[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>
This commit is contained in:
co-authored by
Claude Opus 4.8
parent
f35411ee81
commit
62aa85d9aa
@@ -6,7 +6,7 @@ configs (the block schedule and kernel config are chosen together).
|
||||
import pytest
|
||||
import torch
|
||||
|
||||
from sglang.srt.layers.moe.moe_runner.triton_utils.inkling_moe import (
|
||||
from sglang.kernels.ops.moe.inkling_moe import (
|
||||
SMALL_M_BLOCK_SIZE_M,
|
||||
compute_grouped_gemm_metadata,
|
||||
fused_moe_preprocess,
|
||||
|
||||
@@ -16,16 +16,16 @@ from sglang.kernels.ops.attention import (
|
||||
dsv4_attn_metadata_kernels as attn_metadata_kernels,
|
||||
)
|
||||
from sglang.kernels.ops.speculative import ragged_verify_kernels
|
||||
from sglang.srt.speculative.dspark_components.dspark_planner import (
|
||||
DSparkScheduleConfig,
|
||||
)
|
||||
from sglang.srt.speculative.dspark_components.kernels import (
|
||||
from sglang.kernels.ops.speculative.dspark import (
|
||||
dspark_accept,
|
||||
dspark_attn_metadata,
|
||||
dspark_draft_model,
|
||||
dspark_schedule,
|
||||
dspark_verify_window,
|
||||
)
|
||||
from sglang.srt.speculative.dspark_components.dspark_planner import (
|
||||
DSparkScheduleConfig,
|
||||
)
|
||||
from sglang.srt.speculative.ragged_verify import RaggedVerifyLayout
|
||||
from sglang.test.ci.ci_register import register_cuda_ci
|
||||
from sglang.test.test_utils import CustomTestCase
|
||||
|
||||
@@ -4,6 +4,9 @@ import unittest
|
||||
|
||||
import torch
|
||||
|
||||
from sglang.kernels.ops.speculative.dspark.dspark_schedule import (
|
||||
schedule_verify_lens_topk_from_survival,
|
||||
)
|
||||
from sglang.srt.speculative.dspark_components.dspark_planner import (
|
||||
DSparkScheduleConfig,
|
||||
HostConfidenceBudgetPlanner,
|
||||
@@ -15,9 +18,6 @@ from sglang.srt.speculative.dspark_components.dspark_sps import (
|
||||
SpsAdditiveCostTable,
|
||||
SpsCostTable,
|
||||
)
|
||||
from sglang.srt.speculative.dspark_components.kernels.dspark_schedule import (
|
||||
schedule_verify_lens_topk_from_survival,
|
||||
)
|
||||
from sglang.srt.speculative.ragged_verify import RaggedVerifyLayout
|
||||
from sglang.test.ci.ci_register import register_cpu_ci
|
||||
from sglang.test.test_utils import CustomTestCase
|
||||
|
||||
Reference in New Issue
Block a user