repo stringclasses 454
values | file_path stringlengths 5 201 | extension stringclasses 1
value | content stringlengths 8 509k | num_lines int64 3 16.9k | size_bytes int64 8 511k |
|---|---|---|---|---|---|
sglang | test/registered/ops/test_aiter_allreduce_fusion_amd.py | .py | import csv
import os
import subprocess
import sys
import tempfile
import types
import unittest
from contextlib import ExitStack
from pathlib import Path
from unittest import mock
import torch
from sglang.srt.layers import communicator as comm
from sglang.srt.layers.communicator import LayerCommunicator, ScatterMode
f... | 484 | 17,398 |
sglang | test/registered/dp_engine/test_data_parallelism.py | .py | import time
import unittest
import requests
from sglang.srt.utils import kill_process_tree
from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
from sglang.test.kits.eval_accuracy_kit import GSM8KMixin
from sglang.test.test_utils import (
DEFAULT_MODEL_NAME_FOR_TEST,
DEFAULT_TIMEOUT_FOR_SE... | 72 | 2,074 |
sglang | test/registered/kernels/test_kernel_inventory.py | .py | """CPU-only structural checks for the unified kernel tree."""
from __future__ import annotations
import ast
import importlib.util
import sys
from pathlib import Path
import pytest
import sglang.kernels as kernels
from sglang.test.ci.ci_register import register_cpu_ci
register_cpu_ci(est_time=5, suite="base-a-test-... | 238 | 9,222 |
sglang | test/registered/kernels/test_kda_mtp_cutedsl_replayssm_ring.py | .py | """Parity: CuTe MTP verify kernel's ReplaySSM ring == its own state snapshots.
`fused_kda_decode_mtp_dspark(replayssm_*=...)` switches the CuTe DSpARK verify
kernel to CACHE_RING mode: per draft step it stores post-conv pre-l2norm k,
post-conv v, the log-decay gate gk, and sigmoid(beta) into the per-slot rings
(and sk... | 195 | 7,430 |
sglang | test/registered/kernels/test_fused_op_dispatch.py | .py | """Dispatch-contract tests for the unified ``BaseFusedOp`` (RFC #29630, #26426).
``BaseFusedOp`` replaced ``MultiPlatformOp`` as the single operator
abstraction; these tests pin down the parts of that contract that a refactor
could silently break:
- the priority ladder: explicit ``backend=`` > global forced backend >... | 559 | 18,817 |
sglang | test/registered/kernels/test_jit_cache.py | .py | """CPU-only tests for the JIT build pipeline: ninja generation and the cache.
Everything here runs against synthetic files, so the invariants that matter — a
bad recorded dependency list never causes reuse, differing flags never share a
directory, a moved clone still hits — are checked without a GPU or a compiler.
"""... | 533 | 19,552 |
sglang | test/registered/kernels/test_kda_replayssm_ring_fused.py | .py | """Parity: CACHE_RING verify kernel's ring == the state the verify kernel commits.
This fuses the ReplaySSM ring-write into the recurrent verify kernel
(`fused_sigmoid_gating_delta_rule_update`, CACHE_RING=True): every draft step it
stores the pre-norm k / raw v / in-kernel gate / beta into the per-slot ring, in
place... | 146 | 5,327 |
sglang | test/registered/kernels/test_dcp_lse_combine.py | .py | """Tests for DCP LSE combine kernels.
Covers:
1. Triton LSE combine kernel correctness vs CPU reference (base-e and base-2)
2. Various DCP world sizes (N=1,2,4,8)
3. Edge cases: single shard, dominant LSE, equal LSE, NaN/inf
4. return_lse mode
5. dcp_a2a_lse_reduce with pre-allocated CUDA graph buffers
"""
import uni... | 567 | 19,633 |
sglang | test/registered/kernels/test_kda_replayssm_fold_batched.py | .py | """Parity: layer-batched fold == per-layer loop of commit_kda_replayssm_spec.
The commit-side ReplaySSM fold used to launch commit_kda_replayssm_spec once per
KDA layer (a Python loop -> ~69 tiny eager launches at bs=1, dispatch-bound).
`commit_kda_replayssm_spec_all_layers` packs the layer into the head grid axis so
... | 141 | 4,715 |
sglang | test/registered/kernels/test_kda_replayssm_fold.py | .py | """Parity: KDA ReplaySSM fold-commit vs the recurrent verify kernel.
The fold kernel (`kda_replayssm_exact_fold_kernel`) replays a request's accepted
draft window from a checkpoint to reconstruct the committed SSM state, replacing
the per-step `intermediate_ssm` snapshots. This test pins the fold's committed
state to ... | 148 | 5,855 |
sglang | test/registered/kernels/test_kda_replayssm_ring_ragged.py | .py | """Parity: CACHE_RING ring-write under ragged (varlen) verify layouts.
Packed varlen verify (per-row verify_lens <= gamma): the fold of the first
acc entries of each row's ring must match the kernel's own per-step state,
as in the dense parity test. Covers partial commit (the compact commit shape)
and padding (-1) slo... | 151 | 4,903 |
sglang | test/registered/kernels/benchmark/speculative/bench_spec_topk1.py | .py | """Benchmark CUDA topk=1 speculative decoding helpers."""
