From e9ae3f80779711eef52fa40741ed15cc025fb784 Mon Sep 17 00:00:00 2001
From: Kunshang Ji <kunshang.ji@intel.com>
Date: Wed, 25 Mar 2026 18:14:29 +0800
Subject: [PATCH] [Hardware][XPU] Align memory usage with cuda on xpu (#37029)

Signed-off-by: Kunshang Ji <jikunshang95@gmail.com>
Signed-off-by: Kunshang Ji <kunshang.ji@intel.com>
---
 vllm/platforms/xpu.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/vllm/platforms/xpu.py b/vllm/platforms/xpu.py
index 3bb76425e..3d2c2cca1 100644
--- a/vllm/platforms/xpu.py
+++ b/vllm/platforms/xpu.py
@@ -236,6 +236,7 @@ class XPUPlatform(Platform):
     def get_current_memory_usage(
         cls, device: torch.types.Device | None = None
     ) -> float:
+        torch.xpu.empty_cache()
         torch.xpu.reset_peak_memory_stats(device)
         return torch.xpu.max_memory_allocated(device)