mirror of
https://github.com/kvcache-ai/Mooncake.git
synced 2026-11-05 01:17:55 +00:00
[GH-ISSUE #1254] [Usage]: how to register gpu address with transfer engine #421
Labels
No labels
ascend
auto-closed
bug
bug
CI/Build
Common
documentation
enhancement
feature request
good first issue
help wanted
high priority
MetaX
nightly-failure
pull-request
PyTorch Backend
question
RFC
Roadmap
stale
Store
Tests
Transfer Engine
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference
kvcache-ai/Mooncake#421
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Originally created by @khuonglm on GitHub (Dec 22, 2025).
Original GitHub issue: https://github.com/kvcache-ai/Mooncake/issues/1254
Describe your usage question
Hello. I read that with transfer engine, we could register cuda memory and transfer zero copy between gpu and gpu. I have tried this, but it does not work.
How to resolve this?
mooncake version: 0.3.7.post2
Before submitting a new issue...
@ykwd commented on GitHub (Dec 22, 2025):
It looks like your code is correct. You might want to check whether the issue is caused by other factors. For example, on some systems, accessing RDMA requires root privileges.
@khuonglm commented on GitHub (Dec 22, 2025):
Thank you for your comment. Could you help me to check the details? I think I have root privileges when accessing rdma.
I run a simple code. It seems everything is in good conditions.
This is the rdma information.
@staryxchen commented on GitHub (Dec 22, 2025):
Is the nvidia_peermem driver installed?
@khuonglm commented on GitHub (Dec 22, 2025):
Thank you. it is not installed. I will try to install first and reopen if it is not resolved.