Install

Run the following Vcpkg command to install the port.

vcpkg install libtorch

Usage

Usage details are not available for this port.

Features

libtorch[blas]≥v2.12.0#0
All
BLAS/LAPACK backend (openblas on linux/windows, Accelerate on macOS). On by default; disable with [core,...] when using [mkl] instead.
Default Feature
- All
Feature Dependencies
- blas
  All
- lapack
  All
libtorch[cuda]≥v2.1.2#0
(windows & x64 & !staticcrt) | (linux & x64)
Build with CUDA GPU backend
Feature Dependencies
- cuda
  All
- cudnn
  All
- cudnn-frontend
  All
- gloo
  linux
  - gloo[cuda]
    All
- magma
  All
- nccl
  linux
- nvidia-cutlass
  All
- tensorpipe
  linux
  - tensorpipe[cuda]
    All
libtorch[dist]≥v1.12.1#0
(linux & x64) | windows
Use distributed training/inference (Gloo, MPI, libuv, TensorPipe)
Feature Dependencies
- gloo
  linux
- libuv
  windows | osx
- mpi
  linux
- tensorpipe
  linux
libtorch[gflags]≥v2.7.1#0
All
Build with gflags
Feature Dependencies
- gflags
  All
libtorch[glog]≥v2.7.1#0
All
Build with glog
Feature Dependencies
- glog
  All
libtorch[llvm]≥v2.1.2#0
All
Build with LLVM
Feature Dependencies
- llvm
  All
libtorch[mkl]≥v2.12.0#0
(windows | linux) & x64 & !uwp
Intel performance backend: use oneMKL for BLAS/LAPACK and enable oneDNN (MKLDNN) CPU acceleration via ideep. Mutually exclusive with [blas] (the openblas default-feature) — install as libtorch[core,mkl,...] to skip the openblas backend, otherwise both backends get built and only MKL is linked.
Feature Dependencies
- ideep
  All
- intel-mkl
  All
- onednn
  All
libtorch[vulkan]≥v1.12.1#0
All
Build with Vulkan GPU backend
Feature Dependencies
- shaderc
  All
- vulkan
  All
- vulkan-loader
  All
- vulkan-memory-allocator
  All