Question Multi-GPU windows 10 crashing blue screen at start up

[elvos]

Hello, I have tried a lot of stuff cant figure out any work around seeking help THANK YOU FOR ALL U DO.

so I have a crypto mining rig I have been mining with 4 GPU's for several weeks with no problems. I just added a 5th Graphics card to it today and when I try to boot into windows 10 the computer blue screens pretty quickly I will send pictures/ links with my hardware for info below.

1. I have tried switching out GPU's doesn't matter what order or type of cards or where the cards are placed on the PCI slots it works perfectly with any 4 of the GPU's but as soon as I boot with 5 GPU's plugged in and power connectors connected windows blue screens.

2. I have tried increased virtual memory settings in safe mode restart computer still crashes. but how much would u recommend I have it set at exactly.
3. SAFE MODE everything works fine in safe mode... in safe mode when I look at device management all 5 GPU's are there

4. I tried changing around power cables from the power supply still crashes.
looked at some bios settings tried changing stuff in PCI bios stuff still crashes
5. updated video amd drivers still crashes as soon as the 5th GPU is connected and windows boots
6. tried resetting overclocks and what not still crashes no matter what whenever I try to boot with 5 GPU's connected windows blue screens

Right now I'am just mining crypto with 4 GPU's 1 of the GPU's being the new one I bought just incase there is something wrong with it but everything is going great so far.

BLUE SCREEN ERROR image read out
-Stop code :SYSTEM THREAD EXCEPTION NOT HANDLED
-what failed : amdkmdag.sys

here are my parts link
Mining rig body

https://www.amazon.ca/VTRETU-Ethereum-Motherboard-Supply%EF%BC%88110-264V%EF%BC%89-V01/dp/B09NC3Y2QC/ref=sr_1_10?crid=1GPTOVLC08176&keywords=Crypto+mining+rig&qid=1683738454&sprefix=crypto+mining+ri%2Caps%2C100&sr=8-10

https://m.media-amazon.com/images/I/71rMq2ZToeL._AC_SL1500_.jpg

GPU's
1 RX6800 Sapphire Pulse 16gb
2 RX6800 sapphire Pulse 16gb
3 RX6800 sapphire Pulse 16gb
4 RX6800 Asrock phantom gaming 16gb
5 RX 6800 gigabyte 16gb just purchased today

 

[elvos]

I tried using amd clean utility and DDU to uninstall drivers then during the reinstall process at about 65% windows blue screen crash occurred

 

[elvos]

I just don’t understand why can’t I have a 5th graphics card connected running in the system

 

[ubuysa]

Just speaking purely personally, I think that crypto-mining is one of the most environmentally unfriendly things anyone can engage in. Who pays your electricity bill?

 

[elvos]

I switched back to

Just speaking purely personally, I think that crypto-mining is one of the most environmentally unfriendly things anyone can engage in. Who pays your electricity bill?

I pay my hydro bill I live in an area that has 2nd cheapest hydro in my country

 

[xxxbabyxxx]

- Have you tried changing which Gen Lanes your GPUs are running? (Try higher and lower)
- Above 4G Decoding on/off in BIOS? (I've had rigs not running stable above 2 unless it's ON, and some vis-versa)
- Checked Device Manager/Event Viewer for errors?
- Check your SSD lane connections, many MOBOS (Gigabyte/ASROCK specially) may disable specific lanes depending on where your plug your SSD. (This info would be in the manual)

Just so you know there are motherboards that have their limits in terms of GPUs it can run, even if there's PCI slots available, there wouldn't be enough resources on the bus so it ends up crashing them, or show bad behavior in device manager. (See if you get resources error codes for example)

I've got multiple MOBOs that can all support 4+, but some Mobos can only support up to 3, some 2, etc.

If you're mining, there's a lot of communities that make spreadsheets to know how many GPUs each mobo can handle, (I've got one with 50 Motherboards tested, but unfortunately yours isn't there yet.)

