# Simplified RPM spec file for vLLM - CPU-only initial package %global srcname vllm Name: python-%{srcname} Version: 0.10.0 Release: 1%{?dist} Summary: High-throughput and memory-efficient LLM inference engine (CPU-only) License: Apache-2.0 URL: https://github.com/vllm-project/vllm Source0: https://github.com/vllm-project/vllm/archive/v%{version}.tar.gz#/%{srcname}-%{version}.tar.gz BuildRequires: python3-devel BuildRequires: python3-setuptools >= 77.0.3 BuildRequires: python3-setuptools-scm >= 8.0 BuildRequires: python3-wheel BuildRequires: cmake >= 3.26.1 BuildRequires: ninja-build BuildRequires: gcc-c++ BuildRequires: python3-torch BuildRequires: python3-jinja2 BuildRequires: python3-packaging BuildRequires: python3-regex %description vLLM is a fast and easy-to-use library for LLM inference and serving. This package provides CPU-only inference capabilities. %package -n python3-%{srcname} Summary: %{summary} Requires: python3-torch Requires: python3-numpy Requires: python3-requests %description -n python3-%{srcname} vLLM is a fast and easy-to-use library for LLM inference and serving. %prep %autosetup -n %{srcname}-%{version} %build export VLLM_TARGET_DEVICE=cpu %py3_build %install export VLLM_TARGET_DEVICE=cpu %py3_install %files -n python3-%{srcname} %license LICENSE %doc README.md %{python3_sitelib}/%{srcname}/ %{python3_sitelib}/%{srcname}-*.dist-info/ %{_bindir}/vllm %changelog * Mon Jan 27 2025 Fedora Packager - 0.10.0-1 - Initial simplified package for Fedora