Topic: [ANN] TeamRedMiner v0.10.10 - Ironfish/Kaspa/ZIL/Kawpow/Etchash and More - page 120. (Read 211877 times)

Sometimes deactivating and reactivating in device manager will get a gpu going again without rebooting the machine

anyone have any tips on how to make a watchdog.sh script for linux?

Most users of the watchdog script have it setup to reboot their machine and have the miner setup to auto-start after reboot.
As previous posts have mentioned, you may need to perform additional steps in your watchdog script to make sure the system will reboot without getting stuck.

Then there is no point in Watchdog. Without it, at least, the miner continues to work.

If your GPU has crashed you will most likely need to reboot before it will function again.  Starting a miner with a GPU that is already crashed will usually lead to hangs in the driver, like the situation you're seeing.

Again the miner could not restart. Uptime: 1 days, 16:43:37

[2018-12-05 02:14:42] Stats Uptime: 1 days, 16:43:37
[2018-12-05 02:14:42] GPU 0 [51C, fan 50%] cnv8: 997.9 h/s, avg 986.7 h/s, pool 1.036kh/s a:584 r:0 hw:0
[2018-12-05 02:14:42] GPU 1 [62C, fan 56%] cnv8: 1.021kh/s, avg 1.021kh/s, pool 764.3 h/s a:432 r:0 hw:142
[2018-12-05 02:14:42] GPU 2 [63C, fan 99%] cnv8: 966.2 h/s, avg 966.2 h/s, pool 927.4 h/s a:522 r:0 hw:0
[2018-12-05 02:14:42] GPU 3 [ 0C, fan  0%] cnv8: 912.8 h/s, avg 912.5 h/s, pool 902.3 h/s a:508 r:0 hw:0
[2018-12-05 02:14:42] GPU 4 [62C, fan 66%] cnv8: 985.1 h/s, avg 991.5 h/s, pool 935.2 h/s a:526 r:0 hw:2
[2018-12-05 02:14:42] GPU 5 [64C, fan 99%] cnv8: 1.019kh/s, avg 1.018kh/s, pool 1.003kh/s a:566 r:0 hw:0
[2018-12-05 02:14:42] GPU 6 [64C, fan 67%] cnv8: 1.019kh/s, avg 1.019kh/s, pool 1.021kh/s a:578 r:0 hw:0
[2018-12-05 02:14:42] GPU 7 [66C, fan 70%] cnv8: 1.019kh/s, avg 1.020kh/s, pool 1.022kh/s a:576 r:0 hw:0
[2018-12-05 02:14:42] GPU 8 [64C, fan 67%] cnv8: 1.013kh/s, avg 1.014kh/s, pool 1.049kh/s a:594 r:0 hw:0
[2018-12-05 02:14:42] Total                cnv8: 8.953kh/s, avg 8.947kh/s, pool 8.660kh/s a:4886 r:0 hw:144
[2018-12-05 02:14:53] GPU 8: detected DEAD (10:00.0), will execute restart script watchdog.bat
[2018-12-05 02:14:53] Watchdog triggering miner shutdown after restart script execution.
[2018-12-05 02:14:53] Shutting down...
Watchdog restarting miner in 30 secs.