I'd also suggest checking your manual to see if running certain lanes would disable others, incase the lane for the 5th is disabled.

 

[elvos]

- Have you tried changing which Gen Lanes your GPUs are running? (Try higher and lower)
- Above 4G Decoding on/off in BIOS? (I've had rigs not running stable above 2 unless it's ON, and some vis-versa)
- Checked Device Manager/Event Viewer for errors?
- Check your SSD lane connections, many MOBOS (Gigabyte/ASROCK specially) may disable specific lanes depending on where your plug your SSD. (This info would be in the manual)

Just so you know there are motherboards that have their limits in terms of GPUs it can run, even if there's PCI slots available, there wouldn't be enough resources on the bus so it ends up crashing them, or show bad behavior in device manager. (See if you get resources error codes for example)

I've got multiple MOBOs that can all support 4+, but some Mobos can only support up to 3, some 2, etc.

If you're mining, there's a lot of communities that make spreadsheets to know how many GPUs each mobo can handle, (I've got one with 50 Motherboards tested, but unfortunately yours isn't there yet.)

I'd also suggest checking your manual to see if running certain lanes would disable others, incase the lane for the 5th is disabled.

Ya my 4g decoding is on I tried playing with that on/off makes difference, I also tried upping the bus it was set 32 I increased it to 64 but still crashes. Should I try 128 bus and higher I was not sure if it was totally safe to do

 

[elvos]

If the case is I can’t get any more cards on then I was going to look at making a second rig anyway

 

[xxxbabyxxx]

Ya my 4g decoding is on I tried playing with that on/off makes difference, I also tried upping the bus it was set 32 I increased it to 64 but still crashes. Should I try 128 bus and higher I was not sure if it was totally safe to do

From Experience it either works EASILY or doesn't at all (MOBO SIDE). It works usually simply by making sure the lanes are turned on and you don't have anything turning it off through manual, (example GPU limit, SSD/SATA lanes disabling some PCI etc) OR IT DOESNT.

I'd suggest doing what I said in my main post, and look online to see if anyone got the same BIOS & Mobo as you and is running up to 5 GPUs on it.

What about what I said above in regards to Event viewer? And also Device Manager => Graphics Adapter and see if there's any resources/Error Codes? is GPUZ recognizing every single card properly without any information missing?? GPUZ Sensors too?

- There's a list of things you can try before 100% deciding if you should go a new rig or not. Specially that you have a 2000W PSU that's running efficient cards, where you'd be much happier if it ran that extra RX 6800 instead of buying a new mobo, cpu, ram, ssd, and waste extra ~150W just to run the system to add an extra rx 6800.

 

[elvos]

Okay thank you so much for the suggestions and info I will do some research on my stuff and see if I can get it going. This weekend I figured my 2000w power supply should be good for more cards would like to get it to a least 5-6th card before investing in a whole new system… but it’s all a learning process I’am in to crypto mining for the long haul want to accrue a large holdings then hold into the future hopefully see a large up side in price’s

 

[xxxbabyxxx]

Okay thank you so much for the suggestions and info I will do some research on my stuff and see if I can get it going. This weekend I figured my 2000w power supply should be good for more cards would like to get it to a least 5-6th card before investing in a whole new system… but it’s all a learning process I’am in to crypto mining for the long haul want to accrue a large holdings then hold into the future hopefully see a large up side in price’s

You're welcome and good luck take your time researching one thing at a time! 🙂

Yep that was the issue for me early on, I've had a 1000W Platinum Cooler Master PSU, was only able to run 2x RTX 3070's on one board! ( I wanted to add GTX 1050 Ti and few more back then)

Worst case scenario if it doesn't work, you are still running the PSU without pressuring it, and you can comfortably overclock cards(as you're not worried about the PSU not supporting them, just make sure you do it efficiently). - OR you can always use this 2000w once again in a future BIGGER rig (as you're in it for the long haul as you said ^^), and replace this one with a smaller PSU 🙂

