Gentoo Packages

Search

Install

Install this version:

emerge -a =sci-ml/llama-cpp-0.6644

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

autounmask =sci-ml/llama-cpp-0.6644

Or alternatively:

emerge --autounmask-write -a =sci-ml/llama-cpp-0.6644

Package Information

Description:: Port of Facebook's LLaMA model in C/C++
Homepage:: https://github.com/ggml-org/llama.cpp
License:: MIT

Ebuild Details

Version	EAPI	Keywords	Slot
0.6644	8	~amd64	0

View Raw Ebuild

# Copyright 2025 Gentoo Authors
# Distributed under the terms of the GNU General Public License v2

EAPI=8

ROCM_VERSION="6.3"

inherit cmake rocm

BUILD_NUMBER=6644

DESCRIPTION="Port of Facebook's LLaMA model in C/C++"
HOMEPAGE="https://github.com/ggml-org/llama.cpp"
SRC_URI="https://github.com/ggml-org/llama.cpp/archive/refs/tags/b${BUILD_NUMBER}.tar.gz -> ${P}.tar.gz"

LICENSE="MIT"
SLOT="0"
IUSE="+curl +server"
KEYWORDS="~amd64"

# curl is needed for pulling models from huggingface
# numpy is used by convert_hf_to_gguf.py
CDEPEND="
	curl? ( net-misc/curl:= )
"
DEPEND="${CDEPEND}
	sci-ml/ggml:=
"
RDEPEND="${CDEPEND}
	dev-python/numpy
"

S=${WORKDIR}/llama.cpp-b${BUILD_NUMBER}

src_configure() {
	local mycmakeargs=(
		-DLLAMA_BUILD_TESTS=OFF
		-DLLAMA_BUILD_SERVER=$(usex server ON OFF)
		-DCMAKE_SKIP_BUILD_RPATH=ON
		-DLLAMA_CURL=$(usex curl ON OFF)
		-DBUILD_NUMBER=${BUILD_NUMBER}
		-DLLAMA_USE_SYSTEM_GGML=ON
	)

	cmake_src_configure
}

USE Flags

Manage flags for this package: euse -i <flag> -p sci-ml/llama-cpp | euse -E <flag> -p sci-ml/llama-cpp | euse -D <flag> -p sci-ml/llama-cpp

Global/Standard Flags

curl

server

Inherited Eclasses

cmake

rocm

Dependencies

DEPEND

	curl? ( net-misc/curl:= )

	sci-ml/ggml:=

RDEPEND

	curl? ( net-misc/curl:= )

	dev-python/numpy

Manifest for 0.6644

Type	File	Size	Source URLs
DIST	llama-cpp-0.6644.tar.gz	25868139 bytes	https://github.com/ggml-org/llama.cpp/archive/refs/tags/b6644.tar.gz

sci-ml/llama-cpp - 0.6644 (phackerlay)