from __future__ import annotations
import torch
import triton
import triton.testing
from sglang.kernels.jit.benchmark.utils import (
DEFAULT_DEVICE,
get_benchmark_range,
run_benchmark,
)
from sglang.kernels.ops.speculative.topk1 import draft_topk1... | 167 | 5,329 |
sglang | test/registered/kernels/benchmark/speculative/bench_ngram_update_token_table.py | .py | import torch
import triton
import triton.testing
from sglang.kernels.jit.benchmark.utils import (
DEFAULT_DEVICE,
get_benchmark_range,
run_benchmark_no_cudagraph,
)
from sglang.kernels.ops.speculative.ngram_embedding import (
update_token_table,
update_token_table_decode,
)
from sglang.test.ci.ci_r... | 80 | 2,241 |
sglang | test/registered/kernels/benchmark/speculative/bench_ngram_compute_decode.py | .py | import torch
import triton
import triton.testing
from sglang.kernels.jit.benchmark.utils import (
DEFAULT_DEVICE,
get_benchmark_range,
run_benchmark_no_cudagraph,
)
from sglang.kernels.ops.speculative.ngram_embedding import (
compute_n_gram_ids,
compute_n_gram_ids_decode,
)
from sglang.test.ci.ci_r... | 128 | 3,710 |
sglang | test/registered/kernels/benchmark/activation/bench_activation.py | .py | import torch
import torch.nn.functional as F
from sgl_kernel import gelu_and_mul as gelu_and_mul_aot
from sgl_kernel import gelu_tanh_and_mul as gelu_tanh_and_mul_aot
from sgl_kernel import silu_and_mul as silu_and_mul_aot
from sglang.kernels.jit.benchmark import marker
from sglang.kernels.jit.benchmark.utils import c... | 114 | 4,275 |
sglang | test/registered/kernels/benchmark/kvcache/bench_hicache.py | .py | """Benchmark for HiCache JIT kernel performance.
This benchmark tests the performance of KV cache transfer operations
between GPU and CPU (host pinned memory), comparing:
- SGL AOT Kernel: Pre-compiled transfer_kv kernels from sgl_kernel
- SGL JIT Kernel: JIT-compiled hicache kernels
- PyTorch Indexing: Plain PyTorch ... | 425 | 13,011 |
sglang | test/registered/kernels/benchmark/kvcache/bench_fused_fp8_qkv_kv_cache.py | .py | import torch
from sglang.kernels.jit.benchmark import marker
from sglang.kernels.ops.kvcache.fused_fp8_qkv_kv_cache import fused_fp8_qkv_kv_cache
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(
est_time=6, stage="base-b-kernel-benchmark", runner_config="1-gpu-large"
)
FP8 = torch.float8... | 54 | 1,895 |
sglang | test/registered/kernels/benchmark/kvcache/bench_set_mla_kv_buffer.py | .py | """Benchmark the set_mla_kv_buffer dispatcher.
Compares three providers across a batch-size sweep:
- ``wrapper``: the high-level wrapper exposed by ``set_mla_kv_buffer_triton``
(dispatches to TMA on SM90+, Triton fallback otherwise).
- ``jit_tma``: the JIT CUDA TMA bulk-store kernel directly... | 132 | 3,828 |
sglang | test/registered/kernels/benchmark/kvcache/bench_store_cache.py | .py | import torch
from sglang.kernels.jit.benchmark import marker
from sglang.kernels.jit.benchmark.utils import (
DEFAULT_DEVICE,
create_empty,
create_random,
)
from sglang.kernels.ops.kvcache.kvcache import store_cache
from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
register_cuda_ci(... | 105 | 3,595 |
sglang | test/registered/kernels/benchmark/kvcache/bench_hisparse.py | .py | import itertools
from typing import Dict, Tuple
import torch
import triton
import triton.testing
from sglang.kernels.jit.benchmark.utils import DEFAULT_DEVICE, DEFAULT_DTYPE
from sglang.kernels.ops.kvcache.hisparse import (
copy_cache_planned_mla,
load_cache_to_device_buffer_mla,
)
from sglang.test.ci.ci_regi... | 286 | 9,497 |
sglang | test/registered/kernels/benchmark/kvcache/bench_minimax_store_kv_index.py | .py | """Benchmark: fused MiniMax-M3 KV + index cache store (1 launch) vs the separate
per-buffer index_put_ stores (main K, main V, index K, optional index V)."""
import torch
from sglang.kernels.jit.benchmark import marker
from sglang.kernels.ops.kvcache.minimax_store_kv_index import store_kv_index
from sglang.test.ci.ci... | 71 | 2,088 |
sglang | test/registered/kernels/benchmark/embeddings/bench_vocab_parallel_embedding.py | .py | import torch
import torch.nn.functional as F
import triton
import triton.testing
from sglang.kernels.jit.benchmark.utils import get_benchmark_range, run_benchmark
from sglang.kernels.ops.embeddings.vocab_parallel_embedding import (
vocab_parallel_embedding,
)
from sglang.srt.layers.vocab_parallel_embedding import ... | 187 | 5,408 |
sglang | test/registered/kernels/benchmark/elementwise/bench_add_constant.py | .py | import torch
import triton
import triton.testing
from sglang.kernels.jit.benchmark.utils import (
DEFAULT_DEVICE,
get_benchmark_range,
run_benchmark_no_cudagraph,
)
from sglang.kernels.ops.elementwise.add_constant import (
_jit_add_constant_module,
add_constant,
)
from sglang.test.ci.ci_register im... | 66 | 1,663 |
sglang | test/registered/kernels/benchmark/diffusion/bench_ltx2_qknorm_split_rope.py | .py | import random
import sys
from dataclasses import dataclass
import torch
from sglang.kernels.ops.diffusion.ltx2_qknorm_split_rope import (
ltx2_qknorm_split_rope_cuda,
)
from sglang.test.ci.ci_register import register_cuda_ci
from sglang.utils import is_in_ci
register_cuda_ci(
est_time=30,
stage="base-b-k... | 210 | 6,728 |
