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/kernels/ops/diffusion/test_glm_image_ln_modulate.py | .py | """GLM-Image fused LN+modulate / qk-LN fast paths must stay bit-exact vs eager."""
import pytest
import torch
import sglang.multimodal_gen.runtime.models.dits.glm_image as glm_image
from sglang.multimodal_gen.runtime.models.dits.glm_image import (
_eager_ln_modulate,
_glm_ln_modulate,
_glm_qk_layernorm,
)... | 56 | 2,474 |
sglang | test/registered/kernels/ops/diffusion/test_qwen_image_modulation.py | .py | import sys
import pytest
import torch
import triton
from sglang.kernels.jit.utils import get_ci_test_range
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,
fuse_residual_layernorm_scal... | 229 | 7,294 |
sglang | test/registered/kernels/ops/diffusion/test_ltx2_ada_values.py | .py | import sys
import pytest
import torch
from sglang.kernels.ops.diffusion.triton.ltx2_ada_values import ltx2_ada_values9
from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
register_cuda_ci(est_time=8, stage="base-b-kernel-unit", runner_config="1-gpu-large")
register_amd_ci(est_time=8, suite="nigh... | 93 | 2,893 |
sglang | test/registered/kernels/ops/diffusion/test_bitexact_gate.py | .py | import sys
from types import ModuleType
from unittest.mock import MagicMock, patch
import pytest
import torch
from sglang.kernels.ops.diffusion.bitexact_gate import (
BitExactFusionGate,
flashinfer_rmsnorm_diagnostic_hint,
tensors_equal,
)
from sglang.test.ci.ci_register import register_cpu_ci
from sglang... | 160 | 5,610 |
sglang | test/registered/kernels/ops/diffusion/test_ltx2_rms_norm_modulate.py | .py | """LTX-2 quality=high RMSNorm+modulate fusion: gated, close to eager."""
import sys
import pytest
import torch
from torch import nn
from sglang.kernels.ops.diffusion.ltx2_rmsnorm_modulate import (
fused_ltx2_rms_norm_modulate,
mark_ltx2_rms_norm_modulate_site,
mount_ltx2_rms_norm_modulate,
unmount_lt... | 72 | 2,639 |
sglang | test/registered/kernels/ops/diffusion/test_causal_conv3d_cat_pad.py | .py | import sys
import pytest
import torch
from sglang.kernels.jit.utils import get_ci_test_range
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_conv... | 90 | 2,732 |
sglang | test/registered/kernels/ops/diffusion/test_flux2_vae_fastpath.py | .py | """Focused correctness checks for the FLUX.2 VAE CUDA fast path."""
import sys
import pytest
import torch
import torch.nn as nn
import torch.nn.functional as F
from diffusers.models.upsampling import Upsample2D
from sglang.kernels.ops.diffusion.triton import group_norm_silu_twopass as gn_kernel
from sglang.multimoda... | 65 | 2,337 |
sglang | test/registered/kernels/ops/diffusion/test_qknorm_rope.py | .py | import itertools
import sys
import pytest
import torch
import triton
from sglang.kernels.jit.utils import get_ci_test_range
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=44, stage="base-b-kernel-unit", runner_config="1-gpu-large")
# Nightly is not redundant here: it sets SGLANG_JI... | 441 | 13,922 |
sglang | test/registered/kernels/ops/diffusion/test_fused_ln_modulate.py | .py | import pytest
import torch
import torch.nn as nn
from sglang.kernels.ops.diffusion.fused_ln_modulate import (
can_fuse_ln_modulate,
fused_ln_modulate,
fused_ln_modulate_active,
mark_fused_ln_modulate_site,
mount_fused_ln_modulate,
unmount_fused_ln_modulate,
)
from sglang.test.ci.ci_register imp... | 83 | 3,008 |
sglang | test/registered/kernels/ops/diffusion/test_fused_norm_scale_shift.py | .py | import sys
from typing import Optional, Tuple
import pytest
import torch
from einops import rearrange
from torch import Tensor
from sglang.kernels.ops.diffusion.cutedsl.scale_residual_norm_scale_shift import (
fused_norm_scale_shift,
fused_scale_residual_norm_scale_shift,
validate_scale_shift,
)
from sgla... | 252 | 8,725 |
sglang | test/registered/kernels/ops/diffusion/test_fused_gate_rmsnorm.py | .py | """Core checks for the quality-gated fused gate-RMSNorm path."""
import sys
import pytest
import torch
import torch.nn as nn
import torch.nn.functional as F
from sglang.kernels.ops.diffusion import fused_gate_rmsnorm as fgn
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=4, stage="... | 55 | 2,123 |
sglang | test/registered/kernels/ops/diffusion/test_flydsl_fused_norm.py | .py | import sys
import pytest
import torch
import torch.nn.functional as F
from sglang.test.ci.ci_register import register_amd_ci
register_amd_ci(est_time=30, stage="jit-kernel-unit", runner_config="amd")
DEVICE = "cuda"
D = 5120
EPS = 1e-6
def _ref_rms_norm(x_f32, weight, eps):
var = x_f32.pow(2).mean(-1, keepdim... | 134 | 3,934 |
sglang | test/registered/kernels/ops/diffusion/test_sana_ln_modulate.py | .py | """Sana fused LN+modulate fast path must stay bit-exact vs eager."""
