VSCode
申请步骤
在线开发界面点击添加notebook
申请资源
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
- 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。
- 在Terminal中输入
nvidia-smi
或nvcc -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