multi gpu, multi processes support..

[ashunaveed]

please add for multiple gpu, multiple workers support through cli..

[myhloli]

Thanks for the feedback. From the current implementation, this capability is not completely missing, but the entry point is not a single mineru command with --workers/--gpus. The current multi-worker / multi-GPU orchestration is handled at the mineru-router layer.

mineru-router already supports launching multiple local workers with --local-gpus auto or --local-gpus 0,1,2, and it also supports repeating --upstream-url to aggregate multiple existing mineru-api services. The regular mineru CLI can then submit tasks to that unified entry via --api-url.

A practical setup is:
mineru-router --host 0.0.0.0 --port 8002 --local-gpus 0,1
then
mineru -p <input> -o <output> --api-url http://127.0.0.1:8002

If you want finer GPU control, you can also start multiple mineru-api instances with CUDA_VISIBLE_DEVICES=0/1/... and let the router aggregate them.

So if the goal is “use multiple GPUs / multiple workers in CLI workflows”, the router-based path is already available. If the goal is “add native multi-GPU / multi-worker flags directly to the single mineru command”, that part is still not exposed in a one-command form today.