Skip to content

Support implicit promoting with f16 #321

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Feb 26, 2023
Merged

Support implicit promoting with f16 #321

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
5d3aad7
Support implicit promoting with f16
Benjamin-Philip Jun 8, 2022
f0c63ac
Additional changes for f16 type
umar456 Jul 13, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 3 additions & 0 deletions src/core/arith.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,6 +4,8 @@ use super::defines::AfError;
use super::dim4::Dim4;
use super::error::HANDLE_ERROR;
use super::util::{af_array, HasAfEnum, ImplicitPromote, IntegralType};

use half::f16;
use num::Zero;

use libc::c_int;
Expand Down Expand Up @@ -758,6 +760,7 @@ arith_scalar_spec!(Complex<f64>);
arith_scalar_spec!(Complex<f32>);
arith_scalar_spec!(f64);
arith_scalar_spec!(f32);
arith_scalar_spec!(f16);
arith_scalar_spec!(u64);
arith_scalar_spec!(i64);
arith_scalar_spec!(u32);
Expand Down
20 changes: 20 additions & 0 deletions src/core/data.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,6 +4,7 @@ use super::dim4::Dim4;
use super::error::HANDLE_ERROR;
use super::util::{af_array, c32, c64, dim_t, u64_t, HasAfEnum};

use half::f16;
use libc::{c_double, c_int, c_uint};
use std::option::Option;
use std::vec::Vec;
Expand Down Expand Up @@ -231,6 +232,25 @@ impl ConstGenerator for bool {
}
}

impl ConstGenerator for f16 {
type OutType = f16;

fn generate(&self, dims: Dim4) -> Array<Self::OutType> {
unsafe {
let mut temp: af_array = std::ptr::null_mut();
let err_val = af_constant(
&mut temp as *mut af_array,
f16::to_f64(*self),
dims.ndims() as c_uint,
dims.get().as_ptr() as *const dim_t,
12,
);
HANDLE_ERROR(AfError::from(err_val));
temp.into()
}
}
}

macro_rules! cnst {
($rust_type:ty, $ffi_type:expr) => {
impl ConstGenerator for $rust_type {
Expand Down
3 changes: 3 additions & 0 deletions src/core/num.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,3 +1,4 @@

pub trait Zero {
fn zero() -> Self;
}
Expand Down Expand Up @@ -28,6 +29,7 @@ zero_impl!(i64, 0);
zero_impl!(isize, 0);
zero_impl!(f32, 0.0);
zero_impl!(f64, 0.0);
zero_impl!(half::f16, half::f16::from_f32(0.0));

macro_rules! one_impl {
( $t:ident, $o:expr ) => {
Expand All @@ -51,3 +53,4 @@ one_impl!(i64, 1);
one_impl!(isize, 1);
one_impl!(f32, 1.0);
one_impl!(f64, 1.0);
one_impl!(half::f16, half::f16::from_f32(1.0));
38 changes: 37 additions & 1 deletion src/core/util.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -337,7 +337,7 @@ impl HasAfEnum for f16 {
type AbsOutType = Self;
type ArgOutType = Self;
type UnaryOutType = Self;
type ComplexOutType = Complex<f16>;
type ComplexOutType = Complex<f32>;
type MeanOutType = Self;
type AggregateOutType = f32;
type ProductOutType = f32;
Expand Down Expand Up @@ -468,6 +468,7 @@ macro_rules! implicit {
implicit!(c64, c32 => c64);
implicit!(c64, f64 => c64);
implicit!(c64, f32 => c64);
implicit!(c64, f16 => c64);
implicit!(c64, i64 => c64);
implicit!(c64, u64 => c64);
implicit!(c64, i32 => c64);
Expand All @@ -481,6 +482,7 @@ implicit!(c64, u8 => c64);
implicit!(c32, c64 => c64);
implicit!(c32, f64 => c64);
implicit!(c32, f32 => c32);
implicit!(c32, f16 => c32);
implicit!(c32, i64 => c32);
implicit!(c32, u64 => c32);
implicit!(c32, i32 => c32);
Expand All @@ -494,6 +496,7 @@ implicit!(c32, u8 => c32);
implicit!(f64, c64 => c64);
implicit!(f64, c32 => c64);
implicit!(f64, f32 => f64);
implicit!(f64, f16 => f64);
implicit!(f64, i64 => f64);
implicit!(f64, u64 => f64);
implicit!(f64, i32 => f64);
Expand All @@ -507,6 +510,7 @@ implicit!(f64, u8 => f64);
implicit!(f32, c64 => c64);
implicit!(f32, c32 => c32);
implicit!(f32, f64 => f64);
implicit!(f32, f16 => f32);
implicit!(f32, i64 => f32);
implicit!(f32, u64 => f32);
implicit!(f32, i32 => f32);
Expand All @@ -516,11 +520,26 @@ implicit!(f32, u16 => f32);
implicit!(f32, bool => f32);
implicit!(f32, u8 => f32);