import pytest
import torch
import sglang.multimodal_gen.runtime.models.dits.sana as sana
from sglang.multimodal_gen.runtime.models.dits.sana import (
_eager_ln_modulate,
_sana_ln_modulate,
)
from sglang.test.ci.ci_register import register_cu... | 56 | 2,261 |
sglang | test/registered/kernels/ops/diffusion/test_ernie_norm_scale_shift.py | .py | """ERNIE fused norm/scale/shift fast paths must stay bit-exact vs eager."""
import sys
from unittest.mock import patch
import pytest
import torch
import sglang.multimodal_gen.runtime.models.dits.ernie_image as ernie_image
from sglang.multimodal_gen.runtime.layers.layernorm import RMSNorm
from sglang.multimodal_gen.r... | 136 | 5,044 |
sglang | test/registered/kernels/ops/diffusion/test_stage_profiler_sync.py | .py | """SGLANG_DIFFUSION_SYNC_STAGE_PROFILING must drain the GPU queue at the
timing start of *stage* records too — otherwise a stage that only launches
kernels (DenoisingStage's tail) leaks its queued work into whichever later
stage blocks first, inflating e.g. DecodingStage readings 2-3x."""
import sys
import time
impor... | 51 | 1,958 |
sglang | test/registered/kernels/ops/diffusion/test_varlen_uspattn_equivalence.py | .py | """End-to-end equivalence between USPAttention varlen path and SDPA reference.
Compares the production varlen path (``build_varlen_mask_meta`` +
``fused_pack_qkv`` + ``flash_attn_varlen_func`` + ``fused_scatter_to_padded``)
against ``torch.nn.functional.scaled_dot_product_attention`` with a broadcast
key mask, for inp... | 158 | 5,710 |
sglang | test/registered/kernels/ops/diffusion/test_native_bf16_rmsnorm.py | .py | import pytest
import torch
from sglang.kernels.ops.diffusion.triton.native_bf16_rmsnorm import (
rmsnorm_scale,
rmsnorm_tanh_residual,
)
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")
EPS = 1e-5
def _native_bf16_... | 88 | 3,487 |
sglang | test/registered/kernels/ops/diffusion/test_timestep_embedding.py | .py | import os
import sys
import numpy as np
import pytest
import torch
try:
import tabulate
except Exception:
tabulate = None
from sglang.kernels.jit.utils import get_ci_test_range
from sglang.kernels.ops.diffusion.timestep_embedding import (
timestep_embedding as timestep_embedding_cuda,
)
from sglang.test.... | 185 | 6,031 |
sglang | test/registered/kernels/ops/diffusion/test_scale_shift.py | .py | import sys
import pytest
import torch
from sglang.kernels.ops.diffusion.triton.scale_shift import (
try_fused_scaled_residual_add_exact,
)
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=5, stage="base-b-kernel-unit", runner_config="1-gpu-large")
pytestmark = pytest.mark.skipif(... | 41 | 1,400 |
sglang | test/registered/kernels/ops/diffusion/test_flux_ln_modulate.py | .py | """FLUX.1 fused LN+modulate fast path must stay bit-exact vs eager."""
import pytest
import torch
import sglang.multimodal_gen.runtime.models.dits.flux as flux
from sglang.kernels.ops.diffusion.fused_ln_modulate import (
mark_fused_ln_modulate_site,
mount_fused_ln_modulate,
)
from sglang.multimodal_gen.runtim... | 76 | 2,825 |
sglang | test/registered/kernels/ops/diffusion/test_autoencoder_kl_fastpath.py | .py | """Install-path checks for the generic AutoencoderKL CUDA fast path."""
import sys
import pytest
import torch
from sglang.multimodal_gen.configs.models.vaes.stablediffusion3 import (
StableDiffusion3VAEConfig,
)
from sglang.multimodal_gen.runtime.models.vaes import flux2_vae_cuda_opt as vae_opt
from sglang.multi... | 63 | 2,456 |
sglang | test/registered/kernels/ops/diffusion/test_hunyuanvideo_eager_fusions.py | .py | """HunyuanVideo eager QKV/RoPE and quality-gated QKNorm tests."""
import sys
from unittest.mock import patch
import pytest
import torch
import sglang.kernels.ops.diffusion.hunyuan_qknorm as hunyuan_qknorm
from sglang.kernels.ops.diffusion.hunyuan_qknorm import (
mark_hunyuan_qknorm_site,
mount_hunyuan_qknorm... | 95 | 3,380 |
sglang | test/registered/kernels/ops/diffusion/test_ltx2_qknorm_split_rope.py | .py | import sys
import pytest
import torch
import torch.nn.functional as F
from sglang.kernels.ops.diffusion.ltx2_qknorm_split_rope import (
can_use_ltx2_qknorm_split_rope_cuda,
ltx2_qknorm_split_rope_cuda,
)
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=45, stage="base-b-kerne... | 222 | 6,679 |
sglang | test/registered/kernels/ops/diffusion/test_quality_gate.py | .py | import sys
import pytest
import torch.nn as nn
from sglang.kernels.ops.diffusion.quality_gate import QualityGatedFusion
from sglang.test.ci.ci_register import register_cpu_ci
register_cpu_ci(est_time=2, suite="base-a-test-cpu")
def test_quality_gate_mounts_and_unmounts_all_sites():
fusion = QualityGatedFusion(... | 48 | 1,435 |
sglang | test/registered/kernels/ops/diffusion/test_residual_gate_add.py | .py | import sys
import pytest
import torch
from sglang.kernels.ops.diffusion.residual_gate_add import (
can_use_residual_gate_add_cuda,
residual_gate_add,
residual_gate_add_cuda,
)
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=30, stage="base-b-kernel-unit", runner_config="... | 124 | 4,707 |
sglang | test/registered/kernels/ops/layernorm/test_qknorm.py | .py | import itertools
import sys
import pytest
import torch
import triton
from sglang.kernels.jit.utils import get_ci_test_range
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=37, stage="base-b-kernel-unit", runner_config="1-gpu-large")
# Nightly is not redundant here: it sets SGLANG_JI... | 103 | 3,280 |
sglang | test/registered/kernels/ops/layernorm/test_minimax_m3_rmsnorm.py | .py | # SPDX-License-Identifier: Apache-2.0
"""Reference tests for MiniMax-M3 ROCm Gemma RMSNorm Triton kernels."""
