AHHHHH Nicehash Legacy is making me pull my hair out!! - RESOLVED

[COLT601]

SCROLL DOWN FOR RESOLUTION POST

Nicehash legacy miner keeps crashing. Locks the entire system up forcing me to drive 20 minutes (one way) to manually restart it.

Specs:

XFX AMD Radeon RX 580 8 Gig cards x 7
Asus Z270-A motherboard
Dual EVGA 1000w power supplies
AMD Blockchain drivers
The GPU's are all running on stock settings, no over clocking or any modding. Figured I wanted to get the rig 100% stable before I tried any overclocking.

I had been running only four RX 580's on Nicehash and had it working pretty well (stable at least). Yesterday afternoon I put in new power supplies and added three additional RX 580's.  Now Nicehash is locking up/crashing every couple of hours. Thinking it might be a problem when changing algorithms I tried disabling them all except for equihash and it just locked up again!

The only thing I can find in the log file that might be an issue is:

[ComputeDeviceManager.Query] CudaDevicesDetection found no devices. CudaDevicesDetection returned: CUDA error in func 'CudaDetection::QueryDevices' at line 17 : CUDA driver version is insufficient for CUDA runtime version.


To complicate things further, most of the cards get about a 300 h/s rate but about 50% of the time GPU 6 will only have a 150 or 200 h/s. I've tried replacing the PCI-E riser with no difference.  When I run Claymore miner for intense coin, all 7 GPU's get about 620-650 h/s.

I'm not really sure what else I can do or try at the moment to fix the problem with Nicehash. Can anyone recommend a more stable but equally profitable mining program & pool I can point these GPU's to?  Tried of driving across down tonight so I just ran the ITNS Claymore miner but that will only make me about $7 bucks per 24 hours and seems like a waste of resources for  7 RX 580s. Luckily I get free power (why it's across town) but still.

[1715239242]

1715239242

[1715239242]

1715239242

[1715239242]

1715239242

[1715239242]

1715239242

[1715239242]

1715239242

[nc50lc]

Have you tried running the rig only with the 6 stable GPUs?
Could be the slower one, most of the time, problems are caused by factory defects.

-edit-

Free power? I see, you installed your rig in the basement of a library.

[COLT601]

Have you tried running the rig only with the 6 stable GPUs?
Could be the slower one, most of the time, problems are caused by factory defects.

I haven't, but all the GPU's are stable when I run Claymore Miner for ITNS coin... so I wouldn't think there is a problem with the GPU itself ?

[jillscarbrough]

With RX 580 8Gb will be better if you didn't use Nicehash Legacy, it will give you a little bit different Hashrate with other lower cards. Using Claymore would be better, you can mine Ethas or maybe Equihash coins with your RX 580 8Gb.

[jerrison]

how can one get to start the mining o Nicehash as i have often times been interested in mining but then the skills aren't easily gotten if not impacted upon by someone who knows better

[jillscarbrough]

how can one get to start the mining o Nicehash as i have often times been interested in mining but then the skills aren't easily gotten if not impacted upon by someone who knows better

almost most new miners using Nicehash because it's very easy to use, automatically switch algorithms and payment will be on Bitcoin. and I believe if they enough knowledge to switch manually coin. that's more profit and more safely.

[COLT601]

Have you tried running the rig only with the 6 stable GPUs?
Could be the slower one, most of the time, problems are caused by factory defects.

I haven't, but all the GPU's are stable when I run Claymore Miner for ITNS coin... so I wouldn't think there is a problem with the GPU itself ?

Well I take that back. The Claymore miner crashed too. UGH. Guess I will try removing GPU 6 and see what happens

[jillscarbrough]

First, which Windows version you are using?
Yes, you could try to disable one by one card on Claymore, and see which card causing your systems crash.

[COLT601]

I am running Windows 10 Pro.

I am pretty sure it is the #6. When I saw the machine this morning it was the only one with the fan still spinning (all the others were stopped) and the machine was completely locked up. On the monitor windows had a a prompt that said some time of error. It may have been opencl or something along those lines but i have plenty of virtual memory enabled. When I googled the error it got results indicating from others who've seen it that their power supplies weren't large enough. But I have dual 1000w power supplies so there should be plenty.

What are the chances I got a faulty GPU?

[leonix007]

I am running Windows 10 Pro.

I am pretty sure it is the #6. When I saw the machine this morning it was the only one with the fan still spinning (all the others were stopped) and the machine was completely locked up. On the monitor windows had a a prompt that said some time of error. It may have been opencl or something along those lines but i have plenty of virtual memory enabled. When I googled the error it got results indicating from others who've seen it that their power supplies weren't large enough. But I have dual 1000w power supplies so there should be plenty.

What are the chances I got a faulty GPU?

I guess not, you can possibly check this by exchanging into the working GPU, btw, are the 3 GPU's connected to the other PSU? if yes, try not to use the second PSU and install the 5th GPU aligned with the present 4gpu's. 1000w should handle at least 6 GPU, by then let us know the result.

[jillscarbrough]

I am running Windows 10 Pro.
I am pretty sure it is the #6. When I saw the machine this morning it was the only one with the fan still spinning (all the others were stopped) and the machine was completely locked up....