sglang | test/registered/kernels/benchmark/diffusion/bench_causal_conv3d_cat_pad.py | .py | from dataclasses import dataclass
import torch
from sglang.kernels.jit.benchmark import marker
from sglang.kernels.ops.diffusion.causal_conv3d_cat_pad import (
fused_causal_conv3d_cat_pad_cuda,
)
from sglang.kernels.ops.diffusion.triton.causal_conv3d_pad import (
fused_causal_conv3d_cat_pad as fused_causal_co... | 96 | 2,966 |
sglang | test/registered/kernels/benchmark/diffusion/bench_fused_norm_scale_shift.py | .py | # Benchmarks SGLang fused layernorm/rmsnorm scale shift kernels
# 1. fused_norm_scale_shift
# 2. fused_scale_residual_norm_scale_shift
import itertools
from typing import Tuple
import torch
import triton
import triton.testing
from sglang.kernels.jit.benchmark.utils import run_benchmark_no_cudagraph
from sglang.multim... | 140 | 4,902 |
sglang | test/registered/kernels/benchmark/diffusion/bench_qwen_image_modulation.py | .py | from typing import Tuple
import torch
import triton.testing
from sglang.kernels.jit.benchmark.utils import run_benchmark_no_cudagraph
from sglang.kernels.ops.diffusion.triton.norm import norm_infer
from sglang.kernels.ops.diffusion.triton.scale_shift import (
fuse_layernorm_scale_shift_gate_select01_kernel,
f... | 188 | 6,128 |
sglang | test/registered/kernels/benchmark/diffusion/bench_norm_impls.py | .py | import argparse
import csv
import functools
import importlib
import math
import os
import statistics
import subprocess
import sys
from pathlib import Path
from typing import Callable
import torch
import torch.nn.functional as F
from sglang.kernels.jit.benchmark.utils import DEFAULT_DEVICE
from sglang.kernels.jit.util... | 756 | 26,501 |
sglang | test/registered/kernels/benchmark/diffusion/bench_diffusion_nvfp4_scaled_mm.py | .py | import argparse
import csv
import json
import os
import re
import statistics
from pathlib import Path
from typing import Any, Callable
import flashinfer
import torch
from sglang.kernels.jit.benchmark.utils import DEFAULT_DTYPE
from sglang.kernels.jit.utils import KERNEL_PATH
from sglang.test.ci.ci_register import reg... | 359 | 11,533 |
sglang | test/registered/kernels/benchmark/diffusion/bench_group_norm_silu.py | .py | import argparse
import csv
import statistics
import sys
from dataclasses import dataclass
from pathlib import Path
from typing import Callable
import torch
import torch.nn.functional as F
import triton.testing
from sglang.kernels.ops.diffusion.triton.group_norm_silu import triton_group_norm_silu
from sglang.test.ci.c... | 316 | 10,149 |
sglang | test/registered/kernels/benchmark/diffusion/bench_residual_gate_add.py | .py | import random
import sys
from dataclasses import dataclass
import torch
from sglang.kernels.ops.diffusion.residual_gate_add import residual_gate_add_cuda
from sglang.kernels.ops.diffusion.triton.scale_shift import fuse_scale_shift_kernel
from sglang.test.ci.ci_register import register_cuda_ci
from sglang.utils import... | 117 | 3,830 |
sglang | test/registered/kernels/benchmark/diffusion/bench_qknorm_rope.py | .py | from dataclasses import dataclass
from typing import Tuple
import torch
import triton
import triton.testing
from sglang.kernels.jit.benchmark.utils import (
DEFAULT_DEVICE,
DEFAULT_DTYPE,
get_benchmark_range,
run_benchmark_no_cudagraph,
)
from sglang.test.ci.ci_register import register_cuda_ci
regist... | 193 | 5,576 |
sglang | test/registered/kernels/benchmark/layernorm/bench_qknorm.py | .py | import torch
from sglang.kernels.jit.benchmark import marker
from sglang.kernels.jit.benchmark.utils import create_random
from sglang.kernels.ops.layernorm.norm import fused_inplace_qknorm
from sglang.srt.utils import get_current_device_stream_fast
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda... | 78 | 2,348 |
sglang | test/registered/kernels/benchmark/layernorm/bench_fused_eh_norm.py | .py | from __future__ import annotations
import torch
from sglang.kernels.jit.benchmark import marker
from sglang.kernels.ops.layernorm.fused_eh_norm import fused_eh_norm
from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
register_cuda_ci(
est_time=6, stage="base-b-kernel-benchmark", runner_confi... | 67 | 1,949 |
sglang | test/registered/kernels/benchmark/layernorm/bench_qknorm_across_heads.py | .py | import itertools
from typing import Tuple
import torch
import triton
import triton.testing
from sgl_kernel import rmsnorm
from sglang.kernels.jit.benchmark.utils import run_benchmark
from sglang.kernels.ops.layernorm.norm import fused_inplace_qknorm_across_heads
from sglang.srt.utils import get_current_device_stream_... | 126 | 3,461 |
sglang | test/registered/kernels/benchmark/layernorm/bench_norm.py | .py | import itertools
import torch
import triton
import triton.testing
from flashinfer.norm import fused_add_rmsnorm as fi_fused_add_rmsnorm
from flashinfer.norm import rmsnorm as fi_rmsnorm
from sglang.kernels.jit.benchmark.utils import get_benchmark_range, run_benchmark
from sglang.kernels.ops.layernorm.norm import fuse... | 103 | 3,094 |
sglang | test/registered/kernels/benchmark/quantization/bench_hadamard.py | .py | import itertools
import math
from typing import Tuple
import torch
import torch.nn.functional as F
import triton
import triton.testing
from sglang.kernels.jit.benchmark.utils import (
DEFAULT_DEVICE,
