Bitcoin Forum

Having an issue with ASUS B250 Mining Expert, where windows starts to lag after a 5th card is added. For now I plan to run 12-13 AMD cards.

3 or 4 cards seems ok, but with 5th windows lag is unusable. Opening device manager, seems like its refreshing every ~30 sec as in like looking for devices or something, which is weird.

This I tried so far:
- With & without iGPU
- various slot, card & riser combinations to rule out defective hardware
- installing drivers via windows update & installing manually from AMD's site (DDU cleaned each time).
- update latest BIOS & Asus chipset drivers
- set virtual memory at 36-52GB
- Windows Home vs. Windows Pro


Hardware:
- CPU: Pentium G4560 Kaby Lake Dual-Core 3.5 GHz
- 8 GB RAM / 120gb SSD
- MSI RX 570's (unflashed, yet)
- Windows 10 Pro (latest updates)

Anyone face a similar issue?

Any pointers at this point would be appreciated. Thanks.

That doesnt happen in linux, Windows is a terrible choice for a mining OS, especially with that many cards.

Reimage it with a good linux mining OS and all of your problems will go away.

Well if it would start to hiccup after 7 or 8 then I would agree. But 5 cards should be easy for any mining board. I'm running a bunch of 5 - 7 GPU rigs w/o any issues on Windows. I just think this board is very fussy..

I'm running 16 GB ram on ASUS B250 so maybe that could help.  What about allocating virtual memory?  Not sure if that would help but worth a try.  

And... is it lagging just when mining or all the time?  If it's just when mining maybe try a different algorithm to see if it is any better.  Of course, you'll want to mine the algo you want to mine but at least you will know if that algo is contributing to headaches.  

So... things to try:

Set windows for optimal performance instead of visuals...  http://www.thewindowsclub.com/disable-visual-effects-windows

Update BIOS

More RAM

Allocate virtual memory: https://www.x-plane.com/kb/increasing-virtual-memory-on-windows/

monitor resource usage in task manager to see what may be consuming processing power:  https://www.windowscentral.com/how-identify-and-terminate-high-resource-processes-using-task-manager-windows-10

you must install Windows on GPT partition and boot Windows in UEFI mode to properly use 5+ cards.

Nuts!  I hadn't heard of this requirement.  Got me thinking... how do I know what I have and how do I convert to GPT if I'm MBR???

Found this:

https://www.easyuefi.com/resource/check-mbr-or-gpt.html

and this:

https://www.windowscentral.com/how-convert-mbr-disk-gpt-move-bios-uefi-windows-10

for background:

https://docs.microsoft.com/en-us/windows-server/storage/disk-management/change-an-mbr-disk-into-a-gpt-disk

Guess I know what I'll be doing tonight  :)

Hey guys! I am having exactly the same problem. Here is my thread: https://bitcointalk.org/index.php?topic=4353979.0
I opened another because the problem is not associated with the ASUS B250 Mining Expert. I had it on a Gigabyte too. Check my thread to read what I tried so far.

edit: So I read its a GPT Problem? But windows 10 on a SSD automatically sets UEFI and GPT, or not?
All my other 12 GPU rigs did not require any further settings (but they also are all NVIDIA Rigs hmm...)

Checked. Its GPT and I still have the problem!

Regards

Thanks for replies.

I tried updating BIOS & setting Virtual Memory at 36-52GB (added to orig. post). But problem still remains...

I know this is the correct way to do it with Biostar's mining specific motherboards, so I believe it would be the same for other manufacturers as well. We run 6-8 GPU systems on Windows with more then half of them on Intel Dual Core Celeron or Pentium CPUs and they work fine for us.

I run several Rigs with 12 NVIDIA GPUs on Gigabyte Mainboards and never had a Problem.
So it must have something to do with AMD Cards....

I agree, it must have something to do with AMD cards.

I experienced the same thing before on one of my VEGA rigs. I ran 13-GPUs on a G4560 Petium processor with 4GB RAM. GPUs were 4 VEGAs and 7 RX470s. Opening radeon settings took forever too. I reinstalled everything and used blockchain drivers instead, works perfectly even up until today.

Did you ever find out what was the actual issue?

I've reinstalled Windows 2 times now, doesn't seem to make a difference.

Also, just curious did you re-install with all GPUs connected or connect 1 at a time afterwards?

Hi, I am still on it since it occured 2 days ago for me. I need a solution.
So far, reinstalling Windows or drivers did not make a change. Also I will post mor details in my thread, since it has a more open topic. I opened it to get more attention since the problem is not only related to the B250 Mining Expert. I had the same behaviour on a Gigabyte and on a MSI board.

Thanks for mentioning that. Perhaps its a common issue.

Have you tried opening Task Manager to see if anything is eating up the CPU when lag occurs?

For me it seams to be "Windows audio endpoint builder", but not always.. but killing it or stop/disabling the service does not seem to solve it..

I also checked that. That was the first thing I have done. Nothing suspicious.
Furthermore I just installed the drivers, without the settings and hdmi drivers etc. Did not work either.

So now I will check some BIOS settings. But if this also does not help, I am out of ideas.
Did you find out something I did not mention so far?

For everybody here saying AMD cards don't work on ASUS B250, or that Windows is a shitty solution :)

RIG-A :: 12 x 580 Special Edition GPUs (https://i.imgur.com/v9Ny9mp.png)
RIG-B :: 12 x 580 Special Edition GPUs (https://i.imgur.com/p42KCac.png)
RIG-C :: 12 x 580 Special Edition GPUs (https://i.imgur.com/8be6obx.png)

All use Windows, all running on one 4GB stick of RAM ( proof here (https://i.imgur.com/cHCWUG8.png) ), at about 50 degrees average temps, as you can see in the Claymore miner info, all rigs drawing 1.830 watts each, from 2 pairs of 1400w Delta Electronic server PSUs.

