A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 6, 09:30 UTC
[XPU] Automatically detect target platform as XPU in build. (#37634)
Signed-off-by: huanxing <huanxing.shen@intel.com>
Huanxing committed
Mar 20, 2026 at 13:30 UTC
6951fcd44fdd8a3a4d6b39b19b000604885a295f
1 file changed
+3
setup.py
+3
@@ -54,6 +54,9 @@ elif sys.platform.startswith("linux") and os.getenv("VLLM_TARGET_DEVICE") is Non
54
if torch.version.hip is not None:
55
VLLM_TARGET_DEVICE = "rocm"
56
logger.info("Auto-detected ROCm")
57
+ elif torch.version.xpu is not None:
58
+ VLLM_TARGET_DEVICE = "xpu"
59
+ logger.info("Auto-detected XPU")
60
elif torch.version.cuda is not None:
61
VLLM_TARGET_DEVICE = "cuda"
62
logger.info("Auto-detected CUDA")