//LHS is 16-bit floating point
implicit!(f16, c64 => c64);
implicit!(f16, c32 => c32);
implicit!(f16, f64 => f64);
implicit!(f16, f32 => f32);
implicit!(f16, i64 => f16);
implicit!(f16, u64 => f16);
implicit!(f16, i32 => f16);
implicit!(f16, u32 => f16);
implicit!(f16, i16 => f16);
implicit!(f16, u16 => f16);
implicit!(f16, bool => f16);
implicit!(f16, u8 => f16);

//LHS is 64-bit signed integer
implicit!(i64, c64 => c64);
implicit!(i64, c32 => c32);
implicit!(i64, f64 => f64);
implicit!(i64, f32 => f32);
implicit!(i64, f16 => f16);
implicit!(i64, u64 => u64);
implicit!(i64, i32 => i64);
implicit!(i64, u32 => i64);
Expand All @@ -534,6 +553,7 @@ implicit!(u64, c64 => c64);
implicit!(u64, c32 => c32);
implicit!(u64, f64 => f64);
implicit!(u64, f32 => f32);
implicit!(u64, f16 => f16);
implicit!(u64, i64 => u64);
implicit!(u64, i32 => u64);
implicit!(u64, u32 => u64);
Expand All @@ -547,6 +567,7 @@ implicit!(i32, c64 => c64);
implicit!(i32, c32 => c32);
implicit!(i32, f64 => f64);
implicit!(i32, f32 => f32);
implicit!(i32, f16 => f16);
implicit!(i32, i64 => i64);
implicit!(i32, u64 => u64);
implicit!(i32, u32 => u32);
Expand All @@ -560,6 +581,7 @@ implicit!(u32, c64 => c64);
implicit!(u32, c32 => c32);
implicit!(u32, f64 => f64);
implicit!(u32, f32 => f32);
implicit!(u32, f16 => f16);
implicit!(u32, i64 => i64);
implicit!(u32, u64 => u64);
implicit!(u32, i32 => u32);
Expand All @@ -573,6 +595,7 @@ implicit!(i16, c64 => c64);
implicit!(i16, c32 => c32);
implicit!(i16, f64 => f64);
implicit!(i16, f32 => f32);
implicit!(i16, f16 => f16);
implicit!(i16, i64 => i64);
implicit!(i16, u64 => u64);
implicit!(i16, i32 => i32);
Expand All @@ -586,6 +609,7 @@ implicit!(u16, c64 => c64);
implicit!(u16, c32 => c32);
implicit!(u16, f64 => f64);
implicit!(u16, f32 => f32);
implicit!(u16, f16 => f16);
implicit!(u16, i64 => i64);
implicit!(u16, u64 => u64);
implicit!(u16, i32 => i32);
Expand All @@ -599,6 +623,7 @@ implicit!(u8, c64 => c64);
implicit!(u8, c32 => c32);
implicit!(u8, f64 => f64);
implicit!(u8, f32 => f32);
implicit!(u8, f16 => f16);
implicit!(u8, i64 => i64);
implicit!(u8, u64 => u64);
implicit!(u8, i32 => i32);
Expand All @@ -612,6 +637,7 @@ implicit!(bool, c64 => c64);
implicit!(bool, c32 => c32);
implicit!(bool, f64 => f64);
implicit!(bool, f32 => f32);
implicit!(bool, f16 => f16);
implicit!(bool, i64 => i64);
implicit!(bool, u64 => u64);
implicit!(bool, i32 => i32);
Expand Down Expand Up @@ -652,12 +678,18 @@ impl FloatingPoint for f32 {
true
}
}
impl FloatingPoint for f16 {
fn is_real() -> bool {
true
}
}

///Trait qualifier to accept real data(numbers)
pub trait RealFloating: HasAfEnum {}

impl RealFloating for f64 {}
impl RealFloating for f32 {}
impl RealFloating for f16 {}

///Trait qualifier to accept complex data(numbers)
pub trait ComplexFloating: HasAfEnum {}
Expand All @@ -670,6 +702,7 @@ pub trait RealNumber: HasAfEnum {}

impl RealNumber for f64 {}
impl RealNumber for f32 {}
impl RealNumber for f16 {}
impl RealNumber for i32 {}
impl RealNumber for u32 {}
impl RealNumber for i16 {}
Expand Down Expand Up @@ -830,6 +863,8 @@ impl Fromf64 for u32 { fn fromf64(value: f64) -> Self { value as Self }}
#[rustfmt::skip]
impl Fromf64 for i32 { fn fromf64(value: f64) -> Self { value as Self }}
#[rustfmt::skip]
impl Fromf64 for f16 { fn fromf64(value: f64) -> Self { f16::from_f64(value) }}
#[rustfmt::skip]
impl Fromf64 for u16 { fn fromf64(value: f64) -> Self { value as Self }}
#[rustfmt::skip]
impl Fromf64 for i16 { fn fromf64(value: f64) -> Self { value as Self }}
Expand All @@ -847,6 +882,7 @@ impl IndexableType for u64 {}
impl IndexableType for f32 {}
impl IndexableType for i32 {}
impl IndexableType for u32 {}
impl IndexableType for f16 {}
impl IndexableType for i16 {}
impl IndexableType for u16 {}
impl IndexableType for u8 {}
Expand Down