trexminer / T-Rex

T-Rex NVIDIA GPU miner with web control monitoring page
2.64k stars 439 forks source link

Dual mining ETH+ERGO CUDA_ERROR_ILLEGAL_ADDRESS #909

Open thales51064 opened 2 years ago

thales51064 commented 2 years ago

Hello, I would like a confirmation, is this still possible to dual mining with a 8GB? I've read the wiki on https://github.com/trexminer/T-Rex/wiki/LHR, ETH+ERGO should be fine with 8GB but since I keep crashing on this point I've some doubt, unless the following error is not related?

20211205 00:41:00 conn1: Using protocol: stratum1. 20211205 00:41:00 conn1: Authorizing... 20211205 00:41:00 conn1: Authorized successfully. 20211205 00:41:00 conn2: Extranonce is set to: 5acd 20211205 00:41:00 conn2: Authorizing... 20211205 00:41:00 conn2: Authorized successfully. 20211205 00:41:00 conn2: autolykos2 block: 634462, diff: 4.00 G 20211205 00:41:01 GPU #0: intensity 21.3 20211205 00:41:01 conn1: ethash epoch: 458, block: 13742715, diff: 4.29 G 20211205 00:41:05 GPU #0: generating DAG 4.58 GB for epoch 458 ... 20211205 00:41:10 conn1: ethash epoch: 458, block: 13742716, diff: 4.29 G 20211205 00:41:13 conn1: ethash epoch: 458, block: 13742717, diff: 4.29 G 20211205 00:41:19 GPU #0: DAG generated [crc: 8908df32, time: 14722 ms], memory left: 1.98 GB 20211205 00:41:19 TREX: Can't find nonce with device [ID=0, GPU #0], cuda exception: CUDA_ERROR_ILLEGAL_ADDRESS, try to reduce overclock to stabilize GPU state 20211205 00:41:19 WARN: Miner is going to shutdown... 20211205 00:41:19 Main loop finished. Cleaning up resources...

GPU is a 3060TI 8GB LHR.

Any information/tips/lead is welcome! Thanks in advance.

Edit: I disabled all OC but I'm still getting the same error.

thales51064 commented 2 years ago

It's working on trex 0.24.2 but not on the latest version available (0.24.7). I didn't try the other version tho.

mayronguevara commented 2 years ago

I'm facing the same issue with 3060ti and 3070ti in dual mining mode. I have applied your solution using t-rex 0.24.2. I don't know if T-REX team is able to validate this, seems the issue is with the DAG something changed. i can't be in the network , if not the version 0.24.2 does not work for dual mining, and crashed suddenly.