use ruda_kernel::dsl as kernel_dsl;
use ruda_kernel::dsl::Runtime;
use ruda_kernel::tensor::layout::address_type;
use ruda_kernel::tensor::layout::max_vector_size;
use ruda_kernel::tensor::allocation::empty_device_dtype;
use ruda_kernel::tensor::RudaTensor;
use ruda_core::tensor::TensorMetadata;
use ruda_kernel::dsl::calculate_ruda_count_elemwise;
use ruda_kernel::dsl::prelude::*;
use ruda_kernel::library::tensor::layout::linear::LinearView;
pub trait IntUnaryOpFamily: 'static + Send + Sync {
type Options: LaunchArg;
type Unary<I: Int, N: Size>: IntUnaryOp<I, N, Options = Self::Options>;
}
#[ruda]
pub trait IntUnaryOp<I: Scalar, N: Size>: 'static + Send + Sync {
type Options: LaunchArg;
fn execute(input: Vector<I, N>, options: &Self::Options) -> Vector<I, N>;
}
#[ruda(launch_unchecked, address_type = "dynamic")]
pub fn unary_int<I: Int, N: Size, O: IntUnaryOpFamily>(
input: &LinearView<Vector<I, N>>,
output: &mut LinearView<Vector<I, N>, ReadWrite>,
options: &O::Options,
#[define(I)] _dtype: StorageType,
) {
if !output.is_in_bounds(ABSOLUTE_POS) {
terminate!();
}
output[ABSOLUTE_POS] = O::Unary::<I, N>::execute(input[ABSOLUTE_POS], options);
}
pub fn launch_unary_int<R, O, Args>(tensor: RudaTensor<R>, args: Args) -> RudaTensor<R>
where
for<'a> Args: FnOnce(&'a ()) -> RuntimeArg<O::Options, R>,
R: Runtime,
O: IntUnaryOpFamily,
{
let vector_size = max_vector_size(&tensor);
let client = tensor.client.clone();
let num_elems = tensor.meta.num_elements();
let working_units = num_elems / vector_size as usize;
let ruda_dim = RudaDim::new(tensor.client.properties(), working_units);
let ruda_count = calculate_ruda_count_elemwise(&tensor.client, working_units, ruda_dim);
let dtype = tensor.dtype;
unsafe {
if tensor.can_mut() && tensor.is_nonoverlapping() {
unary_int::launch_unchecked::<O, R>(
&client,
ruda_count,
ruda_dim,
address_type!(tensor),
vector_size,
tensor.clone().into_linear_view(),
tensor.as_linear_view_alias(0),
args(&()),
dtype.into(),
);
tensor
} else {
let output = empty_device_dtype(
tensor.client.clone(),
tensor.device.clone(),
tensor.shape(),
tensor.dtype,
);
unary_int::launch_unchecked::<O, R>(
&client,
ruda_count,
ruda_dim,
address_type!(tensor, output),
vector_size,
tensor.into_linear_view(),
output.clone().into_linear_view(),
args(&()),
dtype.into(),
);
output
}
}
}
pub mod unary_basic_int {
use ruda_kernel::dsl::num_traits::One;
use ruda_kernel::dsl::num_traits::Zero;
use super::*;
pub fn launch<R, Args>(tensor: RudaTensor<R>, args: Args) -> RudaTensor<R>
where
R: Runtime,
for<'a> Args: FnOnce(&'a ()) -> BasicIntUnaryKind,
{
launch_unary_int::<R, BasicIntUnary, _>(tensor, |input| {
BasicIntUnaryOptionsLaunch::new(args(input))
})
}
#[derive(Clone, Copy, Debug, Hash, Eq, PartialEq, serde::Serialize, serde::Deserialize)]
pub enum BasicIntUnaryKind {
BitwiseNot,
Sign,
}
#[derive(RudaLaunch, RudaType)]
struct BasicIntUnaryOptions {
#[ruda(comptime)]
kind: BasicIntUnaryKind,
}
struct BasicIntUnary;
#[ruda]
impl<I: Int, N: Size> IntUnaryOp<I, N> for BasicIntUnary {
type Options = BasicIntUnaryOptions;
fn execute(input: Vector<I, N>, options: &Self::Options) -> Vector<I, N> {
match comptime![options.kind] {
BasicIntUnaryKind::BitwiseNot => !input,
BasicIntUnaryKind::Sign => {
let zero = Vector::zero();
let one = Vector::one();
let minus_one = Vector::new(I::new(-1));
let is_positive = input.greater_than(zero);
let is_negative = input.less_than(zero);
let sign = select_many(is_negative, minus_one, zero);
select_many(is_positive, one, sign)
}
}
}
}
impl IntUnaryOpFamily for BasicIntUnary {
type Options = BasicIntUnaryOptions;
type Unary<I: Int, N: Size> = Self;
}
}