[Kernel] Sweep missed dedicated kernels into kernels.ops (moe/quant siblings + dspark) (RFC #29630) (#32160)

Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>

python/sglang/kernels/ops/moe/sigmoid_gate_topk_renorm.py

@@ -13,15 +13,15 @@ import triton
 import triton.language as tl
 
 from sglang.kernels.jit.utils import is_arch_support_pdl
-from sglang.kernels.ops.moe.inkling_gate_topk_renorm import (
-    inkling_gate_topk_renorm_v2,
-)
-from sglang.srt.environ import envs
-from sglang.srt.layers.moe.moe_runner.triton_utils.gate_topk import (
+from sglang.kernels.ops.moe.gate_topk import (
     fpval_to_key,
     indx_to_key,
     key_to_indx,
 )
+from sglang.kernels.ops.moe.inkling_gate_topk_renorm import (
+    inkling_gate_topk_renorm_v2,
+)
+from sglang.srt.environ import envs
 
 
 @triton.jit

python/sglang/kernels/ops/speculative/dspark/dspark_accept.py

@@ -7,6 +7,7 @@ import torch
 import triton
 import triton.language as tl
 
+from sglang.kernels.ops.speculative.dspark.dispatch import inputs_on_cuda
 from sglang.kernels.ops.speculative.reject_sampling import (
     chain_speculative_sampling_triton,
 )
@@ -16,7 +17,6 @@ from sglang.srt.speculative.dflash_utils import (
     build_dflash_verify_target_probs,
     compute_dflash_correct_drafts_and_bonus,
 )
-from sglang.srt.speculative.dspark_components.kernels.dispatch import inputs_on_cuda
 
 
 class AcceptSampling:

python/sglang/kernels/ops/speculative/dspark/dspark_attn_metadata.py

@@ -7,7 +7,7 @@ import torch
 import triton
 import triton.language as tl
 
-from sglang.srt.speculative.dspark_components.kernels.dispatch import inputs_on_cuda
+from sglang.kernels.ops.speculative.dspark.dispatch import inputs_on_cuda
 from sglang.srt.utils import ceil_align
 
 

python/sglang/kernels/ops/speculative/dspark/dspark_draft_model.py

@@ -8,7 +8,7 @@ import torch.nn.functional as F
 import triton
 import triton.language as tl
 
-from sglang.srt.speculative.dspark_components.kernels.dispatch import inputs_on_cuda
+from sglang.kernels.ops.speculative.dspark.dispatch import inputs_on_cuda
 
 _BLOCK_V = 1024
 _IDX_SENTINEL = tl.constexpr(2147483647)

python/sglang/kernels/ops/speculative/dspark/dspark_schedule.py

@@ -6,7 +6,7 @@ import torch
 import triton
 import triton.language as tl
 
-from sglang.srt.speculative.dspark_components.kernels.dispatch import (
+from sglang.kernels.ops.speculative.dspark.dispatch import (
     inputs_on_cuda,
 )
 

python/sglang/kernels/ops/speculative/dspark/dspark_verify_window.py

@@ -6,8 +6,8 @@ import triton
 import triton.language as tl
 
 from sglang.kernels.ops.speculative.cache_locs import assign_extend_cache_locs_func
+from sglang.kernels.ops.speculative.dspark.dispatch import inputs_on_cuda
 from sglang.srt.managers.schedule_batch import ScheduleBatch
-from sglang.srt.speculative.dspark_components.kernels.dispatch import inputs_on_cuda
 from sglang.srt.speculative.ragged_verify import RaggedVerifyLayout
 