import pytest
import torch
from sglang.srt.utils import is_hip
if not is_hip():
pytest.skip(
"MiniMax-M3 Gemma RMSNorm Triton kernels are ROCm-only.",
allow_module_level=True,
)
if no... | 88 | 2,852 |
sglang | test/registered/kernels/ops/layernorm/test_gemma4_fused_routing.py | .py | """Correctness tests for ``gemma4_fused_routing``.
Compares the Triton-fused routing kernel against the original SGLang
``Gemma4MoE.routing_function`` reference (softmax-of-topk * per_expert_scale).
Run with::
pytest test/registered/kernels/test_gemma4_fused_routing.py -v
Requires a CUDA-capable GPU; skips other... | 107 | 3,698 |
sglang | test/registered/kernels/ops/layernorm/test_rmsnorm_hf.py | .py | """Tests for the JIT rmsnorm_hf kernel (HF LlamaRMSNorm semantics)."""
import itertools
import sys
import pytest
import torch
from sglang.kernels.jit.utils import get_ci_test_range
from sglang.kernels.ops.layernorm.rmsnorm_hf import (
is_supported_rmsnorm_hf_hidden_size,
rmsnorm_hf,
)
from sglang.test.ci.ci_... | 139 | 5,039 |
sglang | test/registered/kernels/ops/layernorm/test_fused_add_rmsnorm.py | .py | import itertools
import sys
import pytest
import torch
from sglang.kernels.jit.utils import get_ci_test_range
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=10, stage="base-b-kernel-unit", runner_config="1-gpu-large")
# Nightly is not redundant here: it sets SGLANG_JIT_KERNEL_RUN_F... | 115 | 3,970 |
sglang | test/registered/kernels/ops/layernorm/test_fused_op_gpu_parity.py | .py | """Every-backend-vs-native parity for BaseFusedOp ops on real GPU (RFC #29630).
For each reworked fused op, run every backend eligible on this platform and
assert it matches the pure-torch ``forward_native`` reference within dtype
tolerance. New backends are picked up automatically.
"""
import pytest
import torch
fr... | 124 | 4,451 |
sglang | test/registered/kernels/ops/layernorm/test_kernels_namespace.py | .py | """GPU-free import / registry / selector tests for ``sglang.kernels`` (RFC #29630)."""
import importlib
import importlib.util
import subprocess
import sys
import pytest
import sglang.kernels as K
import sglang.kernels.fused_op as fo
import sglang.kernels.ops # noqa: F401 -- populate the registry
import sglang.kern... | 237 | 8,630 |
sglang | test/registered/kernels/ops/layernorm/test_mhc_kernels.py | .py | from contextlib import nullcontext
import pytest
import torch
import sglang.kernels.ops.layernorm.mhc as mhc
from sglang.kernels.ops.layernorm.mhc import mhc_fused_post_pre, mhc_post, mhc_pre
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=30, stage="base-b", runner_config="1-gpu-la... | 131 | 4,586 |
sglang | test/registered/kernels/ops/layernorm/test_fused_eh_norm.py | .py | import sys
import pytest
import torch
from sglang.kernels.ops.layernorm.fused_eh_norm import fused_eh_norm
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=45, stage="base-b-kernel-unit", runner_config="1-gpu-large")
pytestmark = pytest.mark.skipif(
not torch.cuda.is_available()... | 122 | 4,510 |
sglang | test/registered/kernels/ops/layernorm/test_fused_op.py | .py | """GPU-free BaseFusedOp + registry / spec unit tests (RFC #29630).
Every-backend-vs-native parity on real hardware lives in
``test_fused_op_gpu_parity.py``.
