Bitcoin Forum

Hi guys !

I have recently built a new mining rig over the course of the weekend however I have come across a few issues after installing the OS.

I currently only have a total of 3 GPU's (looking to add more at a later time) all 3 GPU's are fully recognised by Windows 10 however one of them states an error under device manager:
"This device cannot find enough free resources that it can use. (Code 12)
If you want to use this device, you will need to disable one of the other devices on this system."

Image:
http://i64.tinypic.com/2h38xgl.png

I've installed to the latest drivers, I've even tried moving the cards around on difference PCIe slots on the board however the issue still remains.

My main system specs are as follows:

- Intel Pentium G3260 3.3GHZ CPU
- Asrock H81 BTC PRO R2.0 Mobo
- 4GB DDR3 Hyper X 1866Mhz Memory
- 3 x Nvidia GTX 1060 3GB GPU's
- EVGA 1000W PSU
- 60GB SSD
- Windows 10 Pro 64bit (un-activated)

Other methods I've tried before resulting in posting for advice here:

- Flashed my motherboard to the latest BIOS.
- Increasing virtual memory to 16GB and even tried to look for the TOLUD setting in BIOS but I cant find that setting anywhere.
- Set the PCIe lanes to Gen 1 in the BIOS.
- I've also tried unplugging a GPU in different orders and when two are running together the Code 12 message doesn't appear in device manager, however when 3 are plugged in and powered on the Code 12 error will appear on one of the cards.

All 3 GPU's are working as I've tested them individually. I've also tested the PCI risers individually and it appears they all work too..
Is it possible any of my system specs are being bottle necked somewhere here? Are the newer GTX 1060's to much for a LGA1150 system? Can't figure it out  ???

If there is a guru out there who knows how to resolve this I would be eternally great full as I've spent countless hours on trying to sort this out with no avail.  :'(

Many thanks

enabled 4g decoding?

Checked the image dude. It states that your display adapter does not detect one of card you have added in your PCI slot. Check the riser line from the PCI whether the connectivity lined up properly or not.
In PCI bus it is been connected dude. Try to change the card in different slot whether it is working or not to confirm the issue.

Check this link. It might be useful for you.
https://hardforum.com/threads/gpus-not-recognized-with-pci-e-risers.1936648/

Hey man, thanks for the reply, I've done some research into that and the Asrock PRO BTC R2.0 motherboard doesn't allow you to amend this feature, supposedly the BIOS is pre configured and designed for mining out of the box. (I presume 4g decoding is already enabled)

Hey thanks for the reply bro, I've already tested this by changing the risers in multiple orders but the problem still persists, for some weird reason it just doesn't like having a 3rd GPU on the motherboard. I really cant figure out what the underlying issue is.  :-[  I've also tried fresh installing the drivers in safe mode multiple times thinking it would be software related.

EDIT: updated on a few posts below, appears to be issues moving the risers around.

Hey dude try to install the GPU Drive in the picture your Motherboard can't read properly your GPU try it bro.......

I would reset the BIOS to default and then only disable onboard stuff you don't need through the BIOS.

Okay so I've just come across another issue, I  started to move the risers and around a bit and put them all in a different configuration and now only 1 GPU is showing! (The one in the PCIe x16 lane) Its bizarre, do you guys think its possibly a motherboard issue here? Or maybe the risers are the problem? Or is it because I installed the GPU drivers in the certain PCI slots originally?

http://i63.tinypic.com/2cz9b90.png

EDIT: Re-installed the drivers when the GPU's are in the different PCI slots, all three now show however one still has the Code 12 error within device manager.

Hey man, I did have the factory BIOS in place originally and had the same issues, this is what made me flash the BIOS to the later version in the end but the end result left with the same problem.  :'(


Hey, sorry man I don't quite get what you mean!  :D

So what you want to do is plug in each GPU to the 16x slot to make sure they are working and recognized.  Then when you know they are all working properly you test each Riser package on one GPU to make sure all the risers parts work.  Once you've established those are not the problem then you can dig around the BIOS and windows drivers.  Usually when there is an exclamation mark in device manager you can manually point the GPU to the proper drivers on your hard drive, which usually fixes it.

