[Bug] Fix Batched DeepGEMM (#47884)
Signed-off-by: Robert Shaw <robertgshaw2-redhat@dgx-b200-02.mgmt.accl-001.lab.rdu2.dc.redhat.com> Co-authored-by: Robert Shaw <robertgshaw2-redhat@dgx-b200-02.mgmt.accl-001.lab.rdu2.dc.redhat.com> Co-authored-by: Claude <noreply@anthropic.com>
Robert Shaw committed
Jul 8, 2026 at 09:05 UTC
572b25b03ea5893305d4b25a25150ef129717605
1 file changed
+1
-1
vllm/model_executor/layers/fused_moe/experts/batched_deep_gemm_moe.py
+1
-1
@@ -210,7 +210,7 @@ def persistent_masked_m_silu_mul_quant(
210
DeepGemmQuantScaleFMT.UE8M0,
211
]
212
213
- device_capability = current_platform.get_device_capability(device_id=y.device.index)
213
+ device_capability = current_platform.get_device_capability()
214
assert device_capability is not None
215
cuda_arch = device_capability.to_int()
216