DEFAULT_DTYPE,
get_benchmark_range,
run_benchmark,
)
from sglang.kernels.ops.quantization.hadamard imp... | 122 | 3,483 |
sglang | test/registered/kernels/benchmark/quantization/bench_awq_dequantize.py | .py | import itertools
import torch
import triton
import triton.testing
from sglang.kernels.jit.benchmark.utils import run_benchmark
from sglang.kernels.ops.quantization.awq_dequantize import (
awq_dequantize as jit_awq_dequantize,
)
from sglang.test.ci.ci_register import register_cuda_ci
from sglang.utils import is_in... | 127 | 3,394 |
sglang | test/registered/kernels/benchmark/quantization/bench_per_token_group_quant_8bit_v2.py | .py | import torch
from sgl_kernel import sgl_per_token_group_quant_8bit
from sglang.kernels.jit.benchmark import marker
from sglang.kernels.jit.benchmark.utils import create_random
from sglang.kernels.ops.quantization.fp8_kernel import (
create_per_token_group_quant_fp8_output_scale,
fp8_dtype,
fp8_max,
fp8... | 68 | 1,893 |
sglang | test/registered/kernels/benchmark/quantization/bench_per_tensor_quant_fp8.py | .py | from typing import Optional, Tuple
import torch
import triton
import triton.testing
from sglang.kernels.jit.benchmark.utils import get_benchmark_range, run_benchmark
from sglang.kernels.ops.quantization.per_tensor_quant_fp8 import (
per_tensor_quant_fp8,
)
from sglang.test.ci.ci_register import register_cuda_ci
... | 125 | 3,362 |
sglang | test/registered/kernels/benchmark/quantization/bench_per_token_group_quant.py | .py | from sglang.kernels.jit.benchmark import marker
from sglang.kernels.jit.benchmark.utils import create_empty, create_random
from sglang.kernels.ops.quantization.fp8_kernel import (
create_per_token_group_quant_fp8_output_scale,
fp8_dtype,
fp8_max,
fp8_min,
)
# per_token_group_quant_8bit_v2 is DEPRECATED... | 78 | 2,300 |
sglang | test/registered/kernels/benchmark/quantization/bench_per_token_group_quant_masked.py | .py | import math
import torch
from sglang.kernels.jit.benchmark import marker
from sglang.kernels.jit.benchmark.utils import create_empty, create_random
from sglang.kernels.ops.quantization.fp8_kernel import (
create_per_token_group_quant_fp8_output_scale,
fp8_dtype,
fp8_max,
fp8_min,
)
# per_token_group_... | 124 | 4,095 |
sglang | test/registered/kernels/benchmark/quantization/bench_per_token_quant_fp8.py | .py | import torch
from sgl_kernel import sgl_per_token_quant_fp8 as aot_per_token_quant_fp8
from sglang.kernels.jit.benchmark import marker
from sglang.kernels.jit.benchmark.utils import create_random
from sglang.kernels.ops.quantization.per_token_quant_fp8 import per_token_quant_fp8
from sglang.test.ci.ci_register import ... | 43 | 1,390 |
sglang | test/registered/kernels/benchmark/moe/bench_moe_fused_gate.py | .py | import torch
from sglang.kernels.jit.benchmark import marker
from sglang.kernels.jit.benchmark.utils import create_random
from sglang.kernels.ops.moe.moe_fused_gate import moe_fused_gate, moe_fused_gate_jit
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(
est_time=20, stage="base-b-kernel... | 64 | 2,160 |
sglang | test/registered/kernels/benchmark/moe/bench_post_reorder_deepgemm.py | .py | import torch
from sglang.kernels.jit.benchmark import marker
from sglang.kernels.ops.moe.ep_moe_kernels import (
post_reorder_deepgemm,
post_reorder_triton_kernel,
)
from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
register_cuda_ci(
est_time=8, stage="base-b-kernel-benchmark", runn... | 81 | 2,286 |
sglang | test/registered/kernels/benchmark/moe/bench_renorm.py | .py | import itertools
import sgl_kernel
import torch
import triton
import triton.testing
from sglang.kernels.jit.benchmark.utils import run_benchmark_no_cudagraph
from sglang.test.ci.ci_register import register_cuda_ci
from sglang.utils import is_in_ci
register_cuda_ci(
est_time=5, stage="base-b-kernel-benchmark", ru... | 242 | 8,593 |
sglang | test/registered/kernels/benchmark/communication/bench_symm_mem_all_gather.py | .py | """Benchmark the symmetric-memory multimem all-gather vs NCCL.
Providers:
- ``nccl`` : ``all_gather_into_tensor`` + concat-along-hidden reshape
(what ``tensor_model_parallel_all_gather(dim=-1)`` does)
- ``mm_safe`` : multimem kernel, ``safe=True`` (clones the buffer view)
- ``mm`` ... | 172 | 5,669 |
sglang | test/registered/kernels/benchmark/communication/bench_custom_all_reduce.py | .py | from __future__ import annotations
import atexit
import contextlib
import logging
import os
from typing import Optional
import torch
import torch.distributed as dist
import sglang.srt.distributed.parallel_state as ps
from sglang.kernels.jit.benchmark import marker
from sglang.kernels.jit.benchmark.utils import get_b... | 281 | 9,712 |
sglang | test/registered/kernels/benchmark/communication/bench_tp_qknorm.py | .py | """Benchmark fused TP QKNorm (push-mode custom-AR + RMSNorm) vs the serial
baseline (RMS sum-sq -> pull-mode all-reduce -> RMS apply).
Usage::
# Benchmark on every supported world size (2..8 GPUs):
python benchmark/bench_tp_qknorm.py
# Specific world sizes:
python benchmark/bench_tp_qknorm.py --num-gp... | 234 | 8,026 |
sglang | test/registered/kernels/benchmark/gemm/bench_dsv3_fused_a_gemm.py | .py | """Benchmark for DeepSeek V3 fused QKV-A GEMM: CuTe DSL vs CUDA JIT vs torch.