Bpeмя oжидaния 29[2018-12-05 02:14:54] GPU 1 CN thread 0 exiting.
[2018-12-05 02:14:54] GPU 4 CN thread 0 exiting.
[2018-12-05 02:14:54] GPU 5 CN thread 0 exiting.
[2018-12-05 02:14:54] GPU 2 CN thread 0 exiting.
[2018-12-05 02:14:54] GPU 4 CN thread 1 exiting.
[2018-12-05 02:14:54] GPU 3 CN thread 0 exiting.
[2018-12-05 02:14:54] GPU 6 CN thread 1 exiting.
[2018-12-05 02:14:54] GPU 7 CN thread 0 exiting.
[2018-12-05 02:14:54] GPU 0 CN thread 0 exiting.
[2018-12-05 02:14:54] GPU 1 CN thread 1 exiting.
[2018-12-05 02:14:54] GPU 5 CN thread 1 exiting.
[2018-12-05 02:14:54] GPU 2 CN thread 1 exiting.
[2018-12-05 02:14:54] GPU 3 CN thread 1 exiting.
               28[2018-12-05 02:14:55] GPU 7 CN thread 1 exiting.
[2018-12-05 02:14:55] GPU 0 CN thread 1 exiting.
[2018-12-05 02:14:55] GPU 6 CN thread 0 exiting.
               20[2018-12-05 02:15:04] GPU 8 thread 0 shutdown timed out.
[2018-12-05 02:15:04] GPU 8 thread 1 shutdown timed out.
               19[2018-12-05 02:15:04] Unclean shutdown.
                0
          Team Red Miner version 0.3.8
[2018-12-05 02:15:28] Auto-detected AMD OpenCL platform 0

My first rig at 18.2.1 is working properly.
This one is acting only weird. Will test with different driver probably later.

I need to check the sgminer api. It’s either that it (weird enough) wants that scale, or we have a bug, or the AMD ADL interface is returning weird shit for 18.2.1 .

1150mV is very high, stock settings are often crazy compared to optimal settings. Sounds like you’re being throttled, thermal or power limit, hence the worse performance.

Just something I noticed:

Running a RX570 4GB in combo with 3 RX550 2GB results in ~10% less reported hash on the RX570
Default core voltages of 1150mV for at least the RX570 can be of an issue and give clearly lower hash rates (~20%, or no hash at all) compared to other settings with like 1000-950 mV

More API woes:

Seems like the miner API sometimes report wrong magnitude for GPU voltage. Did not know we power GPUs with Kilovolts

http://i63.tinypic.com/161xrtu.png

Rig is Windows 10, drivers 18.2.1

Nice!

Hmm, this is actually an interesting point.  As is the miner can get hung if the output is blocked, like when you go into that editing mode in windows.  We'll change it in the next version so that the inability to print will not have an effect on mining.

Hmm, that can be possible 

I checked miner before work, so it is possible.

Txs Dragonmike, turning off quick edit mode ;-)

You might possibly have clicked into the miner with your mouse, leaving the cursor there and blocking it from working (known issue). If you press a key or click the cursor away mining resumes.

You can lock your window and prevent this from happening if you right-click on the top bar, then "properties" and uncheck "quick edit mode".

Same here.

Mining for 4 days and than in 7 AM just stop, when I came home about 6 PM just click to miner "h" or something, and miner start mining again 

[2018-12-03 07:15:45] Pool at02.supportxmr.com share accepted. (GPU0) (a:12435 r:0) (36 ms)
[2018-12-03 07:16:06] Stats Uptime: 4 days, 07:15:55
[2018-12-03 18:55:49] Pool at02.supportxmr.com received new job. (job_id: fys23FOZGVBTDnniqeH7zmyUc6fW)
[2018-12-03 18:55:49] Pool at02.supportxmr.com set difficulty to 204697



SupportXMR pool show no hashing for that whole time.

So, I think we have new bug on 0.3.8 

when a gpu hang, the best way is to insert in wacthdog restart file a line with devcon.exe to disable and enable the gpu. it's simple to make it yourself . it's also simple for dev to disable that specific gpu. this is also work with early trm versions , device manager _> disable hang gpu, enable it. trm it's shutdown properly. then u can restart it without unplug the rig.

Sorry, but I do not speak English and use a translator (I am from Russia). I did not understand what you wanted to explain.

You use the "-d #" arguments when starting the miner. But this way, you specify which GPUs to start.

It is possible to make it initialize with less gpus that you specified

For example

My rig have 10 gpus, sometimes one of them fails and it reboots with 9 only

If i specified 10, them the miner crashs and fail to start