"""
import math
import pytest
import torch
import sglang.kernels as K
from sglang.kernels.fused_op import BaseFusedOp
from sglang.kernels.registry import Kerne... | 241 | 6,903 |
sglang | test/registered/kernels/ops/layernorm/test_qknorm_across_heads.py | .py | import itertools
import sys
import pytest
import torch
import triton
from sglang.kernels.jit.utils import get_ci_test_range
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=15, stage="base-b-kernel-unit", runner_config="1-gpu-large")
# Nightly is not redundant here: it sets SGLANG_JI... | 85 | 2,680 |
sglang | test/registered/kernels/ops/layernorm/test_fused_qk_gemma_rmsnorm_gate.py | .py | import itertools
import sys
import pytest
import torch
from sglang.srt.models.utils import fused_qk_gemma_rmsnorm_with_gate
from sglang.test.ci.ci_register import register_amd_ci
register_amd_ci(est_time=20, stage="jit-kernel-unit", runner_config="amd")
def reference_qk_gemma_rmsnorm_with_gate(
q_gate: torch.T... | 152 | 4,981 |
sglang | test/registered/kernels/ops/layernorm/test_rmsnorm.py | .py | import itertools
import sys
import pytest
import torch
from sglang.kernels.jit.utils import get_ci_test_range
from sglang.srt.utils import is_hip
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")
# Nightly i... | 158 | 4,316 |
sglang | test/registered/kernels/ops/quantization/test_per_token_group_quant_8bit_v2.py | .py | import itertools
import pytest
import torch
from sglang.kernels.jit.utils import get_ci_test_range
from sglang.kernels.ops.quantization.per_token_group_quant_8bit_v2 import (
per_token_group_quant_8bit_v2,
)
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=90, stage="base-b-kerne... | 192 | 6,067 |
sglang | test/registered/kernels/ops/quantization/test_awq_marlin_repack.py | .py | import sys
import numpy as np
import pytest
import torch
from sgl_kernel.scalar_type import scalar_types
from sglang.kernels.ops.quantization.awq_marlin_repack import (
awq_marlin_repack as jit_awq_marlin_repack,
)
from sglang.srt.layers.quantization.utils import pack_cols, quantize_weights
from sglang.test.ci.ci... | 111 | 3,395 |
sglang | test/registered/kernels/ops/quantization/test_awq_marlin_moe_repack.py | .py | import sys
import numpy as np
import pytest
import torch
from sgl_kernel.scalar_type import scalar_types
from sglang.kernels.ops.quantization.awq_marlin_repack import (
awq_marlin_moe_repack as jit_awq_marlin_moe_repack,
)
from sglang.srt.layers.quantization.utils import pack_cols, quantize_weights
from sglang.te... | 125 | 4,006 |
sglang | test/registered/kernels/ops/quantization/test_per_tensor_quant_fp8.py | .py | import itertools
import sys
from typing import Optional, Tuple
import pytest
import torch
from sglang.kernels.jit.utils import get_ci_test_range
from sglang.kernels.ops.quantization.per_tensor_quant_fp8 import (
per_tensor_quant_fp8,
)
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_... | 104 | 3,266 |
sglang | test/registered/kernels/ops/quantization/test_per_token_quant_fp8.py | .py | import sys
import pytest
import torch
from sgl_kernel import sgl_per_token_quant_fp8 as aot_per_token_quant_fp8
from sglang.kernels.ops.quantization.fp8_kernel import scaled_fp8_quant
from sglang.kernels.ops.quantization.per_token_quant_fp8 import per_token_quant_fp8
from sglang.test.ci.ci_register import register_cu... | 121 | 4,960 |
sglang | test/registered/kernels/ops/quantization/test_per_token_group_quant.py | .py | """Correctness tests for the trait-driven per_token_group_quant JIT kernel.
The reference is computed in pure PyTorch (the quantization math itself), NOT by
calling the v2 / minimax kernels -- those are being deprecated, so the tests
must outlive them.
Two guard strengths, chosen by what the kernel's numerics can act... | 552 | 23,271 |
sglang | test/registered/kernels/ops/quantization/test_awq_dequantize.py | .py | import itertools
import sys
import pytest
import torch
from sglang.kernels.ops.quantization.awq_dequantize import (
awq_dequantize as jit_awq_dequantize,
)
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=9, stage="base-b-kernel-unit", runner_config="1-gpu-large")
try:
from ... | 171 | 4,758 |
sglang | test/registered/kernels/ops/quantization/test_gptq_marlin.py | .py | import sys
from types import SimpleNamespace
import pytest
import torch
from sgl_kernel.scalar_type import scalar_types
from sglang.kernels.ops.quantization.gptq_marlin import gptq_marlin_gemm
from sglang.srt.layers.quantization.marlin_utils import (
check_marlin_supported,
marlin_make_workspace,
)
from sglan... | 192 | 5,620 |
sglang | test/registered/kernels/ops/quantization/test_hadamard_jit.py | .py | import math
import sys
import numpy as np
import pytest
import torch
import torch.nn.functional as F
from scipy.linalg import hadamard
from sglang.kernels.ops.quantization.hadamard import (
hadamard_transform,
hadamard_transform_12n,
hadamard_transform_20n,
hadamard_transform_28n,
hadamard_transfo... | 428 | 13,467 |
sglang | test/registered/kernels/ops/quantization/test_gptq_marlin_repack.py | .py | import sys
import pytest
import torch
from sgl_kernel.scalar_type import scalar_types
from sglang.kernels.ops.quantization.gptq_marlin_repack import gptq_marlin_repack
from sglang.srt.layers.quantization.utils import (
gptq_quantize_weights,
pack_rows,
sort_weights,
)
from sglang.test.ci.ci_register impor... | 96 | 2,752 |
sglang | test/registered/kernels/ops/moe/test_moe_preprocess.py | .py | """fused_moe_preprocess must be bit-identical to the torch.sort-based path,
and the grouped GEMM must produce identical results under both block_size_m
configs (the block schedule and kernel config are chosen together).
"""
import pytest
import torch
from sglang.kernels.ops.moe.inkling_moe import (
SMALL_M_BLOCK_... | 97 | 3,251 |
sglang | test/registered/kernels/ops/moe/test_moe_lora_align_block_size.py | .py | # Temporarily adapted from https://github.com/vllm-project/vllm/blob/main/tests/lora/test_moe_lora_align_sum.py, will optimize in future refactor
import random
import sys
import pytest
import torch
# ---------------------------------------------------------
# IMPORT PREBUILT KERNEL
# ---------------------------------... | 168 | 6,376 |
sglang | test/registered/kernels/ops/moe/test_inkling_silu_and_mul.py | .py | """Numerics tests for the plain-Triton silu_and_mul (former helion kernels).
