Hash rate drops to 10% - B250

AMD Cards
1

Hi all,

running several ASUS B250 mining expert and this an issue we just can not get around and dont seem to be able to fix it.

Using RX570s from Sapphire, with a simple OC they run between 29 and mhs for Ethereum on Nanopool. So far so good.

13 GPUs are connected to 1 rig, using proper risers.

Out of the blue, ALL cards drop to a max of 3 mhs and there is absolutely no way of recovering.

  1. Rebuild the rig - Done, no success. They mined for 10 min on high speeds, then back to the original issue.
  2. Tested other PSUs - Done, no success.
  3. Reinstall - Done, also no success. Workes for around 10 min, then they crash again.
  4. BIOS update - Done, does not matter which bios we run. Latest and older version show the same error.

Does anyone have any clue? Desperate here!

2

First at all check your BIOS settings. You need set PCI-E to Gen2 for ALL PCI-E slots!

3

OK, just set that.

First thing I notice getting a lot of errors saying couldn’t schedule ib on ring

Followed by a soft lockup - CPU stuck for 22s!

And thats the end of it :slight_smile:

UPDATE:

Created a new USB drive with latest HIVE OS. Running only 4 cars power by 1 PSU. PCI is set to gen3

4

Well, tried some more things:

  • Changed mobo to Gen2 and Gen 3 - both no go
  • First boot is normally successfull and mines at decent rate
  • After first boot, no luck. Keep getting CPU stuck errors, and IB ring errors

So I got pissed…

Bought a new BT250 mobo, new CPU, new RAM, and brand new risers. Took 6 new RX570s out of the box, grabbed a SSD and…

Same shit again.

I have really no idea how to troubleshoot this. Anyone any idea or willing to help?

5

With Gen3 in x1 risers it’s unworkable

7

Well, just tried again, set it to gen2.

Another fresh install on SSD, and IMMEDIATELY IB ring failure.

Stock cards, new risers, everything new out of the box. I have no idea anymore…

8

Update:

Installed Windows 10 on the Mobo, changed nothing else in setup and tadaa, running smooth even with HiveOS.

I seriously have no idea why I can not run HiveOS.

9

For anyone running into the same issue, I just bought 3 other motherboards today.

Gigabyte GA-H110-D3A and just switched all the other hardware over.

BOOM, all issues gone. Smooth sailing on 30MH/s per RX570.

10

Two defective motherboards? Very unlucky or product quality has dropped

11

2?? I wish. I have 6 of them all the same issue. I still believe it has something to do with the mix of hardware

12

So, I have several B250 that I have the same problem with as well. Primarily, the CPU soft lockup garbage. I tried another Linux mining OS and it did the same thing.

The hash rate drop is interesting. I have one rig with all Nvidia cards that I didn’t have the problem with. The other two are AMD and I have the problem with. On one I found if I backed down the OC settings it stabilized. The last one I have to go back to the farm and power it off because the CPU soft lockup has the CPU pegged and it’s not responding to Hive reboot commands.

I found that running all PCIe settings in Gen 2 seem to work the best. I also found that when I’m getting the amdgpu driver errors or CPU soft lockup if I do a hard power cycle (sometimes a couple of times) it’ll come up clean.

I am sure that it’s something to do with this board and Linux. From my research the CPU soft lockup has been attributed to power to the CPU. I don’t know that this is true because I’ve got 2 x Corsair 1500w platinum PSU and it will do this with or without cards plugged in.

Anyway, I’m looking into different CPU settings in the BIOS trying to see if any of them make a difference. The latest BIOS update as of 07/20/2018 didn’t fix the issue, FYI.

Let me know if you figure something out and I’ll do the same.

1 Like
13

Well… nothing worked. I gave up and swapped new mobos. All good since then.

14

What board did you end up going with?

15

I am now using the Gigabyte H110-D3A. No issues whatsoever and they even give me 2mhs per card more!

Tonight I tried for fun turning off a working rig, put the B250 back and immediately all the errors were back.

I still believe its a Linux issue…

16

I’m having similar issues with Asus Motherboards. Specifically the Z270-a and AMD cards

17

I have the Z270E and followed the instructions for the 270A.

Problem solved