Boning Zhang

See https://github.com/open-mmlab/mmcv/blob/4b38ffcf45a6a42e3cb34efd6fa38b9c8fa915b7/docs/zh_cn/get_started/api_reference.md

> could you try `gs.vulkan`? I tried. ```plaintext (device(type='cuda'), 'AMD Radeon Graphics', 7.464061737060547) [Genesis] [19:02:07] [INFO] ╭─────────────────────────────────────────────────────────────────────────────────────╮ [Genesis] [19:02:07] [INFO] │┈┉┈┉┈┉┈┉┈┉┈┉┈┉┈┉┈┉┈┉┈┉┈┉┈┉┈┉┈┉┈┉┈┉┈┉┈┉ Genesis ┈┉┈┉┈┉┈┉┈┉┈┉┈┉┈┉┈┉┈┉┈┉┈┉┈┉┈┉┈┉┈┉┈┉┈┉┈┉│ [Genesis] [19:02:07] [INFO] ╰─────────────────────────────────────────────────────────────────────────────────────╯ [Genesis] [19:02:07] [INFO]...

After reviewing [this issue](https://github.com/ROCm/ROCm/issues/2536), I was able to resolve the problem. Since I am using a `gfx1103` GPU (which I confirmed by running `rocminfo`), I found that setting the environment...

btw, I still get the following warnings when i work with `gs.gpu`: ```plaintext [W 12/22/24 19:25:19.692 11951] [cuda_driver.cpp:load_lib@36] libcuda.so lib not found. [W 12/22/24 19:25:19.692 11951] [misc.py:adaptive_arch_select@758] Arch=[] is not...