The kernels compute silu(gate) * up (* weight) in fp32 with one rounding cast
at the store, in the exact operation order of the helion kernels they
replaced, so bf16 outputs sit within 1 bf16 ulp of a same-order torch fp32
reference (tl.sigmo... | 160 | 5,975 |
sglang | test/registered/kernels/ops/moe/test_fused_topk_deepseek.py | .py | import sys
import pytest
import torch
from sglang.srt.layers.moe.topk import biased_grouped_topk_gpu, biased_grouped_topk_impl
from sglang.srt.utils import get_device
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=40, stage="nightly", runner_config="1-gpu-large")
@pytest.mark.par... | 102 | 3,610 |
sglang | test/registered/kernels/ops/moe/test_marlin_packed_topk_unpack.py | .py | """Precision test for the fused packed-topk unpack triton kernel used by the
Marlin MoE runner (fused-gate-topk support).
The FlashInfer / Inkling fused gate emits PackedTopKOutput -- int32
``(expert_id << 16) | bf16-weight-bits``. The Marlin runner reads topk_ids /
topk_weights separately, so it unpacks with a single... | 91 | 3,144 |
sglang | test/registered/kernels/ops/moe/test_moe_align_block_size.py | .py | import itertools
import sys
import pytest
import torch
import triton
import triton.language as tl
from sglang.kernels.jit.utils import get_ci_test_range
from sglang.kernels.ops.moe.moe_align import moe_align_block_size
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=28, stage="base-... | 364 | 11,329 |
sglang | test/registered/kernels/ops/moe/test_post_reorder_deepgemm.py | .py | import pytest
import torch
from sglang.kernels.ops.moe.ep_moe_kernels import (
post_reorder_deepgemm,
post_reorder_triton_kernel,
)
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=30, stage="base-b-kernel-unit", runner_config="1-gpu-large")
register_cuda_ci(est_time=30, stage... | 91 | 3,128 |
sglang | test/registered/kernels/ops/moe/test_minimax_m3_mxfp8.py | .py | # SPDX-License-Identifier: Apache-2.0
"""Reference-vs-fused unit tests for the MiniMax-M3 ROCm native MXFP8 ops.
Each fused kernel has a slow PyTorch / dequant-to-bf16 reference; these assert
the two agree within tolerance:
* Fused MXFP8 activation quant (Triton) -> torch reference
* Native MXFP8 linear (t... | 269 | 9,811 |
sglang | test/registered/kernels/ops/moe/test_moe_wna16_marlin.py | .py | import itertools
import sys
from types import SimpleNamespace
import pytest
import torch
from sgl_kernel.scalar_type import scalar_types
from sglang.kernels.ops.moe.moe_wna16_marlin import moe_wna16_marlin_gemm
from sglang.srt.layers.moe.fused_moe_triton import moe_align_block_size
from sglang.srt.layers.moe.fused_mo... | 689 | 20,639 |
sglang | test/registered/kernels/ops/moe/test_moe_align_small_numel.py | .py | """Correctness of the single-launch tiny-numel moe_align triton kernel.
The oracle is a plain-torch implementation of the documented contract, so it
does not depend on any other kernel's shape support; the AOT `sgl_kernel` path
is cross-checked on top of it to back the drop-in-replacement claim.
"""
import itertools
... | 215 | 8,461 |
sglang | test/registered/kernels/ops/moe/test_renorm.py | .py | # Adapted from https://github.com/flashinfer-ai/flashinfer/blob/main/tests/test_sampling.py
# and /sgl-workspace/sglang/python/sglang/kernels/aot/tests/test_sampling.py
import sys
import pytest
import torch
from sglang.srt.utils import is_hip
from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
... | 97 | 3,616 |
sglang | test/registered/kernels/ops/moe/test_minimax_quant_scatter.py | .py | import random
import sys
from contextlib import nullcontext
import pytest
import torch
import sglang.srt.layers.moe.moe_runner.deep_gemm as deep_gemm_runner
from sglang.kernels.ops.moe.ep_moe_kernels import (
fill_gateup_input_triton_kernel,
moe_ep_deepgemm_preprocess,
)
from sglang.kernels.ops.quantization.m... | 357 | 11,501 |
sglang | test/registered/kernels/ops/moe/test_sigmoid_gate_mul.py | .py | import sys
import pytest
import torch
from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
register_cuda_ci(est_time=4, stage="base-b-kernel-unit", runner_config="1-gpu-large")
register_amd_ci(est_time=4, stage="jit-kernel-unit", runner_config="amd")
DEVICE = "cuda"
def reference_sigmoid_gate_... | 82 | 2,541 |
sglang | test/registered/kernels/ops/moe/test_sigmoid_gate_mul_broadcast.py | .py | import sys
import pytest
import torch
from sglang.test.ci.ci_register import register_amd_ci
register_amd_ci(est_time=4, suite="jit-kernel-unit-test-amd")
DEVICE = "cuda"
def reference_sigmoid_gate_mul(x, gate):
return x * torch.sigmoid(gate)
# ── element-wise variant ──
@pytest.mark.parametrize("dtype", ... | 124 | 3,522 |
sglang | test/registered/kernels/ops/moe/test_fused_swiglu_epilogue.py | .py | """Correctness of the SwiGLU-in-the-up-GEMM-epilogue MoE fast path.
