numr 0.5.2

High-performance numerical computing with multi-backend GPU acceleration (CPU/CUDA/WebGPU)
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17

//! CUDA implementation of einsum operations.

use crate::error::Result;
use crate::ops::EinsumOps;
use crate::ops::impl_generic::einsum::einsum_impl;
use crate::runtime::cuda::{CudaClient, CudaRuntime};
use crate::tensor::Tensor;

impl EinsumOps<CudaRuntime> for CudaClient {
    fn einsum(
        &self,
        notation: &str,
        inputs: &[&Tensor<CudaRuntime>],
    ) -> Result<Tensor<CudaRuntime>> {
        einsum_impl(self, notation, inputs)
    }
}