Replies: 2 comments
-
|
Hi @ActLan! I'm Dosu and I’m helping the MinerU team. mineru-api 不支持在同一张昇腾910B上用同一个服务端口启动多实例,服务端口无法被多个进程共享,除非通过代理或负载均衡器做连接复用。多实例部署建议每个 mineru-api 实例使用不同端口,并通过环境变量(如 ASCEND_RT_VISIBLE_DEVICES、NPU_VISIBLE_DEVICES)分配独立设备资源,避免资源冲突。参考 性能和并发提升建议:
如果需要企业级异步高并发和多卡负载均衡,可参考 MinerU Tianshu 项目,支持异步任务队列、GPU自动调度和多实例并发。参考 如需具体参数建议或批量处理脚本,可补充说明实际任务规模和瓶颈环节。 To reply, just mention @dosu. Docs are dead. Just use Dosu. |
Beta Was this translation helpful? Give feedback.
0 replies
-
|
新版本已支持昇腾部署,可以看下这个方案https://opendatalab.github.io/MinerU/zh/usage/acceleration_cards/Ascend/ |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
环境:昇腾910B + miner2.6.2 + vllm-ascend 0.10.2 (quay.io/ascend/vllm-ascend:v0.10.2rc1)
环境变量:export MINERU_DEVICE_MODE=npu
export VLLM_USE_V1=1
export MINERU_ENGINE_MODE=vlm-vllm-async-engine
export ASCEND_RT_VISIBLE_DEVICES=7
export MINERU_MODEL_SOURCE=local
export NPU_VISIBLE_DEVICES=7
启动:nohup mineru-api --host 0.0.0.0 --port 8779 >output.log 2>&1 & 该如何提高性能以及增加并发能力,目前用fastapi的服务化接口,backend使用pipeline 我实测并不是异步的,vlm-vllm-async-engine解析也比较慢
Beta Was this translation helpful? Give feedback.
All reactions