pyproject.toml

[project]
name = "nvidia-cudnn-frontend"
dynamic = ["version"]
description = "NVIDIA cuDNN Frontend — Python and C++ Graph API with SOTA attention (SDPA / Flash Attention), MoE grouped GEMM fusions, and FP8/MXFP8 kernels for Hopper, Blackwell, and Rubin GPUs."
readme = "README.md"
requires-python = ">=3.10"
license = {text = "Apache-2.0 AND MIT"}
keywords = [
    "cudnn",
    "cuda",
    "gpu",
    "nvidia",
    "deep-learning",
    "attention",
    "sdpa",
    "flash-attention",
    "transformer",
    "moe",
    "mixture-of-experts",
    "grouped-gemm",
    "fp8",
    "mxfp8",
    "blackwell",
    "rubin",
    "hopper",
    "pytorch",
    "kernel",
    "graph-api",
]
classifiers = [
    "Development Status :: 5 - Production/Stable",
    "Intended Audience :: Developers",
    "Intended Audience :: Science/Research",
    "License :: OSI Approved :: Apache Software License",
    "License :: OSI Approved :: MIT License",
    "Operating System :: POSIX :: Linux",
    "Operating System :: Microsoft :: Windows",
    "Programming Language :: C++",
    "Programming Language :: Python :: 3",
    "Programming Language :: Python :: 3.10",
    "Programming Language :: Python :: 3.11",
    "Programming Language :: Python :: 3.12",
    "Programming Language :: Python :: 3.13",
    "Programming Language :: Python :: 3.14",
    "Environment :: GPU :: NVIDIA CUDA",
    "Topic :: Scientific/Engineering :: Artificial Intelligence",
    "Topic :: Software Development :: Libraries :: Python Modules",
]

dependencies = [
    "nvidia-cutlass-dsl[cu13]>=4.6.2",
    "apache-tvm-ffi>=0.1.11",
]

[project.urls]
"Homepage" = "https://github.com/NVIDIA/cudnn-frontend"
"Documentation" = "https://docs.nvidia.com/deeplearning/cudnn/frontend/latest/"
"Blog" = "https://nvidia.github.io/cudnn-frontend/"
"Repository" = "https://github.com/NVIDIA/cudnn-frontend"
"Bug Tracker" = "https://github.com/NVIDIA/cudnn-frontend/issues"
"Release Notes" = "https://github.com/NVIDIA/cudnn-frontend/releases"

[project.optional-dependencies]
cutedsl = [
    # nvidia-cutlass-dsl and apache-tvm-ffi moved to the required [project]
    # dependencies above; this extra is what is left of the old CuTeDSL group.
    # Naming cuda-python here is belt-and-braces rather than load-bearing: the
    # required nvidia-cutlass-dsl pulls nvidia-cutlass-dsl-libs-base, which itself
    # requires cuda-python>=12.8, so a plain `pip install nvidia-cudnn-frontend`
    # already has it. Keeping it spelled out pins the dependency to this package
    # instead of to a transitive detail of the DSL's packaging, and keeps
    # `pip install nvidia-cudnn-frontend[cutedsl]` resolving for everyone who
    # still writes it that way.
    "cuda-python",
]
comm = [
    # Communication runtimes are grouped by backend so future distributed
    # operation graphs can reuse the same installation extra.
    "nvshmem4py-cu13>=0.3.1",
]
cutile = [
    # The cuTile linear-attention engines. Base cuda-tile only -- its [tileiras]
    # extra pins cuda-toolkit>=13.2,<13.4, and that upper bound would cap the
    # whole environment's toolkit and shut out CUDA 12 entirely. Without it,
    # cuda.tile falls back to a system `tileiras`, the same way this package
    # already leaves GPU wheels to the user. Engines that cannot import the
    # runtime decline in check_support, so a missing compiler costs those
    # engines and nothing else.
    "cuda-tile>=1.4",
]
triton = [
    # Framework-specific Triton FE OSS kernels. Torch remains in the explicit
    # torch dependency group so callers can choose the matching CUDA wheel.
    "triton>=3.7.0; sys_platform == 'linux' and python_version >= '3.10'",
]

[dependency-groups]
dev = [
    "jupyter",
    "numpy",
    "pybind11[global]>=2.13,<3",
    "pytest",
    "pytest-xdist",
    "cuda-python",
    "looseversion",
    "black==26.3.1",
    "clang-format==21.1.6",
]
# Per-framework groups for the type-erased CuTeDSL GEMM APIs (JAX is supported
# by the dense fusions: amax, swiglu, srelu, dsrelu): install with
# `pip install --group torch` / `pip install --group jax`. GPU wheels
# (torch cuXX, jax[cuda12]/[cuda13]) are left to the user.
torch = [
    "torch",
    "torch-c-dlpack-ext",
]
# The jax.jit-compatible XLA custom-call entry points (cudnn.jax, e.g.
# gemm_amax_jax_sm100) build on the CuTeDSL JAX extensions (cutlass.jax,
# shipped with nvidia-cutlass-dsl), which require jax >= 0.5.
jax = [
    "jax>=0.5",
]

[build-system]
requires = ["setuptools>=64", "cmake>=3.18", "ninja==1.11.1.1", "pybind11[global]>=2.13,<3"]
build-backend = "setuptools.build_meta"

[tool.setuptools]
packages = {find = {where = ["python", "."], include = ["cudnn*", "include"], namespaces = true}}
package-dir = {"" = "python", "include" = "include"}
include-package-data = true

[tool.setuptools.dynamic]
version = {attr = "cudnn.__version__"}

[tool.setuptools.package-data]
include = ["**/*"]
"cudnn.moe_ep._megamoe_backend.cutedsl_src" = [
    "LICENSE.Apache-2.0",
    "VENDOR.md",
]
"cudnn.linear_attention.cake.kernels" = ["*.cu", "SHA256SUMS", "UPSTREAM.md"]