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

python/sglang/srt/layers/moe/ep_moe/layer.py

@@ -54,14 +54,11 @@ _is_npu = is_npu()
 _is_fp8_fnuz = is_fp8_fnuz()
 _use_aiter = get_bool_env_var("SGLANG_USE_AITER") and _is_hip
 
-if not _is_npu:
+if not (_is_npu or _is_hip):
     from sgl_kernel import silu_and_mul
 
     from sglang.srt.layers.moe.cutlass_w4a8_moe import cutlass_w4a8_moe
 
-if _is_hip:
-    from vllm._custom_ops import scaled_fp8_quant
-
 if _use_aiter:
     from aiter import ActivationType, QuantType
     from aiter.fused_moe import fused_moe