Hey bro, yes I've already performed these checks however since you mentioned it again I just checked again, all 3 cards work in 16x slot, and each one of the risers work fine and boot into windows visually no problems.

I am debating to do a fresh install Windows 10 as my last resort to see if there is something wrong possibly a corrupt registry file in the OS. Other than that I really cant think of what else I can do to resolve this issue.

One thing to note is i'm not currently using an activated version of Windows 10, I know this should only effect the personalised options and I wouldn't expect it to restrict the amount of VGA cards used.
I am not to sure if anyone else can confirm they use a non activated version of windows 10 with 2+ GPU's running fine can they?

Thanks again guys for your input if any one else has any more solutions please send them this way!!

I don't use an activated windows 10 and run 8 GPU's just fine.  So you never did mention if you disabled everything in bios and made sure everything was set properly.  You should really look up the settings you need to activate/disable as they can hog up pci lanes that could contribute to the error you have.

Hey man, yeah I believe that this particular motherboard is very limited in what you can turn off and turn on, due to it being a "mining" motherboard there isn't really to much I can change as its meant to be preset ready with certain features already enabled or disabled. The only main thing I can change is the PCIe speed to Gen 1 or Gen 2 in which I have tried changing a few times to see if it has made a difference, but the problem still persists.

Only thing left I can think of is did you plug in the molex power connectors to the mobo?  You shouldnt need them if you are using powered risers, but mobo's can be picky.

I did indeed, I've plugged both 4 pin molex connectors in by the PCI slots even though the risers I am using are powered ones, i've tried both with them plugged in and without them plugged in.  :'(

There is a setting called "Share Memory" and it is originally on"Auto". I haven't ever really checked what does it too, but maybe it is the same as TOLUD? Maybe You can change it to something better?

In BIOS go Advanced\Chipset Configuration and find "Share Memory"

Maybe, just maybe it helps...

Hello,

At which speed did you configure your GPU ? Try to put them in PCI Express "Gen1" in the BIOS

It could also be your PSU not big enough, but 1000W should be OK for 3*GTX 1060...

Which generation of risers is it ? Which color ?

Hey, man, speed is set to Gen1 in the BIOS, I worked out a 1000W PSU should be enough to power 6 x 1060's.
These are the risers I purchased from amazon: https://www.amazon.co.uk/gp/product/B074CSZSSD/ref=oh_aui_detailpage_o09_s00?ie=UTF8&psc=1

The build quality of them do not seem to bad, they all work, but do you think these could be a reason why the 3 GPU's are not working in unison?


Hey man, interesting I will give that a go and report back asap.

UPDATE: Tried multiple settings but with no luck  :'(

1000w PSU is pretty borderline for mining with 6 card rig, to be honest (Always give yourself headroom with PSU. Efficiency is better and You can have a nice night sleep knowing that You aren't pushing Your PSU too hard)
Try GEN2 on Your PCI-E lanes.

That is very true!! and yes I tried Gen 2 a still problems with the 3rd GPU  :'(

I really don't know what else to try, I completed another fresh install of windows thinking it would be maybe a reg key that had got corrupt but yeah didn't work..

Debating to buy 3 more PCI risers and possibly RMA the motherboard for a new one or different one..

I've tried to reach out to Asrock but they haven't got back to me yet. Its so frustrating because this motherboard is meant to be a mining board, I wasn't expecting any of these problems...

Use your IGPU and test each pcie slot individually to see if you just have faulty pcie slots.

I RMAd one of my H81 MOBOs on 20.09 (no POST) and I haven't heard anything about it from then... It seems that a lot of AsRock H81 Pro boards are faulty in some stupid way >:(
Hope everything works out for You and You can get a new MOBO fast if this problem doesn't get fixed!

