From 850295881041754184717804104fcaadd2b2129e Mon Sep 17 00:00:00 2001
From: liuzhenwei <zhenweiliu@habana.ai>
Date: Fri, 17 Jul 2026 18:54:34 +0800
Subject: [PATCH] [XPU] support HND layout (#47975)

Signed-off-by: zhenwei-intel <zhenwei.liu@intel.com>
---
 vllm/platforms/xpu.py | 8 --------
 1 file changed, 8 deletions(-)

diff --git a/vllm/platforms/xpu.py b/vllm/platforms/xpu.py
index cd57fb04bd2..80f99acc5ac 100644
--- a/vllm/platforms/xpu.py
+++ b/vllm/platforms/xpu.py
@@ -124,14 +124,6 @@ class XPUPlatform(Platform):
         attn_selector_config: "AttentionSelectorConfig",
         num_heads: int | None = None,
     ) -> str:
-        from vllm.v1.attention.backends.utils import set_kv_cache_layout
-
-        set_kv_cache_layout("NHD")
-        logger.info_once(
-            "Setting VLLM_KV_CACHE_LAYOUT to 'NHD' for XPU; "
-            "only NHD layout is supported by XPU attention kernels."
-        )
-
         # TurboQuant KV cache: route directly to TQ backend
         kv_cache_dtype = attn_selector_config.kv_cache_dtype
         if kv_cache_dtype is not None and kv_cache_dtype.startswith("turboquant_"):