kvcache-ai / kvcache-ai/Mooncake

[Bug]: [ascendDirect] run mooncake-client standalone,--global_segment_size cannot over 32GiB,why?

Open
#3,791 2 comments 0 reactions 1 assignee Claimed by @ascend-direct-dev View on GitHub
bug
Dominant language
C++
Stars
6.6k
Forks
1.2k
Avg merge
3d 5h
Merged PRs (30d)
312

Description

### Bug Report

**command**
mooncake-client --master=xxx --metadata_server="P2PHANDSHAKE" --host=xxx --global_segment_size="100GB" --port=xxx --protocol=ascend --thread=10

error:
Image

**running health when global_segment_size <= 32GB**

**environment**
export ASCEND_AUTO_CONNECT=1
export ASCEND_GLOBAL_RESOURCE_CONFIG='{"comm_resource_config.protocol_desc": ["roce:device"]}'
export HCCL_INTRA_ROCE_ENABLE=1

910B(A2),mooncake-master(v0.3.12.post1)、mooncake-client(v0.3.12.post1)

**reference docs**
https://gitcode.com/cann/hixl/wiki/Mooncake%20KVPool%E6%8C%87%E5%8D%97.md

### Before submitting...

- [ ] Ensure you searched for relevant issues and read the [documentation]

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.