Always research & dont be afraid to ask and you'll learn a lot in short time! Have a nice day

 

[elvos]

I had messed around in the device management it showed

You're welcome and good luck take your time researching one thing at a time! 🙂

Yep that was the issue for me early on, I've had a 1000W Platinum Cooler Master PSU, was only able to run 2x RTX 3070's on one board! ( I wanted to add GTX 1050 Ti and few more back then)

Worst case scenario if it doesn't work, you are still running the PSU without pressuring it, and you can comfortably overclock cards(as you're not worried about the PSU not supporting them, just make sure you do it efficiently). - OR you can always use this 2000w once again in a future BIGGER rig (as you're in it for the long haul as you said ^^), and replace this one with a smaller PSU 🙂

Always research & dont be afraid to ask and you'll learn a lot in short time! Have a nice day

I tried more changes
- looking through all of my bios for settings and stuff to try but still keep blue screen crashing at desktop or even right on boot up same blue screen error

- changing bios settings on each PCI ports gen 1, gen 2 , auto settings tried changing still blue screens same error
- looked in device management devices look fine just an error on the 5th graphics card device about it not being able to start the device event error some details everything else was the same as other devices
- changed more bus speeds still nothing changed
- turned off all the pci ports except the ones iam using 1-5 PCI slot ports still nothing changed

I think I just have a bad motherboard that refuses to do 5 graphics cards. I think I will look into building a second mining rig and building that one out with more graphics cards real shame a waste with the 200w power supply

 

[elvos]

I had messed around in the device management it showed

I tried more changes
- looking through all of my bios for settings and stuff to try but still keep blue screen crashing at desktop or even right on boot up same blue screen error

- changing bios settings on each PCI ports gen 1, gen 2 , auto settings tried changing still blue screens same error
- looked in device management devices look fine just an error on the 5th graphics card device about it not being able to start the device event error some details everything else was the same as other devices
- changed more bus speeds still nothing changed
- turned off all the pci ports except the ones iam using 1-5 PCI slot ports still nothing changed

I think I just have a bad motherboard that refuses to do 5 graphics cards. I think I will look into building a second mining rig and building that one out with more graphics cards real shame a waste with the 200w power supply

O also looked into that GPU-z all my cards are exactly as they are no false detections on the actual hardware all RX6800 detected and correct stats

 

[elvos]

guess my other option would be to pull the power supply out of this rig if I can and use the 2000w in another rig that supports 5+ cards just get a lower watt power supply for this rig enough to power 4 graphics cards could try to rig out the case fans and use the parts as well in another rig build just keep the bare minimum to run the 4 graphics cards

 

[xxxbabyxxx]

guess my other option would be to pull the power supply out of this rig if I can and use the 2000w in another rig that supports 5+ cards just get a lower watt power supply for this rig enough to power 4 graphics cards could try to rig out the case fans and use the parts as well in another rig build just keep the bare minimum to run the 4 graphics cards

Exactly that's what I said above! 🙂 atleast you've got a board that can handle 4x GPU and that's great 😄

You already have an OVERKILL PSU on that rig, so you should feel less bad replacing the PSU and using it on another bigger rig, whilst getting a decent 1000W 1250W PSU on your current one ^^

 

[elvos]

Exactly that's what I said above! 🙂 atleast you've got a board that can handle 4x GPU and that's great 😄

You already have an OVERKILL PSU on that rig, so you should feel less bad replacing the PSU and using it on another bigger rig, whilst getting a decent 1000W 1250W PSU on your current one ^^

ya didn't actually do too bad for all that for 850CAD

iam shopping around and it costs alot more to buy parts

 

[elvos]

my watts meter I have running is saying I'am pulling 825-850watts from the wall mining if I dial up the fans to max all out maybe just under 1000w so if I get a 1200plus watts power supply should do the job. u think?

