[AMD] Remove vllm's scaled_fp8_quant and moe_sum when SGLANG_USE_AITER=1 (#7484)

2025-07-21 17:33:19 -07:00
parent 69adc4f81c
commit e50109f2ed
8 changed files with 156 additions and 69 deletions
--- a/python/sglang/srt/layers/quantization/unquant.py
+++ b/python/sglang/srt/layers/quantization/unquant.py
@@ -37,7 +37,6 @@ _use_aiter = get_bool_env_var("SGLANG_USE_AITER") and _is_hip
 if _use_aiter:
    from aiter import ActivationType
    from aiter.fused_moe import fused_moe
-    from aiter.fused_moe_bf16_asm import ck_moe_2stages
    from aiter.ops.shuffle import shuffle_weight