 

python/sglang/srt/layers/attention/deepseek_v4_backend.py

@@ -33,6 +33,11 @@ from sglang.kernels.ops.attention.dsv4_attn_metadata_kernels import (
     BuildPageTablePositions,
     ExpandPrefillCausally,
 )
+from sglang.kernels.ops.speculative.dspark.dspark_attn_metadata import (
+    BuildBlockSeqLensCausal,
+    BuildDsparkSwaPageIndices,
+    ComputeDsparkWindowGather,
+)
 from sglang.srt.environ import envs
 from sglang.srt.layers.attention.base_attn_backend import AttentionBackend
 from sglang.srt.layers.attention.dsv4.compressor_v2 import (
@@ -54,11 +59,6 @@ from sglang.srt.layers.attention.dsv4.sparse_prefill_utils import (
 from sglang.srt.mem_cache.deepseek_v4_memory_pool import DeepSeekV4TokenToKVPool
 from sglang.srt.model_executor.forward_batch_info import ForwardBatch, ForwardMode
 from sglang.srt.runtime_context import get_parallel
-from sglang.srt.speculative.dspark_components.kernels.dspark_attn_metadata import (
-    BuildBlockSeqLensCausal,
-    BuildDsparkSwaPageIndices,
-    ComputeDsparkWindowGather,
-)
 from sglang.srt.speculative.eagle_utils import per_step_draft_out_cache_loc
 from sglang.srt.speculative.ragged_verify import (
     RaggedVerifyMode,

python/sglang/srt/mem_cache/memory_pool.py

@@ -3361,7 +3361,7 @@ class MHATokenToKVPoolMXFP8(MHATokenToKVPool):
             # payload and the interleaved UE8M0 scales.
             if not self.mxfp8_sf_interleaved or cache_k.dtype == self.store_dtype:
                 raise ValueError("MXFP8 KV cache requires K and V scale tensors.")
-            from sglang.srt.layers.quantization.mxfp8_quant import quant_store_kv_mxfp8
+            from sglang.kernels.ops.quantization.mxfp8_quant import quant_store_kv_mxfp8
 
             quant_store_kv_mxfp8(
                 cache_k,
@@ -3394,7 +3394,7 @@ class MHATokenToKVPoolMXFP8(MHATokenToKVPool):
         """Write per-token UE8M0 K/V scales — interleaved into the FA4
         BlockScaledBasicChunk layout for page_size==128, flat otherwise."""
         if self.mxfp8_sf_interleaved:
-            from sglang.srt.layers.quantization.mxfp8_interleave_sf import (
+            from sglang.kernels.ops.quantization.mxfp8_interleave_sf import (
                 store_sf_interleaved,
             )
 
@@ -3436,7 +3436,7 @@ class MHATokenToKVPoolMXFP8(MHATokenToKVPool):
         # scale rows must travel with their fp8 payload or dequant reads
         # mismatched exponents.
         if self.mxfp8_sf_interleaved:
-            from sglang.srt.layers.quantization.mxfp8_interleave_sf import (
+            from sglang.kernels.ops.quantization.mxfp8_interleave_sf import (
                 store_sf_interleaved,
             )
 

python/sglang/srt/models/deepseek_v4_dspark.py

@@ -9,6 +9,10 @@ import torch.nn.functional as F
 from torch import nn
 
 from sglang.kernels.ops.attention.dsv4 import fused_q_norm_rope, fused_rope_inplace
+from sglang.kernels.ops.speculative.dspark.dspark_draft_model import (
+    BuildStepLocal,
+    CommitKvProj,
+)
 from sglang.srt.configs.deepseek_v4 import DeepSeekV4Config
 from sglang.srt.environ import envs
 from sglang.srt.layers.layernorm import RMSNorm
@@ -40,10 +44,6 @@ from sglang.srt.runtime_context import get_parallel
 from sglang.srt.speculative.dspark_components.dspark_config import (
     parse_dspark_draft_config,
 )
-from sglang.srt.speculative.dspark_components.kernels.dspark_draft_model import (
-    BuildStepLocal,
-    CommitKvProj,
-)
 from sglang.srt.speculative.ragged_verify import (
     RaggedVerifyMode,
     read_ragged_verify_mode,

python/sglang/srt/models/inkling_common/attn.py

@@ -883,7 +883,7 @@ class InklingAttention(nn.Module):
             if prologue_q_descale is not None:
                 extra_attn_kwargs["q_descale"] = prologue_q_descale
             else:
-                from sglang.srt.layers.quantization.mxfp8_quant import to_mxfp8
+                from sglang.kernels.ops.quantization.mxfp8_quant import to_mxfp8
 
