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 13, 14:30 UTC
[Bugfix] Update device name for H200 detection (#28349)
Signed-off-by: Robert Shaw <114415538+robertgshaw2-redhat@users.noreply.github.com>
Robert Shaw committed
Nov 8, 2025 at 14:01 UTC
26990d25dce6562fc3f5dd53f2bde370c944ee28
1 file changed
+2
-2
vllm/model_executor/layers/fused_moe/fused_moe.py
+2
-2
@@ -819,8 +819,8 @@ def get_config_file_name(
819
) -> str:
820
device_name = current_platform.get_device_name().replace(" ", "_")
821
# Set device_name to H200 if a device from the H200 family is detected
822
- if "H200" in device_name:
823
- device_name = "H200"
822
+ if "H200" in device_name.split("_"):
823
+ device_name = "NVIDIA_H200"
824
dtype_selector = "" if not dtype else f",dtype={dtype}"
825
block_shape_selector = (
826
"" if not block_shape or not all(block_shape) else f",block_shape={block_shape}"