That's also a shoot worth taking!

I've had some problems with the exact same PCIe riser you're using

Don't know why, but I never made it work properly, I have to change to this kind of risers (Molex risers) : https://www.geewiz.co.za/54639-large_default/pcie-pci-e-usb-riser-card-molex-and-sata-powered-gpu-mining-1x-16x.jpg

Never explained why  ???

Checked this, everything seems okay here


Thats crazy, im glad you managed to get that sorted!! And thanks, I've literally spent nearly 12 hours today to try resolve this, have never been so stressed in such a long time haha. I think I will literally just RMA the board at this rate. Praying that will solve the issue.


No way really, did you have the same issue as me? Or was it something slightly different, I am also debating to just purchase 3 new PCI risers to see if this would actually make any difference

I know the stress when shit like this happens! Being a student and after building a mining rig that doesn't POST with the hard earned money... Had times like You do right now, just spent days to get something to work. But in the end things worked out, they usually do! ;)

We can only hope my friend..  :D

Really appreciating the input from everyone that is trying to help.. Keep them coming please!

Think google might have found a solution for you.

Hey bro, already tried looking for this setting, its no where to be seen within the BIOS  :-[

UPDATE:

Okay, so I've finally heard back from Asrock.. They attached a later version of the BIOS to me on an email (V1.30B).

Weirdly their latest version that I could find on their website was V1.20B which is currently the one my motherboard is flashed to. They have also now told me that this later version has the "4G Decoding" feature in the settings.. :-X !!

I am going to flash the BIOS later tonight when I have some free time and with a bit of luck I am praying this will resolve everything. I will keep you all updated with the progress!

Glad you actually got some help from Asrock.  Good luck with the flash.

Well to no surprise the flash to the later BIOS didn't work with 4G encoding enabled, I've emailed them back to see if there are any other solutions.  :-[

I've also ordered 3 more PCI risers (molex) type to see if some bizarre reason this will fix it. If that doesn't work then i'm going to RMA the board and try again.

If all else fails I will buy a different motherboard completely, do you guys have any good suggestions on a decent mining socket 1150 board?

You can try BIOSTAR for LGA 1150 socket, but I think its about the same crapshoot as ASRock
https://www.newegg.com/Product/Product.aspx?Item=N82E16813138436

Set UEFI to IGPU as primary.  Use a dummy plug for that HDMI, do not use dummy plugs for other GPUs.
If that doesn't work, increase the physical RAM to 8GB.

Pretty borderline for lamers. For Ethereum mining, it has enough overhead. It has reserve even for core-intensive algorithms.
I have a rig consisting of 4x GTX1060 and 1x GTX1050Ti (60% of GTX1060 in both power and performance) overclocked and power-limited, it delivers 110 MH/s draining 510W from the socket. Factoring the PSU efficiency, I estimate the power consumption at 470W for 4.6x 1060.

Thanks for the advice man, I've tried setting the IGPU as primary and still get the same issues. (See image below).

http://i65.tinypic.com/2d0ifdc.png

I haven't tried upgrading to another 4GB's worth of RAM, but could this really be the reason? Pretty sure i've seen systems running 6 GPU's on 4GB's of RAM. Its an odd one.

I also RMA'd my motherboard thinking this would be the problem but still the same issues :'(

Really cant think what else to do, really stuck in limbo right now... So stressful for a first time rig loool. Debating to try a completely new motherboard maybe a later generation socket than 1150, really not sure its been a nightmare.

Haggyy,

I have the *EXACT* same problem as you.  I've spent the last 12+ hours debugging to no avail.  I've read every post online umpteen times before I saw your post and realized it was recent.  I see the same blasted Code 12 "not enough free resources" error on my cards.

