cubecl-runtime 0.11.0-pre.4

The runtime API of CubeCL: clients, devices, handles and everything a kernel launch needs.
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

use cubecl_common::device::{Device, DeviceId};

/// A device of the CUDA runtime, named by its index.
#[derive(new, Clone, PartialEq, Eq, Default, Hash)]
pub struct CudaDevice {
    /// The index of the GPU among the ones CUDA reports.
    pub index: usize,
}

impl core::fmt::Debug for CudaDevice {
    fn fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result {
        write!(f, "Cuda({})", self.index)
    }
}

impl Device for CudaDevice {
    fn from_id(device_id: DeviceId) -> Self {
        Self {
            index: device_id.index_id as usize,
        }
    }

    fn to_id(&self) -> DeviceId {
        DeviceId {
            type_id: 0,
            index_id: super::index_id(self.index),
        }
    }
}