.

Hi again.
Today news.
I proceeded as you suggested.
The problem persists.
But, I can see a difference at the time it occurs (via RoxTerm):
when gminer starts, the 6 boards are energized, with their fans at 100% and reporting their status (temp, voltage, etc), when connecting to the pool (flux.2miners.com:9090) it receives the first job and there stops the 3 gigabyte’s boards, and its power unit too.
This matter has me very perplexed.
Thanks.

Post a screenshot of your worker overview showing cards, ocs, driver verison, hive version etc.

Hi, the images requested.
The OC PL on 100 is the Last test, It didn’t work either.
Thanks for time.




In this can see when the boards go off:

Hey! what wattage is the powersupply that is connected to the 3 gigabyte cards?

Hi
750 w each of them.
There’re 2 identical: 1 for 3 msi and 1 for 3 gigabyte.
The rig is operational since september 2021, generate Eth with trex until the merge, this problem appeared 15 days ago. The powersupply has been checked obviously.
I have thought about the possibility that, from some new requirement, an additional connection between them was necessary, but I have not found anything about it.
Thank you.

Looks like your issue is psu related. Try swapping parts around and narrow it down

I had tested it before using the new version.
I’ll try it now with it.
Thanks.

Hey sarok! what algorithm are you mining? I’m wondering if, since you were mining ETH the algorithm you are mining now is a lot more power hungry. Could be the psu is getting slightly overtaxed, and the other one is just barely under the maximum. You could try lowering the powerlimit on your 3070’s to like 130 watts just to see if thats the issue. Hope this helps!

Hello.
Exactly.
That’s the problem, I was writing my conclusions to share with you.
I am now using equihash 125/4 for Flux.
The strange thing is that they are set that way (130w) but at startup it ignores it, and they start at maximum (270w).
My conclusions below.
Thank you

Hello.
Well, I think I have found the problem.
I don’t understand how I didn’t try it before…
Disconnect only 1 Gigabyte from the set of 3.
Everything is running as it should.
From this I see that the Gigabyte have a different power scheme than the MSI: 100w, 220w and 270w of the MSI vs 100w, 270w and 270w of the Gigabyte.
My conclusion is that the Gigabytes start at maximum power (270w) exceeding the limit of the PSU (750w) when all 3 are together.
By applying the OC, consumption and temperature drop.
Which doesn’t seem to be done naturally at boot.
And there is the difference with respect to the previous situation.
I will continue to work on this and keep you posted.
Thank you so much for your help.

1 Like

If you set your core clock you don’t need to set a pl.

Well, I didn’t know that.
Thank you very much.

Alright! glad i could help somewhat :slight_smile:

This topic was automatically closed 416 days after the last reply. New replies are no longer allowed.