I'm uisng 6x GTX 1070 cards and 6 powered risers on the AsRock H81 Pro BTC R2.0 motherboard.  I have 8GB RAM in my rig and I can tell you the amount of RAM doesn't matter for this issue, probably not worth you upgrading.  My PSU is a Corsair 1200W platinum, not that it matters.
Even though I'm using 6 cards, I had the same problem with fewer cards.  My first 2 cards worked fine so I ordered more.  After I added 2 more cards (total of 4), I could only get some of them working.  When my last 2 cards arrived (total of 6), 4 out of 6 are working.

4 of my cards work great and I can mine on them no problem.  2 of the cards have the yellow bang in Device Manager.  

I also disabled non-essentials in BIOS, changed out risers with working video cards, reloaded Win10 from scratch, checked and re-checked all connections, set virtual memory to 16,384 Mb, set Bios to gen1, NONE of it works.

Both molex plugs are in the mobo.  I also contacted AsRock, they sent me the 1.30 beta Bios and said turn on the "4G Decoding" setting but it didn't work.

This is my first time building a rig and it's working, but at two-thirds hash rate...

I've re-ordered risers to change them out to see if that's the issue since I see some forum users reporting bad risers causing this.  After trying that, I'm going to return the mobo and get a replacement.

Screenshot: http://oi67.tinypic.com/24q67ia.jpg

Have you made any progress?
We may need to start a therapy support group for the pain and suffering caused...

Russell

well at this point theres really only a few things that are going to cause a not enough free resources

1 would be the cpu doesn't have support enough pci express lanes, number 2 could also be not enough ram, all devices need some ram to some extent even the gpus use motherboard ram, last but not least could be a straight up motherboard problem but I'm leaning twards not enough of either pci express lanes or not enough ram, so my advice would be to add more ram if your using anything less than 8gb, windows page file shouldn't be anything less than 16gb and as some people have mentioned, disable the stuff that isn't needed as those things do take up resources, things like sound are not needed to be active

@russellvb for you since your already running 8gb of system ram, id say the cpu could be your problem, not all cpus support the same amount of pci express lanes, that said if you have either a kaby lake or whats the other one? one of them doesn't support more than 4 gpus and the other works just fine with 6+ gpus.

Hi Russel

I cant believe you're having the same issues to, what a dilemma.. ASRock also advised me to install to a previous version of windows but tbh I really do not think this is the problem.

The amount of hours I've spent doing this to is crazy it hasn't been a smooth experience to say the least lol..  :D

No more progress as of yet I was going to wait until this weekend to get my hands dug into it again. I'm literally debating to change to a completely new motherboard and move to a later generation socket praying that will fix the issues. Please let me know how you get on with the new risers that is one thing I never tried, I have stuck to the same 3 PCI risers that I know for a fact work as they have been tested individually over and over again.

Oh and I noticed you said you were going to replace the mobo however I have already tried this and results are still the same.. Not to sure if that will help save you some time.

Hahahaha and yeah we will definitely need some serious therapy after this haha.

Keep me updated on how you are getting on man.

ALSO: Interesting point CryptoWatcher420 made about the CPU handling more than a certain amount of cards, my CPU in my mining rig is a (Intel Pentium G3260, 3.3GHz, 3MB Cache, LGA1150, 2 Core, 2 Threads) - If anyone is clued up on CPU's and their interaction with GPU's please let me know, is it possible that my CPU cant handle these later generation cards perhaps?

https://bitcointalk.org/index.php?topic=2401059.new#new
Maybe this will help you. One guy came up with a solution!

My CPU is a Intel Celeron G1840 Processor.  It was recommended on a mining forum.  I don't have a different one to swap with, but I could buy a better one.  Would you suggest an i3?

The processor here there is nothing. I am sure that if you replace it then you have the problem not solved. Error code 12 is a problem with the driver. You need to pay attention to software. If you modified the BIOS of your GPU then you will not fit a standard driver. At me such problem was with one card but after updating the driver everything works fine.

https://bitcointalk.org/index.php?topic=2401059.new#new One guy found a way how to probably fix it. Go check it out! :)

Have you tried manually selecting the driver for the erroring card? ie: right click, select driver, etc?