Try to check your Win 10 version, Open Run type Winver and hit enter. I suggest using Win 10 latest update 1709.

If you sure that the GPU 6 is causing your problem, try to disable it when mining, you can disable it using device manager or hit number of your GPU on Claymore if you are using Claymore.

Yes, you need to increase your Virtual Memory, at least 16 Gb.

2000 Watt PSU is enough for your 8 Cards, but maybe something wrong with powering scheme.

[COLT601]

I am running Windows 10 Pro.
I am pretty sure it is the #6. When I saw the machine this morning it was the only one with the fan still spinning (all the others were stopped) and the machine was completely locked up....

Try to check your Win 10 version, Open Run type Winver and hit enter. I suggest using Win 10 latest update 1709.

If you sure that the GPU 6 is causing your problem, try to disable it when mining, you can disable it using device manager or hit number of your GPU on Claymore if you are using Claymore.

Yes, you need to increase your Virtual Memory, at least 16 Gb.

2000 Watt PSU is enough for your 8 Cards, but maybe something wrong with powering scheme.

System info says my Windows 10 build is 16922 version 1709. I only have 7 GPU's and have 56 gigs of virtual memory available.  Just drove the 20 minutes to reboot the miner again since it crashed. I came into the room to find myself staring at the motherboard BIOS screen. There was a big WTF. Exited out of it and it rebooted back into the bios. Cut the power and it restarted into Windows 10.


What does [ComputeDeviceManager.Query] CudaDevicesDetection found no devices. CudaDevicesDetection returned: CUDA error in func 'CudaDetection::QueryDevices' at line 17 : CUDA driver version is insufficient for CUDA runtime version mean??

[COLT601]

Sitting here watching the rig mine for a few minutes and nicehash is restarting itself every 4-5 minutes at the current rate

GetSummary exception: No connection could be made because the target machine actively refused it 127.0.0.1:5000

[COLT601]

And with GPU 6 turned off nicehash crashed again 45 minutes after I left... UGH!! $4500 bucks in this machine and I've spent more on fuel going to restart it than the damn thing has made me.

It was running fine with 4 GPU's that were split between two 750w Corsair PSU's. They would only accommodate 2x 8 pin VGA plugs each so I replaced them with two EVGA 1000w and added 3 more GPU's.

So it looks as if the problem lies somewhere going from 4 GPU's to adding additional. 4G is turned on in the motherboard bios. I've swapped risers, I don't think they are the problem.

[leonix007]

And with GPU 6 turned off nicehash crashed again 45 minutes after I left... UGH!! $4500 bucks in this machine and I've spent more on fuel going to restart it than the damn thing has made me.

It was running fine with 4 GPU's that were split between two 750w Corsair PSU's. They would only accommodate 2x 8 pin VGA plugs each so I replaced them with two EVGA 1000w and added 3 more GPU's.

So it looks as if the problem lies somewhere going from 4 GPU's to adding additional. 4G is turned on in the motherboard bios. I've swapped risers, I don't think they are the problem.

have you tried my suggestion above? just use the single 1000w psu.

[dohfish]

No one advicing to use the latest drivers? I have had a ton of issues with the blockchain drivers, the new adrenaline drivers are night and day in terms of stability for me and my rigs.

I run 8x580's, OC and undervolt and each of my Z270-A motherboards, a single 1000w psu and they are rock solid.

Do yourself the favor, completely reinstall Windows 10 1709, DDU the drivers and install latest adrenaline driver, and then enjoy.

The chances of a faulty card on stock settings is very minimal, more likely to be a riser issue in that case.

[COLT601]

After much trouble shooting and wasting time I finally figured out the problem with my machine. It was a faulty SATA cable that was powering my SSD.

First clue: the miner crashing and the computer restarting to bios. I didn't put two and two together at the time though.

I tried to uninstall the blockchain drivers and reinstall the latest AMD drivers but that turned into a massive clusterfuck of corruption.

Went to format windows and kept getting a bunch of errors. Windows wouldn't reinstall claiming my harddrive was MSR format instead of GPT.  Went into diskpart and had a ton of issues getting diskpart to even recognize the drive. Sometimes it would, sometimes it wouldn't. After about 10 tries I finally got it converted to GPT and began installing windows. 1/2 way through there was an error saying the resources weren't available to install, blah blah.

Tried it again and Windows didn't recognize the SSD at all. Back into diskpart, wouldn't find it. Restarted and it popped up, tried to format back to GPT but got an error. By this time I was heavily suspecting a problem with the SSD, thinking it had gone bad. Tried install one more time and got an error. Gave up and was going to order a new SSD. One of my buddies asked if I had tried a different cable and I told him no, it was a new cable and I figured a bad cable would be very unlikely.

Then I remembered it only booting into BIOS after crashing until it was power cycled again. Hmmm... points to hard drive. Then I remembered when I changed to EVGA power supplies I put the SSD on it's own SATA cable, one I hadn't used before. Hmm, well WTF, might as well give it a shot.

Changed cable. Flawless Windows 10 installation. Loaded latest AMD driver, shut down and plugged in all 7 GPU's. Restarted expecting a bunch of problems as I've always faced before. Nope!! Everything came up perfectly... almost too good to be true.

Machine is now mining ITNS coin and has been stable going on 2 hours. Woohoo!