-
|
使用单机8卡的机器，指定"cuda_ep_cfg.gpu_id": 为0,1,2,3,4,5,6,7，最后还是只会只用0号gpu推理，请问多卡的正确用法是什么呀 # torch gpu
input_args = RapidTableInput(
model_type=ModelType.UNITABLE,
engine_cfg={"use_cuda": True, "cuda_ep_cfg.gpu_id": 1},
) |
Beta Was this translation helpful? Give feedback.
Replies: 2 comments 5 replies
-
|
当前代码仅支持单卡，不支持多卡哈 |
Beta Was this translation helpful? Give feedback.
-
|
那可以指定gpu推理吗 |
Beta Was this translation helpful? Give feedback.
-
|
可以。参见这个：link |
Beta Was this translation helpful? Give feedback.
-
|
是这里吗，我把"cuda_ep_cfg.gpu_id"换成2还是在用第一个gpu推理 # torch gpu
# input_args = RapidTableInput(
# model_type=ModelType.UNITABLE,
# engine_cfg={"use_cuda": True, "cuda_ep_cfg.gpu_id": 1},
# ) |
Beta Was this translation helpful? Give feedback.
-
|
这估计就是bug了，待我下一个版本修复一下吧
…---- 回复的原邮件 ----
| 发件人 | ***@***.***> |
| 发送日期 | 2025年08月18日 17:18 |
| 收件人 | RapidAI/RapidTable ***@***.***> |
| 抄送人 | SWHL ***@***.***>,
Comment ***@***.***> |
| 主题 | Re: [RapidAI/RapidTable] 单机多卡指定gpu推理 (Discussion #133) |
是这里吗，我把"cuda_ep_cfg.gpu_id"换成2还是在用第一个gpu推理
torch gpu
input_args = RapidTableInput(
model_type=ModelType.UNITABLE,
engine_cfg={"use_cuda": True, "cuda_ep_cfg.gpu_id": 1},
)
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you commented.Message ID: ***@***.***>
|
Beta Was this translation helpful? Give feedback.
-
|
刚才调试了一下。正确用法是： # torch gpu
input_args = RapidTableInput(
model_type=ModelType.UNITABLE,
engine_cfg={"use_cuda": True, "gpu_id": 1}, # ⚠️注意：这里只有gpu_id
) |
Beta Was this translation helpful? Give feedback.
-
|
这样可以，感谢 |
Beta Was this translation helpful? Give feedback.
刚才调试了一下。正确用法是：