Also try disabling and then enabling the card a couple of times. Sometimes that jolts them into working (on older ASrock boards anyway)...

If you've installed the drivers several times you might need to ddu them and start again.

Holy crap guys...I made progress.  Thanks to the link provided by smoolae, I have 5 cards working.  Only 1 card has the yellow bang now!

Here's a copy/paste of the post from the other post:
https://bitcointalk.org/index.php?topic=2401059.new#new

make new key "HackFlags" as (DWORD 32) set value 0x600 as (Hex) in
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\PnP\Pci\
and reboot.

User q43a0ef061ej provided this information.  Maybe if I fiddle with things a bit more I can get all 6 cards working!?

AsRock should think about this problem. Because it's affecting only on AsRock motherboards.
Had no problem with Biostar motherboard yet.

Hey guys, I figured mine out.  In Device Manager, I selected "View" > "Devices by connection".  The 1 card with a yellow bang was then listed next to an audio device.  I disabled the audio device and IMMEDIATELY the 6th card started working.  Ahhhh thanks everyone for the help.

haggyy, try the registry patch above, then try viewing in Device Manager and disabling any devices in the same connection/folder, I'd like to hear that yours is working...

Hey thanks for the reply, i've tried this still the same issues.



That's great Russel!!!! So glad to hear that you got all 6 working now!! I will report back i'm going to try and find some time this evening to work on the rig.

That motherboard with 6x1070 word pretty much out of the box for me.  Start with 1 card.  Install windows.  Install drivers.  Add a gpu, boot computer, WAIT 5 MiNUTES even with an SSD!  I swear windows searches the entire globe for drivers even though you have them installed, so after 5 minutes check device manager and you should have 2 cards.

Repeat this process with each card.  Remember to WAIT after you boot, like an abnormal amount of time, sometimes after 5 minutes or so the new card will be recognized.  Each time you shut down and and a card and reboot it will take a bit to recognize it.

If something doesnt get recognized it is 99% your riser.  there is no QC on those risers and about 5-10% are just bad

Hey man thanks for the reply, interesting to know thank you, I will definitely give this a go and start from scratch if all else fails.

Also thank you smoolae for this interesting share will check this out ASAP !  :)

Right guys I've just tried editing the registry however I seem to be getting the same problem, just in case I am doing something completely wrong take a quick look at the screen shot and check that the value is correct.

After creating the registry key, I've rebooted and gone back into device manager, here are the results:

http://i66.tinypic.com/k1v0qt.png

Thanks

hi, haggyy.


IRQ is limited resources.
You can try allocate IRQ to video cards by disabling "some devices consuming IRQ".
update: Sorry. It seems to be a problem of memory mapping, not IRQ.

ty

haggy, your registry setting looks good.

Your screenshot shows an Intel Display Driver.  Go ahead and uninstall that driver and remove the driver files, it isn't helping anything (onboard graphics should be disabled in bios anyhow).

In device manager, did you try sorting "by connection"?  That will hopefully show you the conflicting device using the same IRQ or port/memory address.

I'm cheering for you to get up and running...

hi, guys.

By the way,

I examined the difference between G1840 and G3260.
They had the same number of PCIE lanes.

So, I guess that the problem of PCIE coming from the difference between these two CPUs can be ignored.

actually that's not true, and if you read he hasn't even had a chance to mod the bios, code 12 can relate to a few things, ram being one of them, the cpu being another I use a skylake intel Pentium dual 3ghz cpu, the kabylake cpus are crap if you browsed the forum youd find lots of posts of people having issues with the kabbylake cpus only handling 4 gpus while the skylake ones can handle the full 6+. the problem is people like you giving people like him shitty ass information

they both kabylake or skylake? it does make a difference, if it did not there wouldn't have been numerous people reporting problems with the kabylake cpus

no i3 is just waste of money, get yourself a intel Pentium cpu dual core one, and make sure its the skylake and not the kabylake cpu

hi, guys.

