Bitcoin Forum

Hi everyone,

I have the following rig setup:

Windows 10 Pro - newly installed, fully patched
Control panel performance set to high, never turn off etc
Paging set to 18-20 GB
4GB RAM installed - but tested in 8 GB RAM rig and had same issues
Celeron CPU

3 x MSI Radeon RX 580 GPUs installed. I have tried 2 installed directly onto the motherboard plus 1 riser as well as 1 on motherboard and 2 on riser. It made no difference.

Issue is that if I use AMD's blockchain driver the rig will freeze after running for more than 10 minutes. It doesn't crash it just freezes until I do a hard shut down.

If I install either the latest AMD drivers for the GPUs or an earlier version, the rig stays stable but my hashrates drop from 28.x MH/s to 11-13 MH/s

I also tried 3 different BIOS mods with the same effect.

What is a bit odd is that this rig is about 8 months old - maybe a bit more. It was running fine with all 3 cards then I thought one card was causing an issue so I moved that one card to another machine - it, and this original rig continued to operate fine - till a couple weeks ago.

Any suggestions?

UPDATE:
I have another Windows 10 Pro rig running 2 GPUs. I swapped out the 2 GPUs and tested each of the MSI cards individually. Each GPU caused the computer to freeze after 3-6 minutes of running. The AMD blockchain driver was installed.

Are the cards toast for mining?

April 14/19 UPDATE:
After opening each of the GPUs, cleaning off of the old, dry thermal paste and putting on new....I had the exact same issue as described above.

But then I was watching the fan RPM and they didn't seem to breach the 1300 RPM mark regardless of how hot the GPUs got.  I used MSI Afterburner to spin the fans at about 65% (approx. 2200 RPM) and had no further issues.

So it was definitely related to the GPUs overheating and also to the fans not spinning faster to accommodate the increasing heat.

Thanks for all the responses.

Just install newer drivers (I use 18.6.1) and turn compute mode on.

Here is the link to do it automatically. https://bitcointalk.org/index.php?topic=2815803.0

Reboot after running the tool above...

Since you aren't in compute mode, that's why your getting low hashrate.

You said that the AMD drivers made the rig stable, however, regular AMD drivers wont push the GPU to the limit when mining and the cause might be the GPU's HSF, the entire cooling system (including thermal paste) or the card acting up.

-When was the last time you cleaned the HSF or replaced the Thermal compound?
-Have you tried to mine using the rig installed with only the "Good" GPUs?
---If it also freeze with the good ones, it must be something else (the OS's updates, driver, PSU issues or RAM slot problem).

That got the hashrate up to 28 MH/s but the rig froze after under 10 minutes still. Thanks tho!

Since you test the GPUs to other working rig and the result is the same(freezes) it means the problem is under your GPUs not in your windows OS, ram PSU or other components.

Like the above said your temp might increase too high so you must monitor the temp or much better to clean your GPU and replace a thermal paste to make your GPU fresh again(just do this if it's not under warranty).

Maybe try to return to base BIOS. Also, check your power supply and rysers, I had the same and problem was at the power supply. 

It looks like you are running windows 18xx version now. Check the ULPS and Crossfire status and make sure it's already disabled. Sometimes these two things can mess with your system when you used the newest windows (18xx version) or you decided to update your old Windows (1709 and earlier) to a newer version.

To disable it, run regedit and find EnableCrossFireAutoLink and  EnableUlps then modify the value to "0" , repeat this step for each gpu folder and then restart your system.
(it was usually inside Computer\HKEY_LOCAL_MACHINE\SYSTEM\ControlSet001\Control\Class\{4d36e968-e325-11ce-bfc1-08002be10318}\)

It would be necessary to know more details, like those asked on the last posts... if we need to just guess, I would go for temperature... have you checked the working temperature of the GPUs just before they crash ?
Sometimes people don't care much about the details like cleaning, thermal paste, etc, but as far as hardware longevity is concerned, these are very important details...
Some algos also cause strange behaviour in some equipment... it happened to me that 2 GPUs of same type/RAM, etc but different brand, worked with 50% difference on hashrate when using a certain algo...
good luck

Make a few simple steps that we all do before complaining about rig instability, the few things are, buy the latest version of the risers, update/flash the bios to the latest version, turn off any onboard device that is not needed like the audio card.

Also the Windows installation should be a clean one, not one downloaded from torrents, I had probelms before with a Windows I downloaded from a torrent.

Please check your risers (I had a similar problem and I solved it by replacing one of the risers). There may also be a problem in the power supply.

I had the same problems at first.
Here's what you can do:

STOP WINDOWS UPDATE
RUN - type “services.msc“
Search “Windows Update“
Click “Stop” if status = “Live” or “Checking/Running“
On “Startup type“ select “Disabled”
Apply & Restart