Two claims are invisible at the call site and would break silently under an
innocuous-looking rewrite:
1. Interleaving W13 rows leaves every up-GEMM output column unchanged -- the
permute only decides which column a gate/up pair lands in.
2. The e... | 208 | 7,951 |
sglang | test/registered/kernels/ops/moe/test_moe_topk_sigmoid.py | .py | """
Correctness tests for the moe_topk_sigmoid JIT kernel.
Validates against a pure-PyTorch reference and, when sgl_kernel is available,
cross-checks against the AOT implementation.
"""
import itertools
import os
import sys
from typing import Optional
import pytest
import torch
from sglang.kernels.ops.moe.moe_topk_... | 497 | 18,028 |
sglang | test/registered/kernels/ops/moe/test_fp4_moe.py | .py | # SPDX-License-Identifier: Apache-2.0
from typing import Callable
import pytest
import torch
from flashinfer import fp4_quantize, scaled_fp4_grouped_quantize
from flashinfer.fused_moe import cutlass_fused_moe as flashinfer_cutlass_fused_moe
from sgl_kernel import silu_and_mul
from torch.nn import functional as F
from... | 345 | 10,769 |
sglang | test/registered/kernels/ops/moe/test_moe_fused_gate.py | .py | """Correctness tests for the Triton :func:`moe_fused_gate` router.
The Triton kernel is a drop-in reimplementation of the CUDA fused gate for the
ungrouped case (``num_expert_group == 1``). We validate it three ways:
* against an explicit, definition-based torch reference (documents the math),
* against the CUDA JIT ... | 526 | 18,739 |
sglang | test/registered/kernels/ops/kimi_k3/test_collectives.py | .py | from __future__ import annotations
import atexit
import os
import pytest
import torch
import torch.distributed as dist
import sglang.srt.distributed.parallel_state as ps
from sglang.kernels.jit.utils import cache_once
from sglang.kernels.ops.communication.mp import register_comm_cleanup
from sglang.kernels.ops.kimi_... | 360 | 10,371 |
sglang | test/registered/kernels/ops/kimi_k3/test_compute.py | .py | import unittest
import torch
from sglang.kernels.ops.attention.fla.kda_replayssm_spec_decode import (
commit_kda_replayssm_spec,
)
from sglang.kernels.ops.kimi_k3 import (
situ_and_mul,
situ_and_mul_masked_post_quant,
)
from sglang.kernels.ops.kimi_k3.attn_res import attn_res_fused_tma
from sglang.kernels... | 454 | 14,598 |
sglang | test/registered/kernels/ops/kimi_k3/test_ar_fusion.py | .py | """Correctness test for the K3 MNNVL fused all-reduce (ar_fusion) kernels.
Compares the 1shot multicast-push and the in-place low-SM NVLS 2shot pull
(with and without the fused residual) against
NCCL, bit-exact on small-int bf16 inputs; the fused-RMSNorm pull against a
torch reference; the pull tuning knobs (num_block... | 452 | 17,299 |
sglang | test/registered/kernels/ops/communication/test_tp_qknorm.py | .py | from __future__ import annotations
import atexit
import itertools
import logging
import multiprocessing
import os
from multiprocessing.context import SpawnProcess
from typing import List
import pytest
import torch
import torch.distributed as dist
import triton
import sglang.srt.distributed.parallel_state as ps
from ... | 201 | 6,916 |
sglang | test/registered/kernels/ops/communication/test_amd_deterministic_custom_allreduce.py | .py | """
Test deterministic custom all-reduce kernel behavior with batch size invariance.
This test uses the 1-stage all-reduce kernel which is inherently deterministic
due to fixed accumulation ordering (each GPU reads all data from all GPUs and
reduces locally in a fixed order - no atomics, no race conditions).
Note: Th... | 255 | 8,335 |
sglang | test/registered/kernels/ops/communication/test_symm_mem_all_gather.py | .py | """Correctness test for the symmetric-memory multimem all-gather kernel.
Compares ``all_gather_inner`` (concat-along-hidden multimem.st gather) against
NCCL all-gather for a sweep of token counts, hidden widths, and the
``safe`` / ``skip_entry_sync`` knobs, in both eager and CUDA-graph modes.
Usage::
# Run on th... | 168 | 5,996 |
sglang | test/registered/kernels/ops/communication/test_custom_all_reduce.py | .py | """Correctness test for the JIT custom all-reduce (v2) kernel.
Compares the JIT custom all-reduce output against NCCL all-reduce for a sweep
of tensor sizes, dtypes, and algorithms, in both eager and CUDA-graph modes.
Usage::
# Run the test on the default world sizes (2, 4, 8 GPUs):
python tests/test_custom_... | 244 | 8,527 |
sglang | test/registered/kernels/ops/communication/test_amd_nccl_allreduce_determinism.py | .py | """
Test to confirm non-determinism of default NCCL all-reduce with batch size invariance.
This test uses the default torch.distributed.all_reduce (NCCL) which can be
NON-DETERMINISTIC due to tree-based reduction algorithms that don't guarantee
fixed accumulation order for bfloat16/float16.
