From 4573c855b7f8903321573470388c723aab41c8d8 Mon Sep 17 00:00:00 2001
From: LI SHENGYONG <49200266+shenchuxiaofugui@users.noreply.github.com>
Date: Fri, 21 Nov 2025 14:24:54 +0800
Subject: [PATCH] [Readme] EPLB Support Scenarios (#4314)

### What this PR does / why we need it?
Add information on the scope of EPLB support.
### Does this PR introduce _any_ user-facing change?
no
### How was this patch tested?


- vLLM version: v0.11.0
- vLLM main:
https://github.com/vllm-project/vllm/commit/2918c1b49c88c29783c86f78d2c4221cb9622379

---------

Signed-off-by: shenchuxiaofugui <1311027364@qq.com>
---
 vllm_ascend/ops/fused_moe/fused_moe.py | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/vllm_ascend/ops/fused_moe/fused_moe.py b/vllm_ascend/ops/fused_moe/fused_moe.py
index f82d341a..4eee35d0 100644
--- a/vllm_ascend/ops/fused_moe/fused_moe.py
+++ b/vllm_ascend/ops/fused_moe/fused_moe.py
@@ -256,6 +256,11 @@ class AscendFusedMoE(FusedMoE):
             self.moe_load = torch.zeros(local_num_experts,
                                         dtype=torch.int64).npu()
 
+        eplb_enable = self.dynamic_eplb or (self.expert_map_path is not None)
+        if eplb_enable and (not isinstance(self.quant_method,
+                                           AscendW8A8DynamicFusedMoEMethod)):
+            raise ValueError("Eplb supports only w8a8_dynamic quantization.")
+
         self.moe_config.num_experts = self.global_num_experts
         self.moe_config.num_local_experts = self.local_num_experts
         self.moe_config.original_num_experts = num_experts