Both G1840 and G3260 belong to Haswell.

Also,
I looked into the details of HackFlags.

Unlike my expectation,
This option works to increase the limit register of "buffer size of enumerating PCI Express bridge devices"  from the initial value.

For example,
If you install an unusual number of video cards (like mining rig) with the default limit register value,
"PCIE device enumeration buffer" overflows due to too many PCIE devices,
Code12 exception seems to occur.

In addition to setting HackFlags,
disabling "some unnecessary PCIE devices" will reduce the occurrence of overflow.

Perhaps, Updating the "PCI Express root port" driver may also help you.
(ASRock H81 Pro BTC R2.0's "PCI Express root port" have more fit driver at WindowsUpdate named "intel(r) 8 series/c220 series pci express root port".)
Update: Sorry. Maybe this does not matter.

ty

Thanks so much for all the useful information guys, I am going to attempt to work on the rig sometime this evening or tomorrow and I will report back asap!

haggyy,

Unfortunately, we had a tomboy ..
Just now, I hope you will solve it well.

good luck.

update:

If disabling "some unnecessary PCIE device" does not solve the problem,
"PCIE device enumeration buffer" may not be the only problem.
and It may also be a problem of TOLUD.


1 or 2 is suspicious.

Even if you disable "some unnecessary PCIE devices",
It may not be enough with only 4GB RAM to drive several GPUs.

GREAT NEWS! I've got it working!! 3 GPU's running successfully, wow cant thank you all enough for the help.

Russel I literally done what you said, went onto device manager and sorted by connection and disabled the audio adaptor to the GPU that was flagged with the yellow exclamation mark and then BOOM it worked. I can finally relax... And now look to get 3 more GPU's added to the works.. Thanks again to everyone to helped  ;D ;D

Great to hear that! :)

Woo hoo! Great news haggyy! Glad it worked.

I'm getting the same problem with one of my 3 cards showing up under an audio controller, but even if I disable the audio controller, it still shows code 12
https://imgur.com/a/xKXcL

Yoledo,

It may not make a difference, but have you tried disabling "NVIDIA High Definition Audio" instead of disabling the controller?  (or maybe try to disable both, starting with the child device.

Other than that, you could try the other suggestions further up the thread and see if that helps..

Russell

This worked for my Asrock H81 PRO BTC R2.0 with 2x1060 and 1x1080ti cards! Thanks a lot!!

I have the same board and am having the same issues. I've tried everything
Enable /disable igpu in bios and device manager
Disable audio HD in device manager
HackFlags 600
The knowledge base article
Enabled/disabling rendering in bios

Why wont anything work....

Hi p1r473,

Your error could likely be due to a recent Windows update.  After windows rolled out the CPU bugfix a couple weeks ago, one of my GPUs stopped working again.  If you roll back this windows update, it may work.  (Uninstalling the windows update got me all 6 GPUs again).  Note you should be careful what you use the machine for if you remove the windows update patch.

Good luck,
Russell

hi, russellvb.

please refer this thread.
https://bitcointalk.org/index.php?topic=2401059.0;all

Windows Update on Jan/3/2018[KB4056892] might fix this problem!
You no longer need HackFlags nor Disable Audio Devices!

If you try install 1709 and patched up to [KB 4056892] and the rig successfully recognized the GPUs well without HackFlags,
I no longer recommend you to use 1703.

ty

Hi,
All my rigs with asrock H81 pro BTC R2 have the same issues with differents GPU since automatic fucking windows upgrade lol...
So will try this KB and let all of you know if it works well.
Thanks to all for the post. I hope it wall work :)

It works perfectly with this KB.
Thank you.
Will post the info on telegram group for my mining pool.
Happy mining