 

[elvos]

1300w power supply I might get to swap out the 2000w over kill

https://www.newegg.ca/evga-supernova-1300-g2-120-g2-1300-xr-1300w/p/N82E16817438011?Item=9SIA92GJV75276

 

[elvos]

You're welcome and good luck take your time researching one thing at a time! 🙂

Yep that was the issue for me early on, I've had a 1000W Platinum Cooler Master PSU, was only able to run 2x RTX 3070's on one board! ( I wanted to add GTX 1050 Ti and few more back then)

Worst case scenario if it doesn't work, you are still running the PSU without pressuring it, and you can comfortably overclock cards(as you're not worried about the PSU not supporting them, just make sure you do it efficiently). - OR you can always use this 2000w once again in a future BIGGER rig (as you're in it for the long haul as you said ^^), and replace this one with a smaller PSU 🙂

Always research & dont be afraid to ask and you'll learn a lot in short time! Have a nice day

Okay so interesting thing for the fun I was up all night I was watching a video for gaming benchmarks noticed the RX 6800 GPU is better then my RTX 2080ti for gaming performance FPS. I have currently in my gaming pc RTX 2080ti so just for fun I switched them out... 6800 in gaming pc and 2080ti as 5th mining rig GPU. BAM I'am mining again now with 5 GPU's slapped the RTX 2080ti in my mining rig installed Nvidia drivers no problem install after burner and started mining it detected the 2080ti and all 4 other RX 6800 now I'am mining with 5 GPU's all at once so now curious is it a AMD software issue with adrenaline software and clearly not windows or motherboard being the problem is does AMD software just not like more then 4 cards or something.

When I think about it now I was able to get the 5 6800's hocked up I tried to install AMD drivers and it would fail and crash right at about 65% driver installing completion. always the GPU driver count start error event notes seemed like

Now I want to try to put another nvidia card into my mining rig for a 6th card lol, would have thought mixing nvidia and amd gpu's would be the problem but guess not maybe a RTX 3060 maybe the way to go is for 8 GPU's ill run 4 3060's 4 RX 6800's together in the rig slowly add one card at a time see if the mining rig will be able to handle it for now I might test out some other GPU's i have laying around see if one or more might work also just test out the 4 RX 6800's and 1 RTX 2080ti see if it will continue to mine with out issues pulling about 1150watts from the wall right now

 

[xxxbabyxxx]

Okay so interesting thing for the fun I was up all night I was watching a video for gaming benchmarks noticed the RX 6800 GPU is better then my RTX 2080ti for gaming performance FPS. I have currently in my gaming pc RTX 2080ti so just for fun I switched them out... 6800 in gaming pc and 2080ti as 5th mining rig GPU. BAM I'am mining again now with 5 GPU's slapped the RTX 2080ti in my mining rig installed Nvidia drivers no problem install after burner and started mining it detected the 2080ti and all 4 other RX 6800 now I'am mining with 5 GPU's all at once so now curious is it a AMD software issue with adrenaline software and clearly not windows or motherboard being the problem is does AMD software just not like more then 4 cards or something.

When I think about it now I was able to get the 5 6800's hocked up I tried to install AMD drivers and it would fail and crash right at about 65% driver installing completion. always the GPU driver count start error event notes seemed like

Now I want to try to put another nvidia card into my mining rig for a 6th card lol, would have thought mixing nvidia and amd gpu's would be the problem but guess not maybe a RTX 3060 maybe the way to go is for 8 GPU's ill run 4 3060's 4 RX 6800's together in the rig slowly add one card at a time see if the mining rig will be able to handle it for now I might test out some other GPU's i have laying around see if one or more might work also just test out the 4 RX 6800's and 1 RTX 2080ti see if it will continue to mine with out issues pulling about 1150watts from the wall right now

That's awesome! ikr sometimes you could be doing everything correct and an error still shows up.

Sometimes it could be Windows not acting nice to too many GPUs at once (hence why some people just go to HiveOs), Sometimes it could also be simply the miner itself for example causing troubles, perhaps overclocks on a specific algorithm etc! glad you got it working!