ruda-tensor-dispatch 0.2.4

Runtime backend dispatch for RUDA tensors.
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
31
32
33
34
35
36
37
38
39

use ruda_tensor::graph::{BackendIr, HandleKind, TensorHandle};
use crate::{Dispatch, DispatchTensor};

impl BackendIr for Dispatch {
    type Handle = HandleKind<Self>;

    fn float_tensor(handle: TensorHandle<Self::Handle>) -> DispatchTensor {
        match handle.handle {
            HandleKind::Float(tensor) => tensor,
            other => panic!("Expected float handle, got {}", other.name()),
        }
    }

    fn int_tensor(handle: TensorHandle<Self::Handle>) -> DispatchTensor {
        match handle.handle {
            HandleKind::Int(tensor) => tensor,
            other => panic!("Expected integer handle, got {}", other.name()),
        }
    }

    fn bool_tensor(handle: TensorHandle<Self::Handle>) -> DispatchTensor {
        match handle.handle {
            HandleKind::Bool(tensor) => tensor,
            other => panic!("Expected boolean handle, got {}", other.name()),
        }
    }

    fn quantized_tensor(handle: TensorHandle<Self::Handle>) -> DispatchTensor {
        match handle.handle {
            HandleKind::Quantized(tensor) => tensor,
            other => panic!("Expected quantized handle, got {}", other.name()),
        }
    }

    fn float_tensor_handle(tensor: DispatchTensor) -> Self::Handle { HandleKind::Float(tensor) }
    fn int_tensor_handle(tensor: DispatchTensor) -> Self::Handle { HandleKind::Int(tensor) }
    fn bool_tensor_handle(tensor: DispatchTensor) -> Self::Handle { HandleKind::Bool(tensor) }
    fn quantized_tensor_handle(tensor: DispatchTensor) -> Self::Handle { HandleKind::Quantized(tensor) }
}