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 15, 16:00 UTC
[ROCm] Use device name in the warning (#15838)
Signed-off-by: Gregory Shtrasberg <Gregory.Shtrasberg@amd.com>
Gregory Shtrasberg committed
Apr 1, 2025 at 03:10 UTC
c7e63aa4d84de4f0b076d2974d30cd1cd34a4191
1 file changed
+1
-1
vllm/engine/arg_utils.py
+1
-1
@@ -1530,7 +1530,7 @@ class EngineArgs:
1530
# Non-CUDA is supported on V1, but off by default for now.
1531
not_cuda = not current_platform.is_cuda()
1532
if not_cuda and _warn_or_fallback( # noqa: SIM103
1533
- current_platform.device_type):
1533
+ current_platform.device_name):
1534
return False
1535
#############################################################
1536