Skip to main content

VSCode

申请步骤

  1. 在线开发界面点击添加notebook

  2. 申请资源

  3. 当状态为Running时代表VSCode已创建成功,可以进入VSCode进行开发和训练

使用技巧

  1. 该开发环境会自动挂载/mnt/$(username)和/archives/$(username)目录,可以将/mnt/$(username)当作个人workspace使用,把需要持久化的内容放入/archive/$(username)目录保存。
  2. 在Terminal中输入 nvidia-sminvcc -V查看GPU及驱动。
    root@test-notebook:/mnt/admin# nvidia-smi
    Mon Oct 10 10:15:50 2022
    +-----------------------------------------------------------------------------+
    | NVIDIA-SMI 460.91.03 Driver Version: 460.91.03 CUDA Version: 11.2 |
    |-------------------------------+----------------------+----------------------+
    | GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
    | Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
    | | | MIG M. |
    |===============================+======================+======================|
    | 0 Tesla T4 On | 00000000:00:07.0 Off | 0 |
    | N/A 25C P8 9W / 70W | 0MiB / 15109MiB | 0% Default |
    | | | N/A |
    +-------------------------------+----------------------+----------------------+
    +-----------------------------------------------------------------------------+
    | Processes: |
    | GPU GI CI PID Type Process name GPU Memory |
    | ID ID Usage |
    |=============================================================================|
    | No running processes found |
    +-----------------------------------------------------------------------------+
    root@test-notebook:/mnt/admin# nvcc -V
    nvcc: NVIDIA (R) Cuda compiler driver
    Copyright (c) 2005-2019 NVIDIA Corporation
    Built on Sun_Jul_28_19:07:16_PDT_2019
    Cuda compilation tools, release 10.1, V10.1.243