                 q_mxfp = to_mxfp8(q.view(num_tokens, self.num_tp_heads, self.head_dim))
                 q = q_mxfp.data.view(num_tokens, -1)

python/sglang/srt/models/inkling_common/dense_mlp.py

@@ -382,7 +382,7 @@ class InklingBatchDenseMLP(nn.Module, FusedMoELoadingMixin):
 
     def _swiglu(self, y_st2f: torch.Tensor, gammas_st: torch.Tensor) -> torch.Tensor:
         # Helion's kernel can produce NaNs for small shared-expert batches.
-        from sglang.srt.layers.moe.moe_runner.triton_utils.inkling_moe import (
+        from sglang.kernels.ops.moe.inkling_moe import (
             silu_and_mul_triton,
         )
 

python/sglang/srt/models/inkling_common/moe.py

@@ -10,20 +10,13 @@ from torch import nn
 from triton.language.extra import libdevice
 
 from sglang.kernels.jit.utils import is_arch_support_pdl
+from sglang.kernels.ops.moe.gate_topk import gate_topk
 from sglang.kernels.ops.moe.inkling_gate_topk_renorm import (
     ensure_gate_gemv_fused_scratch,
     inkling_gate_gemv,
     inkling_gate_gemv_fused,
 )
-from sglang.srt.configs.inkling import InklingModelConfig
-from sglang.srt.distributed import (
-    get_tensor_model_parallel_group,
-)
-from sglang.srt.environ import GateGemvMode, envs
-from sglang.srt.layers.moe import get_moe_runner_backend
-from sglang.srt.layers.moe.fused_moe_triton.layer import FusedMoE
-from sglang.srt.layers.moe.moe_runner.triton_utils.gate_topk import gate_topk
-from sglang.srt.layers.moe.moe_runner.triton_utils.inkling_moe import (
+from sglang.kernels.ops.moe.inkling_moe import (
     FUSED_PREPROCESS_WIN_TOKENS,
     compute_grouped_gemm_metadata,
     fused_moe_preprocess,
@@ -34,9 +27,16 @@ from sglang.srt.layers.moe.moe_runner.triton_utils.inkling_moe import (
     select_grouped_gemm_block_m,
     silu_and_mul_helion,
 )
-from sglang.srt.layers.moe.moe_runner.triton_utils.sigmoid_gate_topk_renorm import (
+from sglang.kernels.ops.moe.sigmoid_gate_topk_renorm import (
     sigmoid_gate_topk_renorm,
 )
+from sglang.srt.configs.inkling import InklingModelConfig
+from sglang.srt.distributed import (
+    get_tensor_model_parallel_group,
+)
+from sglang.srt.environ import GateGemvMode, envs
+from sglang.srt.layers.moe import get_moe_runner_backend
+from sglang.srt.layers.moe.fused_moe_triton.layer import FusedMoE
 from sglang.srt.layers.moe.topk import PackedTopKOutput, StandardTopKOutput
 from sglang.srt.layers.moe.utils import RoutingMethodType
 from sglang.srt.layers.quantization.base_config import QuantizationConfig

python/sglang/srt/speculative/dspark_components/dspark_draft.py

@@ -7,6 +7,9 @@ from typing import Optional
 import msgspec
 import torch
 
+from sglang.kernels.ops.speculative.dspark.dspark_draft_model import (
+    SampleStepTokens,
+)
 from sglang.srt.environ import envs
 from sglang.srt.managers.schedule_batch import ScheduleBatch
 from sglang.srt.model_executor.forward_batch_info import (
@@ -18,9 +21,6 @@ from sglang.srt.runtime_context import get_parallel
 from sglang.srt.speculative.dflash_info_v2 import DFlashDraftInputV2
 from sglang.srt.speculative.draft_worker_common import make_draft_input_v2
 from sglang.srt.speculative.dspark_components.dspark_planner import VerifyWindow
-from sglang.srt.speculative.dspark_components.kernels.dspark_draft_model import (
-    SampleStepTokens,
-)
 from sglang.srt.speculative.spec_info import (
     SpeculativeAlgorithm,
     spec_scale_global_num_tokens,