Rigs pic ( all are the same ): https://i.imgur.com/jbxfS4W.jpg

So please, stop with the 16GB RAM tips, or the GPT UEFI partitions, or whatever other crazy idea comes to mind.

Hey mate,
nobody says it is not working in general! We are saying it does not work for us. And we try to find out why.

Maybe you can help us find the problem. I have 4x of the same 580 Special Edition cards. right now 3 of them are working with 5x 1070 Tis. When I add the 4th 580, system lags.

What drivers do you use?
Did you change anything special in general?

Regards

I spent days on BIOS settings, nothing seemed to work. Hopefully you can find something I missed.

The only additional thing that comes to mind is to swap the Pentium CPU for i3.

I just had i3-8100 delivered, going to give it a try. Will share results in a bit.

Here's a crazy thought, can Windows activation possibly have anything to do with this?

Trying on un-activated Win 10 instance..

I am using an i3. It does not help at all...
Also (just a coincidence) this one Rig I am using has an activated windows.
So both points should not be relevant.  :(

It occurs AFTER installing drivers, right?
So maybe we have to try even older drivers?

Can you install the Blockchain drivers? I am getting blue screen.

That board is one of the most problematic ones in Windows, I have been able to successfully run 11 cards though with only 8 GB of HyperX RAM DDR4 and an G4400 processor. It is definitely a driver problem from what I see. However I would suggest the following:

Update BIOS of motherboard, turn all PCH in BIOS in Gen1 as a riser slot is Gen1 . Install Windows , UEFI one , you can download it anywhere for free and buy a sticker if you want it activated (but that is not relevant). Install only 1 card and install latest AMD drivers from the official website. Make sure to disable windows updates at services.msc , shut down your PC. Plug all the cards you have an turn it on, it should recognise all cards without any issues.

This was my personal experience with this board and it recognised up to 11 RX 580 cards.

Yeah I feel we're describing the same issue. It happens as soon as driver installation is finished. Very often the installation does not complete and rig will freeze, then reboot by itself.

I haven't tried original Blockchain drivers since current drivers have it.

Have you tried connecting GPU's 1 by 1 and setting each GPU to compute mode before connecting the next?

Need 16gb ram.  I had the same issue with 2 of my 13 gpu rig and 16 gb with 32 gb virtual mem solved the problem

I have 16 GB RAM and virtual memory is 96GB minimum. So thats not the problem.

If you open Device Manager and leave it, does it refresh for you every 15-30 sec, as in like scanning for device?

Mine seems to do so..

Yes it does that.
I will try tomorrow a clean install, then putting in one vega, install drivers, reboot, check and so on. Erverybody says that should work.

Ok, a bit of progress to report, on the Device Manager refreshing issue. Seems to be resolved by unplugging unused pcie usb cables that were still plugged in to the motherboard. Or by setting 3 available BIOS settings to Gen 1. I made a bunch of changes together..

So the Device Manager now behaves normally, but lag is still there. I have to say it is much better then before, at least windows is somewhat use-able. After all this time spent, I'll consider this a small "an improvement"..

Its windows 10 problem.

Windows 10 Pro
1703 running fine
upgraded to 1709 running fine
1709 upgrade 1803 random freezing, start menu not working, etc

Then done:-
clean install 1803 random freezing, start menu not working, etc
clean install 1709 random freezing, start menu not working, etc
clean install 1703 running fine again

Ok, so I will try a bunch of things on 1803 and then will switch back to 1703 if you say it works.
Funny thing is that I had the startmenu not working problem on my NVIDIA rigs. Rebooting several times helped and then the problem disappeared. It was gone since 1803.  ???

Ok, so I got it up and running!
Thank you toptek12 for the useful script.

The thing I did not notice and that I was up to try next was the Crossfire settings. I remembered that someone told me something about this for AMD cards (these were my first AMD cards for mining). You have to turn Crossfire and Ultra Low Power mode off in order to get the cards to work. This is possible via the registry, what takes some time to set. Then I found out that the AMD-Compute-n-Stability-Script_v2.0 that toptek12 mentioned in another thread does exactly that in a conveniant way. So I used it (DL from Anoraktech).

Step by step:
1.) Remove any drivers in Safe mode with DDU.
2.) Reboot, get into BIOS and make sure everything related to Link speeds is on Gen1. Disable the 2 virtualisation settings. Enable above 4G encoding.
3.) Boot.
4.) Shut down.
5.) Plug in ONE single Vega.
6.) Start up, install recent AMD drivers (ONLY drivers!).
7.) Reboot, check if everything works.
8.) Shut down.
9.) Plug in all the other cards. I just had the riser disconnected for convenience.
10.) Reboot, open Device Manager and wait until every card is installed and no triangles or unknown devices are shown.
11.) DONT REBOOT! Apply the AMD-Compute-n-Stability-Script_v2.0 (Both parts, read the readme.)
12.) Only after this is done reboot.

Everything should work fine now.

Regards

Bingo! Just tried this, and can confirm this works.

This is the most frustrating motherboard I ever had to work on.

Many thanks for sharing the steps!

Glad to hear.
I can recommend OverdriveN Tool for configuration. If you dont install the AMD settings...

Increase windows OS virtual memory and also pls decide to stop using windows OS and start using mining OS like hiveOS, this will erase any memory issues you having when using windows OS

Does this have anything to do with my motherboard with 6x pcie slots that only boots with 5 GPUs? Anything over 5 GPU the motherboard won't boot up at all 🤔🤔 so weird and confusing