[Kernel] Enable Kimi-K3 SiTU on the CuteDSL MoE backend and the SM107 low-latency GEMM plan

vllm/model_executor/layers/fused_moe/experts/flashinfer_cutedsl_moe.py

@@ -62,6 +62,8 @@ def __init__(
         self.gemm1_alpha = quant_config.gemm1_alpha
         self.gemm1_beta = quant_config.gemm1_beta
         self.gemm1_clamp_limit = quant_config.gemm1_clamp_limit
+        self.situ_beta = moe_config.activation_situ_beta
+        self.situ_linear_beta = moe_config.activation_situ_linear_beta
 
     def process_weights_after_loading(self, layer: torch.nn.Module) -> None:
         layer.w13_weight_scale_2.data.mul_(layer.w13_input_scale)
@@ -101,6 +103,7 @@ def _supports_activation(activation: MoEActivation) -> bool:
             MoEActivation.SWIGLUOAI,
             MoEActivation.SWIGLUOAI_UNINTERLEAVE,
             MoEActivation.RELU2_NO_MUL,
+            MoEActivation.SITU,
         )
 
     @staticmethod
@@ -171,6 +174,19 @@ def apply(
                 "swiglu_beta": self.gemm1_beta,
                 "swiglu_limit": self.gemm1_clamp_limit,
             }
+        elif activation == MoEActivation.SITU:
+            # The cute_dsl kernel keys SiTU on situ_beta and requires
+            # activation_type to stay a base type (ActivationType.Situ is
+            # rejected by normalize_cute_dsl_moe_activation_type), so the
+            # Swiglu base type is passed below and SiTU rides the betas.
+            if self.situ_beta is None:
+                raise ValueError(
+                    "SITU activation requires moe_config.activation_situ_beta"
+                )
+            swiglu_params = {
+                "situ_beta": self.situ_beta,
+                "situ_linear_beta": self.situ_linear_beta,
+            }
         swiglu_kwargs = {k: v for k, v in swiglu_params.items() if v is not None}
 
         flashinfer_cute_dsl_fused_moe_nvfp4(
@@ -190,6 +206,8 @@ def apply(
             num_local_experts=self.local_num_experts,
             local_expert_offset=self.local_expert_offset,
             moe_output=output,
-            activation_type=activation_to_flashinfer_int(activation),
+            activation_type=activation_to_flashinfer_int(
+                MoEActivation.SILU if activation == MoEActivation.SITU else activation
+            ),
             **swiglu_kwargs,
         )

vllm/models/kimi_k3/nvidia/low_latency_gemm.py

@@ -1,6 +1,6 @@
 # SPDX-License-Identifier: Apache-2.0
 # SPDX-FileCopyrightText: Copyright contributors to the vLLM project
-"""Kimi-K3 decode GEMM selection for unquantized BF16 on SM90/SM100/SM103.
+"""Kimi-K3 decode GEMM selection for unquantized BF16 on SM90/SM100/SM103/SM107.
 
 Dispatch is purely by local ``(N, K)`` shape and token count ``M`` — the module
 name plays no role. Each measured shape maps to a :class:`ProjectionSpec`
@@ -12,7 +12,10 @@
 :data:`KIMI_K3_PROJECTIONS` was tuned on B300 (SM103),
 :data:`KIMI_K3_PROJECTIONS_SM100` on B200 (SM100), and
 :data:`KIMI_K3_PROJECTIONS_SM90` on H200 (SM90). The per-(shape, M) winners
-genuinely differ between the parts, so the tables must not be merged.
+genuinely differ between the parts, so the tables must not be merged. SM107
+(Rubin) reuses the SM103 table: the plan was validated end-to-end on SM107
+hardware, but the per-M crossovers have not been re-measured there and may
+deserve their own table once retuned.
 """
 
 from __future__ import annotations
@@ -626,9 +629,14 @@ def _is_sm103() -> bool:
     return current_platform.is_device_capability((10, 3))
 
 
+def _is_sm107() -> bool:
+    return current_platform.is_device_capability((10, 7))
+
+
 def _low_latency_table() -> dict[tuple[int, int], ProjectionSpec] | None:
     """Measured dispatch table for the current device, or None if unsupported."""
-    if _is_sm103():
+    if _is_sm103() or _is_sm107():
+        # SM107 reuses the SM103 table; see the module docstring.
         return KIMI_K3_PROJECTIONS
     if current_platform.is_device_capability((10, 0)):
         return KIMI_K3_PROJECTIONS_SM100