Skip to main content

cum_sum_forward

teeny_kernels::nn::tensor::reduction

Function cum_sum_forward 

Source 
pub fn cum_sum_forward<T: Triton, D: Num, const BLOCK_INNER: i32>(
    x_ptr: T::Pointer<D>,
    y_ptr: T::Pointer<D>,
    n_inner: i32,
    n_outer: i32,
)
where
    T::I32Tensor: Tensor<i32, 1> + Comparison<i32, BoolTensor = T::BoolTensor>,
    T::Pointer<D>: AddOffsets<i32, 1, T::I32Tensor, Output = T::Tensor<T::Pointer<D>>>,
Expand description

Forward: y = cumsum(x, axis=0) over a 1-D block Each CTA handles one complete row (n_inner elements).