Run on SM90+ (Hopper or later):
python test/registered/jit/benchmark/bench_dsv3_fused_a_gemm.py
"""
import torch
import torch.nn.functional as F
import triton.testing
from sglang.kernels.jit.benchmark import marker
from sglang.kernels.... | 87 | 2,496 |
sglang | test/registered/kernels/benchmark/gemm/bench_dsv3_router_gemm.py | .py | """Benchmark for DeepSeek V3 router GEMM (JIT kernel vs torch).
Run on a Hopper (SM90+) GPU:
python -m sglang.kernels.jit.benchmark.bench_dsv3_router_gemm
"""
import torch
import torch.nn.functional as F
from sglang.kernels.jit.benchmark import marker
from sglang.kernels.jit.benchmark.utils import create_random
... | 54 | 1,704 |
sglang | test/registered/kernels/benchmark/gemm/bench_bf16xfp32_router_gemm.py | .py | """Benchmark for the HPC-Ops bf16xfp32 router GEMM (HPC-Ops vs cublas fp32).
`linear_bf16_fp32` computes `x[m, k](bf16) @ w[n, k](fp32)^T`. The `hpc`
provider requires HPC-Ops (https://github.com/Tencent/hpc-ops) installed and
a Hopper GPU (sm90a); it decomposes the fp32 weight into two cached bf16
halves and runs bot... | 69 | 2,278 |
sglang | test/registered/kernels/benchmark/gemm/bench_fp8_blockwise_gemm.py | .py | from __future__ import annotations
import sys
import torch
import triton
from sglang.kernels.jit.benchmark.utils import get_benchmark_range, run_benchmark
from sglang.kernels.ops.gemm.fp8_blockwise_gemm import fp8_blockwise_scaled_mm
from sglang.srt.utils import is_sm120_supported
from sglang.test.ci.ci_register imp... | 105 | 3,402 |
sglang | test/registered/kernels/benchmark/attention/bench_clamp_position.py | .py | import itertools
import torch
import triton
import triton.testing
from sglang.kernels.jit.benchmark.utils import (
DEFAULT_DEVICE,
get_benchmark_range,
run_benchmark,
)
from sglang.kernels.ops.attention.clamp_position import clamp_position_cuda
from sglang.srt.utils import get_compiler_backend
from sglang... | 68 | 1,823 |
sglang | test/registered/kernels/benchmark/attention/bench_minimax_decode_topk.py | .py | """Benchmark: MiniMax-M3 single-stage radix-select decode topk (JIT CUDA) vs the
2-stage split-K Triton baseline (_topk_index_partial_kernel + _topk_index_merge_kernel).
Both consume the decode score tensor [num_heads, batch, max_seqblock] and produce
topk_idx [num_heads, batch, topk]. The JIT kernel is one launch wit... | 125 | 3,778 |
sglang | test/registered/kernels/benchmark/attention/bench_rope.py | .py | import itertools
import torch
import triton
import triton.testing
from sglang.kernels.jit.benchmark.utils import (
DEFAULT_DEVICE,
DEFAULT_DTYPE,
get_benchmark_range,
run_benchmark,
)
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(
est_time=6, stage="base-b-kernel-benchm... | 311 | 8,656 |
sglang | test/registered/kernels/benchmark/attention/bench_mla_kv_pack_quantize_fp8.py | .py | """Bench the hybrid ``mla_kv_pack_quantize_fp8`` against an inlined naive Triton baseline."""
import itertools
from typing import Tuple
import torch
import triton
import triton.language as tl
import triton.testing
from sglang.kernels.jit.benchmark.utils import (
DEFAULT_DEVICE,
DEFAULT_DTYPE,
DEFAULT_QUA... | 218 | 6,100 |
sglang | test/registered/kernels/benchmark/attention/bench_dsv4_fp4_indexer.py | .py | from __future__ import annotations
import sys
import torch
import triton
from sglang.benchmark.bench_utils import run_bench
from sglang.kernels.jit.benchmark.utils import get_benchmark_range
from sglang.srt.utils import is_sm100_supported, is_sm120_supported
from sglang.test.ci.ci_register import register_cuda_ci
r... | 180 | 5,945 |
sglang | test/registered/kernels/benchmark/attention/bench_fused_qknorm_rope.py | .py | """
Benchmark: fused_qknorm_rope JIT vs AOT (sgl_kernel)
Measures throughput (µs) for fused_qk_norm_rope across typical
LLM configurations (head_dim × num_heads × num_tokens).
Run:
python test/registered/jit/benchmark/bench_fused_qknorm_rope.py
"""
import itertools
import torch
import triton
import triton.testi... | 270 | 8,672 |
sglang | test/registered/kernels/benchmark/attention/bench_sparse_mla_q8kv8_prefill_sm90.py | .py | from __future__ import annotations
import math
import torch
import triton
import triton.testing
from sglang.kernels.jit.benchmark.utils import run_benchmark_no_cudagraph
from sglang.kernels.ops.attention.sparse_mla_q8kv8_prefill_sm90 import (
sparse_mla_q8kv8_prefill_fwd,
)
from sglang.srt.utils import is_sm90_s... | 155 | 4,689 |
sglang | test/registered/kernels/benchmark/attention/bench_topk.py | .py | import torch
from sglang.kernels.jit.benchmark import marker
from sglang.kernels.ops.attention.dsv4.topk import (
plan_topk_v2,
topk_transform_512,
topk_transform_512_v2,
)
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(
est_time=120, stage="base-b-kernel-benchmark", runner_c... | 91 | 3,367 |
sglang | test/registered/kernels/benchmark/attention/bench_online_c128_mtp.py | .py | """Benchmark online c128 MTP write-prefix kernel."""
