adapt to dsv32 on dcu

2025-09-30 18:37:31 +08:00
parent 8f7453e3af
commit 852a49c5cc
159 changed files with 7211 additions and 7782 deletions
--- a/python/sglang/srt/layers/quantization/w4afp8.py
+++ b/python/sglang/srt/layers/quantization/w4afp8.py
@@ -19,6 +19,10 @@ from sglang.srt.layers.quantization.unquant import UnquantizedLinearMethod
 from sglang.srt.layers.quantization.utils import is_layer_skipped
 from sglang.srt.utils import is_npu, set_weight_attrs

+_is_npu = is_npu()
+if not _is_npu:
+    from sglang.srt.layers.moe.cutlass_w4a8_moe import cutlass_w4a8_moe
+
 if TYPE_CHECKING:
    from sglang.srt.layers.moe import MoeRunnerConfig
    from sglang.srt.layers.moe.ep_moe.layer import EPMoE