python/sglang/srt/speculative/dspark_components/dspark_kv_inject.py

@@ -3,10 +3,10 @@ from typing import Optional
 import torch
 
 from sglang.kernels.ops.speculative.cache_locs import assign_extend_cache_locs_func
-from sglang.srt.managers.schedule_batch import ScheduleBatch
-from sglang.srt.speculative.dspark_components.kernels.dspark_verify_window import (
+from sglang.kernels.ops.speculative.dspark.dspark_verify_window import (
     BuildCommitInjectLayout,
 )
+from sglang.srt.managers.schedule_batch import ScheduleBatch
 from sglang.srt.speculative.ragged_verify import RaggedVerifyLayout
 
 

python/sglang/srt/speculative/dspark_components/dspark_planner.py

@@ -7,6 +7,10 @@ import msgspec
 import torch
 
 from sglang.kernels.ops.speculative.cache_locs import assign_extend_cache_locs_func
+from sglang.kernels.ops.speculative.dspark.dspark_schedule import (
+    ScheduleVerifyLensTopk,
+    compute_sort_survival,
+)
 from sglang.srt.distributed import get_tp_group
 from sglang.srt.environ import envs
 from sglang.srt.layers.dp_attention import is_dp_attention_enabled
@@ -31,10 +35,6 @@ from sglang.srt.speculative.dspark_components.dspark_sps import (
 from sglang.srt.speculative.dspark_components.dspark_sts import (
     load_sts_calibration_from_path,
 )
-from sglang.srt.speculative.dspark_components.kernels.dspark_schedule import (
-    ScheduleVerifyLensTopk,
-    compute_sort_survival,
-)
 from sglang.srt.speculative.ragged_verify import (
     RaggedVerifyLayout,
     RaggedVerifyMode,

python/sglang/srt/speculative/dspark_components/dspark_verify.py

@@ -5,6 +5,23 @@ from typing import Optional
 import msgspec
 import torch
 
+from sglang.kernels.ops.speculative.dspark.dspark_accept import (
+    AcceptGreedy,
+    AcceptSampling,
+    FinalizeAcceptLens,
+    SelectMixedAccept,
+    SoftmaxTemp,
+    accept_greedy_triton,
+    finalize_accept_lens_triton,
+)
+from sglang.kernels.ops.speculative.dspark.dspark_verify_window import (
+    BuildCommitInjectLayout,
+    BuildOutTokens,
+    BuildRaggedVerifyWindow,
+    RaggedVerifyWindow,
+    ScatterCompactToStrided,
+    scatter_compact_to_strided_into,
+)
 from sglang.srt.layers.logits_processor import LogitsProcessorOutput
 from sglang.srt.managers.schedule_batch import ScheduleBatch
 from sglang.srt.model_executor.forward_batch_info import CaptureHiddenMode, ForwardMode
@@ -19,23 +36,6 @@ from sglang.srt.speculative.dspark_components.dspark_planner import (
     VerifyWindow,
     apply_logits_adjustments_strided,
 )
-from sglang.srt.speculative.dspark_components.kernels.dspark_accept import (
-    AcceptGreedy,
-    AcceptSampling,
-    FinalizeAcceptLens,
-    SelectMixedAccept,
-    SoftmaxTemp,
-    accept_greedy_triton,
-    finalize_accept_lens_triton,
-)
-from sglang.srt.speculative.dspark_components.kernels.dspark_verify_window import (
-    BuildCommitInjectLayout,
-    BuildOutTokens,
-    BuildRaggedVerifyWindow,
-    RaggedVerifyWindow,
-    ScatterCompactToStrided,
-    scatter_compact_to_strided_into,
-)
 from sglang.srt.speculative.ragged_verify import RaggedVerifyLayout