ByteDance-Seed / ByteDance-Seed/Bagel
多GPU加载后,运行 Image Understanding, GPU不能满负荷运行。
- Dominant language
- Python
- Stars
- 6.2k
- Forks
- 545
- PR merge metrics
- No merged PRs in 30d
Description
使用本项目的代码, Image Understanding 一个3M的表格,8卡 H20。使用到了6张卡,每张卡的使用率在9-10%,速度并不快。
nvtop 如下:
3784743 root 1 Compute 9% 5372MiB 4% 0% 2224MiB
3784743 root 2 Compute 9% 5372MiB 4% 0% 2224MiB
3784743 root 3 Compute 9% 5372MiB 4% 0% 2224MiB
3784743 root 4 Compute 9% 5372MiB 4% 0% 2224MiB
3784743 root 5 Compute 10% 5370MiB 4% 0% 2224MiB
3784743 root 0 Compute 6% 4804MiB 3% 100% 2224MiB
是否正常?如何配置能够提升GPU使用率?
Contributor guide
No contributing guide indexed for this repository
Research direction
Start from the Image Understanding entry point and its multi-GPU configuration, then reproduce the reported 3M-table workload on eight H20 GPUs while monitoring utilization with nvtop. Done means determining whether the low utilization and six-GPU usage are expected, and identifying a documented configuration or a confirmed implementation issue.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning, performance
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100