from __future__ import annotations
import itertools
from dataclasses import dataclass
import torch
import triton
import triton.testing
from sglang.kernels.jit.benchmark.utils import (
DEFAULT_DEVICE,
get_benchmark_range,
run_benchmark,
run_benchma... | 166 | 5,058 |
sglang | test/registered/kernels/benchmark/attention/bench_concat_mla.py | .py | import itertools
import torch
import triton
import triton.testing
from sgl_kernel import concat_mla_absorb_q as aot_absorb_q
from sgl_kernel import concat_mla_k as aot_k
from sglang.kernels.jit.benchmark.utils import run_benchmark
from sglang.kernels.ops.attention.concat_mla import concat_mla_absorb_q as jit_absorb_q... | 165 | 4,425 |
sglang | test/registered/kernels/benchmark/attention/bench_minimax_qknorm_rope.py | .py | """Benchmark: fused MiniMax-M3 Gemma-RMSNorm + partial NeoX RoPE (1 in-place
launch) vs the unfused path (GemmaRMSNorm(q) + GemmaRMSNorm(k) + rotary_emb,
3 launches + intermediates). Main attention branch, per-rank TP8 shape (nq=8, nk=1).
"""
import torch
from sglang.kernels.jit.benchmark import marker
from sglang.ke... | 149 | 4,740 |
sglang | test/registered/kernels/benchmark/kv_canary/bench_scatter_req_token_ids.py | .py | from __future__ import annotations
from dataclasses import dataclass
import torch
import triton
import triton.testing
from sglang.kernels.jit.benchmark.utils import (
DEFAULT_DEVICE,
get_benchmark_range,
run_benchmark_no_cudagraph,
)
from sglang.kernels.ops.kv_canary.scatter_req_token_ids import (
la... | 100 | 3,015 |
sglang | test/registered/kernels/benchmark/kv_canary/bench_write.py | .py | from __future__ import annotations
from typing import Tuple
import torch
import triton
import triton.testing
from sglang.kernels.jit.benchmark.kv_canary.utils import (
RING_CAPACITY,
SWA_WINDOW,
BenchCase,
build_fast_matrix_cases,
build_full_matrix_cases,
cases_to_x_vals,
make_real_kv_sou... | 315 | 10,138 |
sglang | test/registered/kernels/benchmark/kv_canary/bench_verify.py | .py | from __future__ import annotations
from typing import Tuple
import torch
import triton
import triton.testing
from sglang.kernels.jit.benchmark.kv_canary.utils import (
RING_CAPACITY,
SWA_WINDOW,
BenchCase,
build_fast_matrix_cases,
build_full_matrix_cases,
cases_to_x_vals,
make_real_kv_sou... | 319 | 9,597 |
sglang | test/registered/kernels/benchmark/kv_canary/bench_plan.py | .py | from __future__ import annotations
from dataclasses import dataclass
from typing import Optional, Tuple
import torch
import triton
import triton.testing
from sglang.kernels.jit.benchmark.kv_canary.utils import (
POOL_AXIS,
SWA_WINDOW,
BenchCase,
build_fast_matrix_cases,
build_full_matrix_cases,
... | 348 | 10,839 |
sglang | test/registered/kernels/ops/test_kimi_k3_prerequisite_ops.py | .py | """Representative parity coverage for the lightweight Kimi-K3 prerequisites."""
import unittest
import torch
from sglang.kernels.ops.attention.concat_mla import concat_mla_absorb_q
from sglang.kernels.ops.attention.fla.fused_sigmoid_gating_recurrent import (
fused_sigmoid_gating_delta_rule_update,
)
from sglang.... | 486 | 18,772 |
sglang | test/registered/kernels/ops/speculative/test_gather_spec_extras.py | .py | from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
register_cuda_ci(est_time=10, stage="base-b", runner_config="1-gpu-small")
register_amd_ci(est_time=10, suite="nightly-amd-kernel-1-gpu", nightly=True)
import unittest
import torch
from sglang.kernels.ops.speculative.gather_spec_extras import ... | 216 | 8,676 |
sglang | test/registered/kernels/ops/speculative/test_boundary_kv_fix_kernels.py | .py | from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=15, stage="base-b", runner_config="1-gpu-small")
"""Boundary-KV fix kernels (SGLANG_ENABLE_MTP_BOUNDARY_KV_FIX) vs a pure-torch reference.
