please mention which GPU: "WATCHDOG: T-Rex has a problem with GPU, terminating..." #20
Comments
|
There should be a message saying which GPU is idle prior to that. Please upgrade to 0.11.0 and send me the full log if the issue occurs again |
|
That was the first error/warning message in the session. Regardless, I don't see a reason why you shouldn't display the GPU# in there (or don't you know it?). |
|
If there was no message with GPU# as you've just said, then yes, the miner doesn't know which GPU caused the problem. The behaviour you're describing is not expected and appears to be a bug that needs investigation. If you start the miner with |
|
example from mining BCD. There were no Xid errors reported by the driver in kern.log either. Seems odd that the miner can't identify which card, it's definitely not all the 9 cards as t-rex seems to suggest -- it's the only miner with this problem from all the miners that I have been using in the past 2+ years. |
|
We've been chasing an issue where the miner stops hashing after a dev fee session, and the log you provided indicates it might be the same issue. In this case however the watchdog correctly did its job and restarted the miner, but we would like to fix the root cause. Would you be willing to help us with the investigation? If so we'll prepare a build which will produce an extra debugging info, so if you could run it and then send us the log file, that would be much appreciated. How long does it usually take for the problem to show itself? Which CUDA version do you use? |
|
CUDA 10.0 Sadly, my time is very limited to help with testing and I don't mine with t-rex all the time either, but I can give it a try (I'm just not promising anything). |
|
Please try 0.12.0 when you have time, there is a chance that the error is fixed, although we are not 100% sure. |
|
I have the same issue as outlined above, is there any new information on how to resolve the error? |
|
I'm using v0.19.1 and have the same issue. The problems occurs right after miner starting. This is my output:
I tried to add different gpu parameters such an indexing or gpu indexes setting but it didn't help. How can I solve it? |
|
@OverchenkoDev it would be best if the miner showed which GPU, but it's still not doing that. If you are on Linux, then you can Nnote that the explanation of the Xid errors doesn't always help you debug which particular o/c you need to bring down (mem,gpu,pow) |
|
@aleqx That's what I see after this command:
Find it difficult to understand this Also I used bencmark mode and there was no problems. Output:
|
|
WARN: WATCHDOG: T-Rex has a problem with GPU, terminating... |
|
I found out if your time changes on your computer this error will be caused.. just an fyi.. |
You give the error:
But it doesn't say which GPU#, so I dont' know which one is the culpri tin order to reduce overclock for that particular GPU only.
Could you please include the GPU# in the error?
The text was updated successfully, but these errors were encountered: