Package python3-module-torch-cuda: Information
Source package: python3-module-torch-cuda
Version: 2.10.0-alt2
Build time: Mar 2, 2026, 08:58 PM in the task #409614
Category: Development/ML
Report package bugHome page: https://pytorch.org/
License: BSD-3-Clause
Summary: Tensors and dynamic neural networks in Python with strong acceleration support (with CUDA support)
Description:
Tensors and dynamic neural networks in Python with strong acceleration support (with CUDA support). PyTorch is an optimized tensor library for deep learning using GPUs and CPUs.
List of RPM packages built from this SRPM:
libtorch-cuda (x86_64, aarch64)
libtorch-cuda-cpu (x86_64, aarch64)
python3-module-torch-cuda (x86_64, aarch64)
python3-module-torch-cuda-devel (x86_64, aarch64)
libtorch-cuda (x86_64, aarch64)
libtorch-cuda-cpu (x86_64, aarch64)
python3-module-torch-cuda (x86_64, aarch64)
python3-module-torch-cuda-devel (x86_64, aarch64)
Maintainer: Nikita Shmatko
Last changed
March 2, 2026 Nikita Shmatko 2.10.0-alt2
- Built torch-cuda on aarch64. - Vendored mimalloc for better aarch64 memory allocation.
Feb. 24, 2026 Nikita Shmatko 2.10.0-alt1
- Updated to 2.10.0 version. - Used bundled pybind11 insted of system. - Vendored cutlass files for examples used in build. - Minor specfile fixes.
Feb. 13, 2026 Nikita Shmatko 2.9.1-alt1
- Updated to version 2.9.1. - Enabled building with system-provided NVTX. - Dropped copying of valgrind headers, used system ones. - Updated package descriptions for libtorch and libtorch-cuda for clarity. - Relocated heavy .so files and their runtime deps to /usr/lib64, added symlinks in site-packages. - Renamed package to python3-module-torch-cuda and subpackages libtorch and libtorch-cuda to libtorch-cuda-cpu and libtorch-cuda.