Covers the three pieces behind the pool-free chain-MTP boundary-KV exactness
fix (widened draft-... | 230 | 8,788 |
sglang | test/registered/kernels/ops/speculative/test_spec_topk1.py | .py | from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=10, stage="base-b", runner_config="1-gpu-small")
import unittest
import torch
from sglang.kernels.ops.speculative.topk1 import draft_topk1_postprocess
from sglang.test.test_utils import CustomTestCase
def _make_logits_with_unique_ar... | 164 | 6,211 |
sglang | test/registered/kernels/ops/speculative/test_ngram_embedding.py | .py | import sys
import pytest
import torch
from sglang.kernels.ops.speculative.ngram_embedding import (
compute_n_gram_ids,
compute_n_gram_ids_decode,
update_token_table,
update_token_table_decode,
)
from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
register_cuda_ci(est_time=30, sta... | 142 | 4,607 |
sglang | test/registered/kernels/ops/activation/test_activation.py | .py | import sys
import pytest
import torch
import torch.nn.functional as F
from sglang.kernels.jit.utils import get_ci_test_range
from sglang.kernels.ops.activation.activation import (
SUPPORTED_ACTIVATIONS,
relu2,
run_activation,
)
from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
regi... | 214 | 7,280 |
sglang | test/registered/kernels/ops/kvcache/test_set_mla_kv_buffer.py | .py | import sys
import pytest
import torch
from sglang.kernels.jit.utils import get_ci_test_range
from sglang.kernels.ops.kvcache.set_mla_kv_buffer import (
can_use_set_mla_kv_buffer,
set_mla_kv_buffer,
)
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=30, stage="base-b-kernel-un... | 127 | 4,394 |
sglang | test/registered/kernels/ops/kvcache/test_hicache.py | .py | import sys
import pytest
import torch
from sglang.kernels.ops.kvcache.hicache import can_use_write_back_jit_kernel
from sglang.srt.mem_cache.memory_pool import MHATokenToKVPool, MLATokenToKVPool
from sglang.srt.mem_cache.pool_host.common import (
ALLOC_MEMORY_FUNCS,
alloc_with_pin_memory,
)
from sglang.srt.me... | 501 | 17,486 |
sglang | test/registered/kernels/ops/kvcache/test_store_cache.py | .py | import itertools
import sys
import pytest
import torch
from sglang.kernels.jit.utils import get_ci_test_range
from sglang.kernels.ops.kvcache.kvcache import can_use_store_cache, store_cache
from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
register_cuda_ci(est_time=28, stage="base-b-kernel-uni... | 273 | 11,259 |
sglang | test/registered/kernels/ops/kvcache/test_hicache_page_first_write_back.py | .py | """Unit tests for the page_first + ``kernel`` JIT HiCache write-back / load path.
This file specifically exercises the JIT staged write-back and load kernels that
accept a CPU-resident destination index and stage through device memory
(``staged_write_back.cuh`` / ``hicache.cuh``). Unlike ``test_hicache.py`` (which
is ... | 260 | 8,861 |
sglang | test/registered/kernels/ops/kvcache/test_minimax_store_kv_index.py | .py | """Correctness for the fused MiniMax-M3 KV + index cache store kernel.
Verifies the single fused launch writes the main K/V, the index K, and the
optional index V into their pools at out_cache_loc rows exactly as the separate
index_put_ stores would, for both value modes and int32/int64 indices.
"""
import pytest
imp... | 80 | 2,546 |
sglang | test/registered/kernels/ops/kvcache/test_hisparse.py | .py | import sys
import pytest
import torch
from sglang.kernels.ops.kvcache.hisparse import (
load_cache_to_device_buffer_dsv4_mla,
load_cache_to_device_buffer_mla,
transfer_cache_dsv4_mla,
)
from sglang.srt.utils import is_cuda, is_hip, is_npu, is_xpu
from sglang.test.ci.ci_register import register_amd_ci, reg... | 759 | 28,157 |
sglang | test/registered/kernels/ops/kvcache/test_kvcacheio_asymmetric.py | .py | import sys
from types import SimpleNamespace
import pytest
import torch
from sglang.srt.mem_cache.pool_host.mha import AsymmetricMHATokenToKVPoolHost
from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
register_cuda_ci(est_time=10, stage="base-b", runner_config="1-gpu-large")
register_amd_ci(est... | 214 | 7,853 |
sglang | test/registered/kernels/ops/kvcache/test_fused_fp8_qkv_kv_cache.py | .py | import pytest
import torch
from sglang.kernels.ops.kvcache.fused_fp8_qkv_kv_cache import fused_fp8_qkv_kv_cache
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=40, stage="base-b-kernel-unit", runner_config="1-gpu-large")
register_cuda_ci(est_time=40, stage="base-b-kernel-unit", runne... | 92 | 3,275 |
sglang | test/registered/kernels/ops/embeddings/test_vocab_parallel_embedding.py | .py | import types
import pytest
import torch
import torch.nn.functional as F
from sglang.kernels.ops.embeddings.vocab_parallel_embedding import (
vocab_parallel_embedding,
)
from sglang.srt.layers.quantization.unquant import UnquantizedEmbeddingMethod
from sglang.srt.layers.vocab_parallel_embedding import (
VocabP... | 143 | 5,218 |
sglang | test/registered/kernels/ops/mamba/test_sconv_cache.py | .py | import pytest
import torch
from sglang.srt.models.inkling_common.kernels.sconv import (
HIS_PREFIX,
HIS_ZEROS,
PAD_SLOT_ID,
causal_conv1d,
fused_decode_sconv_metadata,
fused_extend_sconv_metadata,
update_sconv_cache,
)
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_c... | 172 | 5,317 |
sglang | test/registered/kernels/ops/mamba/test_sconv_extend_metadata.py | .py | """fused_extend_sconv_metadata must be bit-identical to the unfused prep.
The unfused reference is the exact op sequence the extend metadata prep
+ precompute_helion_extend_metadata used to launch: zeros + cumsum + slice-copy
(or arange + ones for verify) + the has_initial_state compare, then != PAD, &,
clamp, long, t... | 178 | 5,856 |
sglang | test/registered/kernels/ops/mamba/test_sconv_decode_metadata.py | .py | """fused_decode_sconv_metadata must be bit-identical to the unfused prep.
The unfused reference is the exact op sequence the decode metadata prep
used to launch: two arange calls + ones + precompute_helion_decode_metadata
(!= PAD, &, clamp, long, arange x2).
"""
import pytest
import torch
from sglang.srt.models.inkl... | 74 | 2,692 |
sglang | test/registered/kernels/ops/mamba/test_fused_replay_state_indices.py | .py | """fused_replay_state_indices must be bit-identical to the unfused prep.
The unfused reference is the exact op sequence ``_replay_metadata`` used to
launch for the static hybrid pool:
req_pool_indices[valid_bs:total_bs] = 0 # zero padded rows (side effect)
mamba_indices = mapping[req_pool_indices] ... | 193 | 7,820 |
sglang | test/registered/kernels/ops/mamba/test_transfer_mamba.py | .py | """Unit tests for the Mamba JIT transfer kernel.