This test compares:
1. Def... | 213 | 6,861 |
sglang | test/registered/kernels/ops/gemm/test_dsv3_router_gemm.py | .py | """Tests for JIT dsv3_router_gemm kernel."""
import itertools
import sys
import pytest
import torch
from sglang.kernels.jit.utils import (
get_ci_test_range,
get_jit_cuda_arch,
is_hip_runtime,
)
from sglang.kernels.ops.gemm.dsv3_router_gemm import dsv3_router_gemm
from sglang.test.ci.ci_register import r... | 69 | 2,143 |
sglang | test/registered/kernels/ops/gemm/test_minimax_fused_qkv_index_gemm.py | .py | import pytest
import torch
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=30, stage="base-b-kernel-unit", runner_config="4-gpu-b200")
dev = "cuda"
def _pack_weight_scale(scale_u8: torch.Tensor) -> torch.Tensor:
from sglang.srt.layers.deep_gemm_wrapper.configurer import DEEPGE... | 70 | 2,065 |
sglang | test/registered/kernels/ops/gemm/test_fp8_blockwise_gemm.py | .py | import sys
from typing import Optional, Type
import pytest
import torch
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 import register_cuda_ci
register_cuda_ci(
est_time=30,
stage="base-b",
run... | 91 | 3,190 |
sglang | test/registered/kernels/ops/gemm/test_cutedsl_bf16_gemm.py | .py | """Tests for the CuTe DSL TGV BF16 GEMM kernel."""
import sys
import pytest
import torch
from sglang.kernels.jit.utils import (
get_ci_test_range,
get_jit_cuda_arch,
is_hip_runtime,
)
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=30, stage="base-b-kernel-unit", runner... | 84 | 2,968 |
sglang | test/registered/kernels/ops/gemm/test_chunked_sgmv_cuda_graph.py | .py | # Copyright 2023-2026 SGLang Team
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writin... | 295 | 9,680 |
sglang | test/registered/kernels/ops/gemm/test_cutedsl_dsv3_fused_a_gemm.py | .py | """Tests for the CuTe DSL DeepSeek-V3 fused-A GEMM kernel."""
import sys
import pytest
import torch
from sglang.kernels.jit.utils import (
get_ci_test_range,
get_jit_cuda_arch,
is_hip_runtime,
)
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=30, stage="base-b-kernel-un... | 54 | 1,719 |
sglang | test/registered/kernels/ops/gemm/test_dsv3_fused_a_gemm.py | .py | """Tests for JIT dsv3_fused_a_gemm kernel."""
import sys
import pytest
import torch
import torch.nn.functional as F
from sglang.kernels.jit.utils import (
get_ci_test_range,
get_jit_cuda_arch,
is_hip_runtime,
)
from sglang.kernels.ops.gemm.dsv3_fused_a_gemm import dsv3_fused_a_gemm
from sglang.test.ci.ci... | 51 | 1,691 |
sglang | test/registered/kernels/ops/attention/test_c128_v2.py | .py | from __future__ import annotations
import sys
from typing import Tuple, Union
import pytest
import torch
import triton
from sglang.kernels.ops.attention.dsv4 import compress_forward
from sglang.srt.utils import get_device
from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
from sglang.test.kerne... | 280 | 9,212 |
sglang | test/registered/kernels/ops/attention/test_deepseek_v4_compress_plan_draft_pad.py | .py | """Kernel-level tests for the DSV4 compress write-plan (`plan_prefill`).
`plan_w` decides which tokens' raw KV get persisted into the compress-state ring
for a *future* compression window to read. A speculative verify batch plans from
the optimistic `seq_len = prefix + num_draft_tokens` but rolls back to
`prefix + acc... | 218 | 10,013 |
sglang | test/registered/kernels/ops/attention/test_cp_prefix_len_fa3_parity.py | .py | """
FA3 parity test for `prepare_context_parallel_metadata`.
Drives the real function and feeds its `kv_len_prev/next_tensor` into FA3
via `flash_attn_with_kvcache`. Compares per-rank CP output against a
full-sequence FA3 reference computed over the unpadded `(prefix + extend)`
KV. Any discrepancy indicates the metada... | 144 | 5,445 |
sglang | test/registered/kernels/ops/attention/test_mla_cp_fa3_parity.py | .py | """FA3 numerical parity for MLA prefill CP.
Verifies the rank-local zigzag-split FA3 path (``_mla_cp_attn`` +
``cp_attn_forward_extend`` in ``flashattention_backend.py``) matches a
single non-CP ``flash_attn_with_kvcache`` over the full sequence.
Single-process, single-layer, pre-populated paged KV cache. Requires
FA... | 211 | 6,690 |
sglang | test/registered/kernels/ops/attention/test_fused_verify_triton_gdn.py | .py | """Tests for fused sigmoid gating delta rule MTP kernel (GDN target_verify).
Compares the fused kernel `fused_sigmoid_gating_delta_rule_update` against
the reference two-step implementation:
1. g, beta = fused_gdn_gating(A_log, a, b, dt_bias)
2. o = fused_recurrent_gated_delta_rule_update(q, k, v, g, beta, ...... | 239 | 6,995 |
sglang | test/registered/kernels/ops/attention/test_dsa_metadata.py | .py | import unittest
import torch
from sglang.kernels.ops.attention.dsa_metadata import (
fused_dsa_decode_metadata,
fused_dsa_draft_extend_metadata,
fused_dsa_target_verify_metadata,
)
from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
from sglang.test.test_utils import CustomTestCase
r... | 528 | 20,176 |
sglang | test/registered/kernels/ops/attention/test_flash_attention_3_only_qv.py | .py | # Adapted from sgl-flash-attn hopper/test_attn_kvcache.py::test_flash_attn_kvcache_only_qv
# Covers the only_qv (NoPE) decode path that FA3 adds for sparse MLA on SM90:
# the QK^T matmul is skipped and attention is computed as softmax(qv * V) over
# a paged V cache (no K cache, no rope).
import math
import sys
import ... | 101 | 3,346 |
sglang | test/registered/kernels/ops/attention/test_minimax_decode_topk.py | .py | """Correctness tests for the MiniMax-M3 single-stage radix-select decode topk.
