launch_ref

cubecl_linalg::matmul::kernels::naive

Function launch_ref 

Source 
pub fn launch_ref<R: Runtime, E: Numeric>(
    client: &ComputeClient<R::Server, R::Channel>,
    lhs: &TensorHandleRef<'_, R>,
    rhs: &TensorHandleRef<'_, R>,
    out: &TensorHandleRef<'_, R>,
) -> Result<(), MatmulLaunchError>
Expand description

Matrix multiplication using memory coalescing algorithm with custom cube dimensions