Verifies kernel backup (D2H) and load (H2D) correctness for
``MambaPoolHost`` via the ``io_backend='kernel'`` path, across both
supported layouts and multiple index scenarios.
"""
import sys
import threading
from types import SimpleNamespace
import pytest
import torch... | 350 | 12,919 |
sglang | test/registered/kernels/ops/elementwise/test_add_constant.py | .py | import sys
import pytest
import torch
from sglang.kernels.ops.elementwise.add_constant import add_constant
from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
register_cuda_ci(est_time=45, stage="base-b-kernel-unit", runner_config="1-gpu-large")
register_amd_ci(est_time=8, stage="jit-kernel-unit... | 42 | 1,404 |
sglang | test/registered/kernels/ops/diffusion/test_flux2_eager_fusions.py | .py | """FLUX.2 eager fusions must be bit-exact for real packed/view layouts."""
import unittest
from unittest.mock import patch
import torch
import torch.nn.functional as F
import sglang.multimodal_gen.runtime.models.dits.flux_2 as flux2
from sglang.multimodal_gen.runtime.models.dits.flux_2 import (
_flux2_norm_modul... | 102 | 3,960 |
sglang | test/registered/kernels/ops/diffusion/test_diffusion_nvfp4_scaled_mm.py | .py | import sys
import flashinfer
import pytest
import torch
from sglang.multimodal_gen.runtime.layers.quantization import (
modelopt_quant as diffusion_modelopt_quant,
)
from sglang.multimodal_gen.runtime.layers.quantization.modelopt_quant import (
ModelOptFp4Config,
ModelOptFp4LinearMethod,
)
from sglang.mul... | 448 | 16,120 |
sglang | test/registered/kernels/ops/diffusion/test_wan_vae_fastpath.py | .py | """Wan VAE decoder fast path: fused-kernel numerics and gate dispatch
(the lossless off-path must stay bit-exact)."""
import sys
import pytest
import torch
import torch.nn as nn
import torch.nn.functional as F
from sglang.kernels.ops.diffusion.triton.wan_rmsnorm_silu import wan_rmsnorm_silu
from sglang.multimodal_ge... | 80 | 2,901 |
sglang | test/registered/kernels/ops/diffusion/test_fused_linear_gelu.py | .py | """Core checks for the quality-gated linear + tanh-GELU fusion."""
import sys
import pytest
import torch
import torch.nn as nn
import torch.nn.functional as F
from sglang.kernels.ops.diffusion import fused_linear_gelu as gelu
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=4, stage... | 87 | 3,125 |
sglang | test/registered/kernels/ops/diffusion/test_group_norm_silu.py | .py | import sys
import pytest
import torch
import torch.nn as nn
import torch.nn.functional as F
from sglang.kernels.ops.diffusion.group_norm_silu import apply_group_norm_silu
from sglang.kernels.ops.diffusion.triton.group_norm_silu import triton_group_norm_silu
from sglang.test.ci.ci_register import register_amd_ci, regi... | 105 | 3,368 |
sglang | test/registered/kernels/ops/diffusion/test_modulate_scale_shift.py | .py | import pytest
import torch
from sglang.kernels.ops.diffusion.modulate_scale_shift import (
can_use_modulate_scale_shift_cuda,
modulate_scale_shift,
modulate_scale_shift_cuda,
)
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=30, stage="base-b-kernel-unit", runner_config="... | 59 | 2,112 |
sglang | test/registered/kernels/ops/diffusion/test_ulysses_qkv.py | .py | import sys
import pytest
import torch
from sglang.kernels.ops.diffusion.triton.ulysses_qkv import (
pack_qkv_destination_major,
)
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=8, stage="base-b-kernel-unit", runner_config="1-gpu-large")
pytestmark = pytest.mark.skipif(not torch... | 55 | 1,813 |
sglang | test/registered/kernels/ops/diffusion/test_usp_relayout.py | .py | """Bitwise tests for the generic Ulysses output head-merge fast path."""
import sys
from unittest.mock import patch
import pytest
import torch
from sglang.kernels.ops.diffusion.usp_relayout import (
can_use_usp_merge_heads,
usp_merge_heads,
)
from sglang.test.ci.ci_register import register_cuda_ci
register_... | 70 | 2,080 |
sglang | test/registered/kernels/ops/diffusion/test_varlen_pack_pad.py | .py | """Numerical correctness for fused varlen pack/scatter Triton kernels.
Bit-exact comparison against the equivalent PyTorch ops (index_select,
zeros + index_copy_) across bf16/fp16 and several shape/mask cases.
"""
import pytest
import torch
from sglang.kernels.jit.utils import get_ci_test_range
from sglang.kernels.o... | 198 | 7,234 |
sglang | test/registered/kernels/ops/diffusion/test_diffusion_modelopt_fp8_scaled_mm.py | .py | import sys
import pytest
import torch
from sglang.kernels.ops.quantization.fp8_kernel import static_quant_fp8
from sglang.multimodal_gen.runtime.layers.quantization.modelopt_quant import (
ModelOptFp8Config,
ModelOptFp8LinearMethod,
)
from sglang.srt.layers.quantization.fp8_utils import (
cutlass_fp8_supp... | 142 | 4,803 |
sglang | test/registered/kernels/ops/diffusion/test_wan_causal_cache.py | .py | """Wan causal VAE data-movement kernels: the fused conv-input builder and the
fused DupUp3D shortcut add must be bitwise identical to the aten op chains
they replace (they are pure data movement plus zero fill / one fp32 add)."""
import sys
import pytest
import torch
import torch.nn.functional as F
from sglang.kerne... | 165 | 6,212 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.