Run commandof docker CUDA sample docker image:ĭocker run -gpus all -env NVIDIA_DISABLE_REQUIRE=1 nvcr.io/nvidia/k8s/cuda-sample:nbody nbody -gpu -benchmark ~$ ls /dev/ autofs kmsg mapper ram11 ram8 sg3 tty13 tty23 tt圓3 tty43 tty53 tty63 vcsa block kvm mem ram12 ram9 shm tty14 tty24 tt圓4 tty44 tty54 tty7 vcsa1 bsg loop-control net ram13 random stderr tty15 tty25 tt圓5 tty45 tty55 tty8 vcsu btrfs-control loop0 null ram14 rtc0 stdin tty16 tty26 tt圓6 tty46 tty56 tty9 vcsu1 console loop1 nvram ram15 sda stdout tty17 tty27 tt圓7 tty47 tty57 ttyS0 vfio cpu_dma_latency loop2 ppp ram2 sdb tty tty18 tty28 tt圓8 tty48 tty58 ttyS1 vhost-net cuse loop3 ptmx ram3 sdc tty0 tty19 tty29 tt圓9 tty49 tty59 ttyS2 vsock dxg loop4 pts ram4 sdd tty1 tty2 tt圓 tty4 tty5 tty6 ttyS3 zero fd loop5 ram0 ram5 sg0 tty10 tty20 tt圓0 tty40 tty50 tty60 urandom full loop6 ram1 ram6 sg1 tty11 tty21 tt圓1 tty41 tty51 tty61 vcs fuse loop7 ram10 ram7 sg2 tty12 tty22 tt圓2 tty42 tty52 tty62 vcs1 Ls /dev/ on my WSL 2.0 host: (you can see dev/dxg is visible to the host and available to docker containers) | GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. | GPU GI CI PID Type Process name GPU Memory | | Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. | GPU Name TCC/WDDM | Bus-Id Disp.A | Volatile Uncorr. I'm starting the docker image with these run parameters:ĭocker run -d -gpus all -env NVIDIA_DISABLE_REQUIRE=1 -name='trex-miner' -e WALLET=(myethaddress) -e SERVER=stratum+ssl://:5555 -e WORKER=rig0 -e ALGO=ethash -p '4067:4067/tcp' ptrfrll/nv-docker-trex:cuda11 The version of t-rex I'm using is T-Rex NVIDIA GPU miner v0.19.11 - Ģ0210305 22:41:09 WATCHDOG: initialize error, memory dump: eyJhIjoiWWFRTlpnQW5HcnBYaW4yUDlETENReDdHT1hrZ1dKNm40TkhqR3NWUzloN0RWMFUxdW45R3M2UFdhcmF5Y2RFeVllVzZQWU1VUjZ3SGREbnFWdFRwdm5rYU9YZElLWEhPY1dMalZpam1QVmMzU2s1dmpFbTF5MjNOd1VwL05nY3Q4Z0dJZ3owaU81ZW82ZG0xWjkwYTZBPT0iLCJiIjoiVS9meWRFNEU4OCtqQUNzOXUzRVhhUT09In0K When the docker image spins up it refused initially to start up due to CUDA version discrepancies but passing the -env NVIDIA_DISABLE_REQUIRE=1 flag bypasses CUDA version check and the miner starts up ok but then crashes with a memory dump on initialisation. My host has latest NVIDIA CUDA drivers installed (v465.42) and the GPU is visible within my WSL 2.0 kernel (Ubuntu running on WSL2 kernel 5.4.91-microsoft-standard-WSL2) I used the tutorial available here to get this all working I am on Windows 10 dev build 21H1 21327.1000 and trying to mine on Ethermine using t-rex in a docker image using WSL 2.0 on Docker Desktop 3.10.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |