LI SHENGYONG 0331f16a50 [EPLB] Reduce the memory used for heat aggregation (#6729) ...

### What this PR does / why we need it?
If dist.all_gather is used directly, 2 x HCCL_BUFFSIZE memory will be
consumed, but the actual memory required for hotspot aggregation is less
than 1 MB. Therefore, a separate small communication domain is created
for it.

### Does this PR introduce _any_ user-facing change?

### How was this patch tested?
Original：

![1](https://github.com/user-attachments/assets/8880b461-c26f-497c-9a05-2ca60cc46aa4)
Current：

![2](https://github.com/user-attachments/assets/c9da32b5-9200-4fa2-aff9-d8c4978ac602)


- vLLM version: v0.15.0
- vLLM main:
9562912cea

Signed-off-by: shenchuxiaofugui <1311027364@qq.com>

2026-02-24 18:02:24 +08:00

..

adaptor

[EPLB] Avoiding eplb's dependency on a specified model (#6528)

2026-02-10 15:58:44 +08:00

core

[EPLB][Bugfix] Bugfix for ineffective dynamic eplb (#6653)

2026-02-24 14:43:04 +08:00

__init__.py

Dynamic Expert Load Balance with Zero-like-overhead (#2956)

2025-09-17 10:36:43 +08:00

eplb_updator.py

[EPLB] Reduce the memory used for heat aggregation (#6729)

2026-02-24 18:02:24 +08:00

utils.py

[EPLB] Avoiding eplb's dependency on a specified model (#6528)

2026-02-10 15:58:44 +08:00