[Kernel] Migrate scattered MoE kernels to sglang.kernels (RFC #29630, Phase 2.5, 2/7) (#30786)

Co-authored-by: Claude Fable 5 <noreply@anthropic.com>

test/registered/jit/benchmark/bench_post_reorder_deepgemm.py

@@ -1,7 +1,7 @@
 import torch
 
 from sglang.jit_kernel.benchmark import marker
-from sglang.srt.layers.moe.ep_moe.kernels import (
+from sglang.kernels.ops.moe.ep_moe_kernels import (
     post_reorder_deepgemm,
     post_reorder_triton_kernel,
 )

test/registered/jit/minimax/test_minimax_quant_scatter.py

@@ -8,7 +8,7 @@ from sglang.jit_kernel.minimax_quant_ue8m0 import (
     per_token_quant_fp8_ue8m0,
     per_token_quant_fp8_ue8m0_scatter,
 )
-from sglang.srt.layers.moe.ep_moe.kernels import fill_gateup_input_triton_kernel
+from sglang.kernels.ops.moe.ep_moe_kernels import fill_gateup_input_triton_kernel
 from sglang.test.ci.ci_register import register_cuda_ci
 
 register_cuda_ci(est_time=20, stage="base-b-kernel-unit", runner_config="4-gpu-b200")

test/registered/jit/test_post_reorder_deepgemm.py

@@ -1,7 +1,7 @@
 import pytest
 import torch
 
-from sglang.srt.layers.moe.ep_moe.kernels import (
+from sglang.kernels.ops.moe.ep_moe_kernels import (
     post_reorder_deepgemm,
     post_reorder_triton_kernel,
 )

test/registered/moe/test_fused_append_remap_per_rank_shared_slots.py

@@ -11,7 +11,7 @@ import unittest
 
 import torch
 
-from sglang.srt.layers.moe.moe_runner.triton_utils.fused_moe_triton_kernels import (
+from sglang.kernels.ops.moe.fused_moe_triton_kernels import (
     fused_append_remap_shared_experts_deepep,
     fused_append_shared_experts,
 )

test/registered/moe/test_zero_experts.py

@@ -2,7 +2,7 @@ import unittest
 
 import torch
 
-from sglang.srt.layers.moe.ep_moe.kernels import zero_experts_compute_triton
+from sglang.kernels.ops.moe.ep_moe_kernels import zero_experts_compute_triton
 from sglang.test.ci.ci_register import register_cuda_ci
 from sglang.test.test_utils import CustomTestCase