安装sudo apt install -y stress-ng lm-sensors smartmontools nvme-cli fio memtester git build-essentialsudo sensors-detectsensorsCPU 半个小时stress-ng --cpu 0 --cpu-method matrixprod --verify --metrics-brief --timeout 30m | tee cpu_stress_30m.log内存一个小时stress-ng --vm 4 --vm-bytes 75% --vm-method all --verify --metrics-brief --timeout 1h | tee mem_stress_1h.loggpu半个小时git clone https://github.com/wilicc/gpu-burn.gitcd gpu-burnmake./gpu_burn 1800 | tee gpu_burn_30m.log如果是有其他算力服务器可以选择make COMPUTE12.0跑的时候看GPUwatch -n 1 nvidia-smiwatch -n 1 ‘date; echo “ CPU / BOARD ”; sensors; echo; echo “ GPU ”; nvidia-smi’跑完看logdmesg -T | grep -iE “error|fail|thermal|xid|nvrm|mce|hardware”