torch-sys-plus 0.18.2

Low-level FFI bindings for the PyTorch C++ api (libtorch).
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30

use libc::c_int;

extern "C" {
    /// Returns the number of CUDA devices available.
    pub fn atc_cuda_device_count() -> c_int;

    /// Returns true if at least one CUDA device is available.
    pub fn atc_cuda_is_available() -> c_int;

    /// Returns true if CUDA is available, and CuDNN is available.
    pub fn atc_cudnn_is_available() -> c_int;

    /// Sets the seed for the current GPU.
    pub fn atc_manual_seed(seed: u64);

    /// Sets the seed for all available GPUs.
    pub fn atc_manual_seed_all(seed: u64);

    /// Waits for all kernels in all streams on a CUDA device to complete.
    pub fn atc_synchronize(device_index: i64);

    /// Returns true if CUDNN is enabled by the user.
    pub fn atc_user_enabled_cudnn() -> c_int;

    /// Enable or disable CUDNN.
    pub fn atc_set_user_enabled_cudnn(b: c_int);

    /// Sets CUDNN benchmark mode.
    pub fn atc_set_benchmark_cudnn(b: c_int);
}