Gentoo Packages

dev-python/comfy-kitchen - 0.2.14 (zGentoo)

Search

Install

Install this version:

emerge -a =dev-python/comfy-kitchen-0.2.14

If this version is masked, you can unmask it using the autounmask tool or standard emerge options:

autounmask =dev-python/comfy-kitchen-0.2.14

Or alternatively:

emerge --autounmask-write -a =dev-python/comfy-kitchen-0.2.14

Package Information

Description:
Fast kernel library for Diffusion inference with multiple compute backends
Homepage:
https://github.com/Comfy-Org/comfy-kitchen
License:
Apache-2.0

Ebuild Details

Version EAPI Keywords Slot
0.2.14 8 ~amd64 0
View Raw Ebuild 
# Copyright 2026 Gentoo Authors
# Distributed under the terms of the GNU General Public License v2

EAPI=8

DISTUTILS_USE_PEP517=setuptools
DISTUTILS_SINGLE_IMPL=1
PYTHON_COMPAT=( python3_{12..14} )

inherit distutils-r1

DESCRIPTION="Fast kernel library for Diffusion inference with multiple compute backends"
HOMEPAGE="https://github.com/Comfy-Org/comfy-kitchen"
SRC_URI="https://github.com/Comfy-Org/${PN}/archive/refs/tags/v${PV}.tar.gz -> ${P}.gh.tar.gz"

S="${WORKDIR}/${P}"

LICENSE="Apache-2.0"
SLOT="0"
KEYWORDS="~amd64"

# No CUDA USE flag — we always build CPU-only for Gentoo.
# The triton and eager backends work for both CUDA and ROCm at runtime.
# The compiled CUDA extension requires CUDA 12.8+ and nanobind and is not
# needed for standard ComfyUI operation.

RDEPEND="
	|| (
		sci-ml/pytorch[${PYTHON_SINGLE_USEDEP}]
		sci-ml/pytorch-rocm[${PYTHON_SINGLE_USEDEP}]
		sci-ml/pytorch-cuda[${PYTHON_SINGLE_USEDEP}]
	)
"

PATCHES=(
	"${FILESDIR}/${P}-optional-cuda.patch"
)

python_compile() {
	COMFY_KITCHEN_BUILD_NO_CUDA=1 distutils-r1_python_compile
}

Inherited Eclasses

distutils-r1

Dependencies

RDEPEND

	|| (
		sci-ml/pytorch[]
		sci-ml/pytorch-rocm[]
		sci-ml/pytorch-cuda[]
	)

Manifest for 0.2.14

Type File Size Source URLs
DIST comfy-kitchen-0.2.14.gh.tar.gz 182159 bytes https://github.com/Comfy-Org/comfy-kitchen/archive/refs/tags/v0.2.14.tar.gz