The kernel selects, per (head, batch) row, the indices of the ``topk`` largest
block scores among the row's first ``num_blocks = ceil(seq_len / block_size)``
entries, front-packing valid block ids and ``-1``-padding the tail. This mirrors
t... | 146 | 5,875 |
sglang | test/registered/kernels/ops/attention/test_kda_helion.py | .py | from __future__ import annotations
import inspect
import sys
import pytest
import torch
from sglang.kernels.ops.attention.fla.fused_recurrent import (
fused_recurrent_kda_packed_decode,
)
from sglang.kernels.ops.attention.fla.fused_recurrent_linear_replayssm import (
fused_recurrent_linear_replayssm_decode,
... | 1,027 | 33,854 |
sglang | test/registered/kernels/ops/attention/test_topk_v2.py | .py | """Correctness tests for the DeepSeek-V4 (DSA indexer) JIT top-k transform v2.
The v2 kernel selects the per-row top-k of ``scores`` (ragged ``seq_lens``) and
writes the page-table transform of the selected raw indices into the output. We
validate against ``torch.topk`` with a small tolerance for boundary ties (the
fp... | 300 | 12,345 |
sglang | test/registered/kernels/ops/attention/test_paged_mqa_metadata.py | .py | """Unit tests for paged_mqa_metadata JIT kernel.
Verifies byte-equal correctness against a pure-PyTorch reference oracle
across the shape envelope. Output is int32 ``[num_sm + 1, 2]`` — a
deterministic partition table — so equality is strict (``torch.equal``,
no atol/rtol).
Test groups:
1. ``test_matches_pytorch_ref... | 278 | 10,823 |
sglang | test/registered/kernels/ops/attention/test_concat_mla.py | .py | import itertools
import sys
import pytest
import torch
import triton
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=17, stage="base-b-kernel-unit", runner_config="1-gpu-large")
def torch_concat_mla_k(
k: torch.Tensor, k_nope: torch.Tensor, k_rope: torch.Tensor
) -> None:
... | 175 | 5,642 |
sglang | test/registered/kernels/ops/attention/test_cute_dsl_fp8_paged_mqa_logits.py | .py | # SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: Apache-2.0
import sys
import pytest
import torch
from sglang.kernels.ops.attention.dsa import cutedsl_paged_mqa_logits, pick_dsl_expand
from sglang.srt.layers.attention.dsa.utils import (
... | 235 | 7,607 |
sglang | test/registered/kernels/ops/attention/test_dsv4_indexer_quant.py | .py | """Correctness tests for the DeepSeek-V4 DSA indexer fp8-quant Q kernel and its
V3.2/GLM rope-first variant, after the grid-stride + occupancy scheduling
optimization of ``fused_q_indexer_rope_hadamard_quant``.
Covers both template configs that share the kernel:
- fused_q_indexer_rope_hadamard_quant (V4: rope on tr... | 190 | 7,740 |
sglang | test/registered/kernels/ops/attention/test_minimax_qknorm_rope.py | .py | """Correctness for the fused MiniMax-M3 Gemma-RMSNorm + partial NeoX RoPE kernel.
Verifies the in-place fused kernel reproduces GemmaRMSNorm((1+w)) + partial NeoX
RoPE to the bf16 round-off floor, leaves V untouched, and matches sglang's RoPE
convention (cos|sin cache, neox pairs (i, i+rotary_dim/2)).
"""
import pyte... | 179 | 7,031 |
sglang | test/registered/kernels/ops/attention/test_sparse_mla_q8kv8_prefill_sm90.py | .py | from __future__ import annotations
import math
import sys
import pytest
import torch
from sglang.srt.utils import is_sm90_supported
from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(est_time=240, stage="base-b-kernel-unit", runner_config="1-gpu-large")
DTYPE_FP8 = torch.float8_e4m3fn
D_V = ... | 725 | 26,953 |
sglang | test/registered/kernels/ops/attention/test_cutedsl_gdn.py | .py | """Tests for CuTe DSL fused sigmoid gating delta rule kernel (GDN)."""
import sys
import numpy as np
import pytest
import torch
from sglang.test.ci.ci_register import register_cuda_ci
try:
import cuda.bindings.driver as cuda_driver
import cutlass # noqa: F401
from cutlass.cute.runtime import from_dlpac... | 312 | 10,680 |
sglang | test/registered/kernels/ops/attention/test_kda_fused_decode.py | .py | """Kimi-K3 fused KDA decode must match the existing unfused decode chain.
The fused kernel replaces:
causal_conv1d_update -> kda_packed_decode -> sigmoid-gated RMSNorm
This file covers the local head layouts used by Kimi-K3 TP8/TP16/TP32:
H = 12/6/3. The H=6 and H=3 cases are the branches added by the fixed-head... | 216 | 5,662 |
sglang | test/registered/kernels/ops/attention/test_dsa_transform_index.py | .py | import unittest
from unittest.mock import patch
import torch
import sglang.kernels.ops.attention.dsa.transform_index as transform_index_module
from sglang.kernels.ops.attention.dsa.transform_index import (
transform_index_page_table_decode_fast,
transform_index_page_table_prefill_fast,
)
from sglang.test.ci.c... | 241 | 8,069 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.