Watchdog: GPU # hangs in opencl call, exit (nVidia 1060 3GB OC)

stevej

member

Activity: 118

Merit: 10

I just fixed this today on my rig.
The last 4 days all of a sudden it started dropping GPU' s from mining.
Tried everything and then this morning put a new PSU in,
It worked straight away and no more opencl errors.
The old psu was 4 years old so I guess it was getting tired.
So the fix is get a new PSU

darkneorus

jr. member

Activity: 238

Merit: 3

maybe your PSU is tired and needs a vacation?

Skitch

newbie

Activity: 1

Merit: 0

I am experiencing the same problem. My rig was running stable for weeks before the Widows 10 Update crashed everything.

I have watchdog set. The rig will run for a few minutes, then crash because of a OpenCL call on a random GPU. It happens to a different GPU each time. The miner then restarts and the entire process happens all over again.

I tried going into the Widows performance settings and increasing my virtual memory from 16 to 32. This however did NOT fix the problem.

I am currently running (6) RX570's.

Does anyone have any recommendations on how to fix this?? Please??!

chasedg1

newbie

Activity: 27

Merit: 0

Quote from: chasedg1 on March 15, 2018, 09:09:35 AM

I'm experiencing the same issue on 2 different machines that have been running fine for close to a year... until about 2 weeks ago. I believe that it's a Windows 10 issue. For some reason Windows forced security updates even though I had them turned off. One machine will allow me to run 4 cards but not 5/6. I isolated different cards on different risers and it hangs no matter which 5 are operational so I know it's not riser/card. I uninstalled/reinstalled driver, reinstalled windows, reinstalled driver.. nothing. Has anyone come across this issue and been able to get their machine working?

I'm running RX 580s on Windows 10.

A third machine is now doing the same thing. Is anyone else experiencing this? Additionally windows doesn't allow you to go back to a previous build with the most recent update.

chasedg1

newbie

Activity: 27

Merit: 0

I'm experiencing the same issue on 2 different machines that have been running fine for close to a year... until about 2 weeks ago. I believe that it's a Windows 10 issue. For some reason Windows forced security updates even though I had them turned off. One machine will allow me to run 4 cards but not 5/6. I isolated different cards on different risers and it hangs no matter which 5 are operational so I know it's not riser/card. I uninstalled/reinstalled driver, reinstalled windows, reinstalled driver.. nothing. Has anyone come across this issue and been able to get their machine working?

I'm running RX 580s on Windows 10.

inverted

newbie

Activity: 19

Merit: 0

I'm stilling looking for the solution if anyone could help me?

inverted

newbie

Activity: 19

Merit: 0

Quote from: Xazax310 on January 26, 2018, 09:47:58 AM

That's usually an OC error additionally you may have a bad riser. Try downclocking all cards. I found my 1060 3GBs run at 400-600 memory(depending on brand) and +100 core. The Best way to test a riser is to put as much power as the cards can handle, mine a GPU intenstive coin like Zcash and see how long it takes before it crashes.

Method you wanna take for this is,

Plug 1 card in, run the miner, max power no OC. Run for at least one hour.
Plug 2nd card in, etc
and so on until you find one that fails, swap the riser. If they all pass then start OCing all of them. First try your settings then if they fail there back off until they no longer crash.

Mind you this is time intensive. Problem solving always is when you have an unknown "Watchdog" error like that.

Thanks. Could you let me know what do you mean by maximum power?

Xazax310

member

Activity: 246

Merit: 24

That's usually an OC error additionally you may have a bad riser. Try downclocking all cards. I found my 1060 3GBs run at 400-600 memory(depending on brand) and +100 core. The Best way to test a riser is to put as much power as the cards can handle, mine a GPU intenstive coin like Zcash and see how long it takes before it crashes.

Method you wanna take for this is,

Plug 1 card in, run the miner, max power no OC. Run for at least one hour.
Plug 2nd card in, etc
and so on until you find one that fails, swap the riser. If they all pass then start OCing all of them. First try your settings then if they fail there back off until they no longer crash.

Mind you this is time intensive. Problem solving always is when you have an unknown "Watchdog" error like that.

inverted

newbie

Activity: 19

Merit: 0

Quote from: adaseb on January 26, 2018, 04:43:49 AM

Could be anything

Either a bad riser, either your overclocked too much, undervolted too much, your PSU is crap, or it just crashed due to bad luck.

Just set the watchdog on and just let claymore reboot the software and it should work again.

What overcloak and undervolt settings do you recommend?

Also, How to do so: "Just set the watchdog on and just let claymore reboot the software and it should work again." ?

Thanks.

adaseb

legendary

Activity: 3808

Merit: 1723

Could be anything

Either a bad riser, either your overclocked too much, undervolted too much, your PSU is crap, or it just crashed due to bad luck.

Just set the watchdog on and just let claymore reboot the software and it should work again.

inverted

newbie

Activity: 19

Merit: 0

Quote from: clems on January 26, 2018, 04:34:39 AM

Quote from: inverted on January 26, 2018, 04:27:02 AM

Hello, I am getting Watchdog: GPU # in opencl call, exit error in my GPU mining rig.

My setup is:

GPU: 6 x 1060 3GB OC
Motherboard: Gigabyte GA-Z270P-D3
CPU: Intel G4400
RAM: 4 GB DDR4 Kingston
SSD: Kingston 120 GB (Sata)
PSU: Corsair HX850 + RM750X

I am posting the screenshots:

Claymore Error: https://i.imgur.com/p3yovy1.png

MSI Afterburner: https://i.imgur.com/G7wjz0B.png

Please help.

Thanks.

I guess you OC la little bit to much Memory maybe.

How much do you recommend?

By the way, I had this error once with all the GPU from 0-5 when it was all on stock (reset).

clems

member

Activity: 171

Merit: 10

Quote from: inverted on January 26, 2018, 04:27:02 AM

Hello, I am getting Watchdog: GPU # in opencl call, exit error in my GPU mining rig.

My setup is:

GPU: 6 x 1060 3GB OC
Motherboard: Gigabyte GA-Z270P-D3
CPU: Intel G4400
RAM: 4 GB DDR4 Kingston
SSD: Kingston 120 GB (Sata)
PSU: Corsair HX850 + RM750X

I am posting the screenshots:

Claymore Error: https://i.imgur.com/p3yovy1.png

MSI Afterburner: https://i.imgur.com/G7wjz0B.png

Please help.

Thanks.

I guess you OC la little bit to much Memory maybe.

inverted

newbie

Activity: 19

Merit: 0

Hello, I am getting Watchdog: GPU # in opencl call, exit error in my GPU mining rig.

My setup is:

GPU: 6 x 1060 3GB OC
Motherboard: Gigabyte GA-Z270P-D3
CPU: Intel G4400
RAM: 4 GB DDR4 Kingston
SSD: Kingston 120 GB (Sata)
PSU: Corsair HX850 + RM750X

I am posting the screenshots:

Claymore Error: https://i.imgur.com/p3yovy1.png

MSI Afterburner: https://i.imgur.com/G7wjz0B.png

Please help.

Thanks.

Topic: Watchdog: GPU # hangs in opencl call, exit (nVidia 1060 3GB OC) (Read 366 times)