kvcache-ai/Mooncake

Fork 0

mirror of https://github.com/kvcache-ai/Mooncake.git synced 2026-11-05 01:17:55 +00:00

[GH-ISSUE #1123] [RFC]: Support hugepage memory allocation via mmap to bypass RDMA memory registration limits #375

New issue

Closed

opened 2026-10-29 21:13:33 +00:00 by bsakaguchi · 6 comments

bsakaguchi commented

2026-10-29 21:13:33 +00:00

Owner

Originally created by @xiaguan on GitHub (Nov 27, 2025).
Original GitHub issue: https://github.com/kvcache-ai/Mooncake/issues/1123

Changes proposed

Many RDMA environments have memory registration limits, causing registration failures when users try to start instances with large memory configurations.

Add support for users to specify hugepage-based memory allocation using mmap, which can bypass RDMA memory registration limits.

Before submitting a new issue...

Make sure you already searched for relevant issues and read the documentation

Originally created by @xiaguan on GitHub (Nov 27, 2025). Original GitHub issue: https://github.com/kvcache-ai/Mooncake/issues/1123 ### Changes proposed Many RDMA environments have memory registration limits, causing registration failures when users try to start instances with large memory configurations. Add support for users to specify hugepage-based memory allocation using `mmap`, which can bypass RDMA memory registration limits. ### Before submitting a new issue... - [ ] Make sure you already searched for relevant issues and read the [documentation](https://kvcache-ai.github.io/Mooncake/)

bsakaguchi

2026-10-29 21:13:33 +00:00

closed this issue
added the
good first issue
label

bsakaguchi commented

2026-10-29 21:13:34 +00:00

Author

Owner

@xiaguan commented on GitHub (Nov 27, 2025):

Welcome to contribute! 🎉

@xiaguan commented on GitHub (Nov 27, 2025): Welcome to contribute! 🎉

bsakaguchi commented

2026-10-29 21:13:34 +00:00

Author

Owner

@stmatengss commented on GitHub (Nov 27, 2025):

Could you add this feature to #414 ? @xiaguan

@stmatengss commented on GitHub (Nov 27, 2025): Could you add this feature to #414 ? @xiaguan

bsakaguchi commented

2026-10-29 21:13:34 +00:00

Author

Owner

@xiaguan commented on GitHub (Nov 28, 2025):

Could you add this feature to #414 ? @xiaguan

Good, let me add this

@xiaguan commented on GitHub (Nov 28, 2025): > Could you add this feature to [#414](https://github.com/kvcache-ai/Mooncake/issues/414) ? [@xiaguan](https://github.com/xiaguan) Good, let me add this

bsakaguchi commented

2026-10-29 21:13:34 +00:00

Author

Owner

@dongb0 commented on GitHub (Feb 20, 2026):

It seems this feature has already been implemented in #1300. Should we close this issue, or is there anything still needs to be done? I'm happy to help.

@dongb0 commented on GitHub (Feb 20, 2026): It seems this feature has already been implemented in #1300. Should we close this issue, or is there anything still needs to be done? I'm happy to help.

bsakaguchi commented

2026-10-29 21:13:34 +00:00

Author

Owner

@stmatengss commented on GitHub (Feb 27, 2026):

It seems this feature has already been implemented in #1300. Should we close this issue, or is there anything still needs to be done? I'm happy to help.

True. It is done.

@stmatengss commented on GitHub (Feb 27, 2026): > It seems this feature has already been implemented in [#1300](https://github.com/kvcache-ai/Mooncake/pull/1300). Should we close this issue, or is there anything still needs to be done? I'm happy to help. True. It is done.

bsakaguchi commented

2026-10-29 21:13:34 +00:00

Author

Owner

@stmatengss commented on GitHub (Feb 27, 2026):

It seems this feature has already been implemented in #1300. Should we close this issue, or is there anything still needs to be done? I'm happy to help.

Maybe you can try to optimize the memory management.

@stmatengss commented on GitHub (Feb 27, 2026): > It seems this feature has already been implemented in [#1300](https://github.com/kvcache-ai/Mooncake/pull/1300). Should we close this issue, or is there anything still needs to be done? I'm happy to help. Maybe you can try to optimize the memory management.