1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372
#![allow(incomplete_features)]
//#![cfg_attr(not(test), no_std)]
#![feature(const_trait_impl)]
#![feature(generic_const_exprs)]
#![feature(trait_alias)]
#![feature(const_fn_floating_point_arithmetic)]
#![feature(portable_simd)]
#![feature(associated_type_bounds)]
#![feature(round_ties_even)]
#![feature(float_gamma)]
#![feature(lazy_cell)]
//! # Custom Float
//!
//! This crate adds a custom floating point number type, `Fp<U, SIGN_BIT, EXP_SIZE, INT_SIZE, FRAC_SIZE, EXP_BASE>`, where the bit size of the exponent and mantissa can be set separately.
//!
//! `U` is the underlying unsigned integer type which is used to represent the number.
//!
//! `SIGN_BIT` is wether or not the number has a sign bit.
//!
//! `EXP_SIZE` is the size of the exponent in bits.
//!
//! `INT_SIZE` is the size of the integer part of the mantissa in bits. If zero, then the integer bit is implicit.
//!
//! `FRAC_SIZE` is the size of the fractional part of the mantissa in bits.
//!
//! `EXP_BASE` is the base of the exponent.
//!
//! The total bit size of `U` must be greater or equal to `SIGN_BIT` + `EXP_SIZE` + `INT_SIZE` + `FRAC_SIZE` to contain the entire number.
//!
//! The bit layout is as follows:
//! ```txt
//! No data: | Sign: | Exponent: | Integer: | Fractional: |
//! < .. > | <SIGN_BIT> | <EXP_SIZE> | <INT_SIZE> | <FRAC_SIZE> |
//! ```
//!
//! The value of a real floating-point number is the following:
//! ```txt
//! x = (-1)**sign*EXP_BASE**(exponent - bias)*mantissa
//! ```
//!
//! where the bias equals
//! ```txt
//! bias = 2**(EXP_SIZE - 1) - 1
//! ```
//!
//! If the exponent has the maximum value, the number is either infinity or NaN.
//!
//! The number then automatically implements `num::Float`, and supports all ordinary floating point operations.
//!
//! This allows simple implementation of special floating point types, such as TensorFloat, IEEE754 Quadruple/binary128, Fp80, and BFloat16.
//!
//! The accuracy of all of the floating point operations are not perfect, but work well enough to be usable. Various plots showing the accuracy of basic functions are shown in the [plots](https://github.com/sigurd4/custom_float/tree/master/plots) subfolder.
//!
//! All floats can be converted into each other painlessly, though the conversion may produce rounding errors or unbounded outputs when converting to a float with lesser resolution.
//!
//! # Examples
//!
//! ```rust
//! #![feature(generic_const_exprs)]
//!
//! use custom_float::Fp;
//!
//! type FpSingle = Fp<u32, true, 8, 0, 23, 2>;
//!
//! let two = FpSingle::from(2);
//! let four = FpSingle::from(4);
//!
//! assert_eq!(two + two, four);
//! ```
use core::fmt::Debug;
use num_traits::{CheckedNeg, CheckedShl, CheckedShr, PrimInt, Signed, Unsigned};
moddef::moddef!(
pub mod {
ati,
g_711,
google,
ibm,
ieee754,
intel,
khronos,
nvidia,
pixar
},
flat(pub) mod {
fp
},
mod {
util,
plot for cfg(test)
}
);
pub trait UInt = Unsigned + PrimInt + CheckedShl + CheckedShr + Debug;
pub trait Int = Signed + PrimInt + CheckedShl + CheckedShr + CheckedNeg;
#[cfg(test)]
mod tests
{
#![allow(unused)]
use std::{ops::{Range, RangeBounds}, time::{Instant, SystemTime}};
use array_math::{ArrayMath, ArrayOps};
use linspace::LinspaceArray;
use num::Complex;
use num_traits::{Float, One, ToPrimitive, Zero};
use crate::{ati::Fp24, g_711::FpG711, google::{Bf16, Bf32, Bf8}, ibm::{HFpLong, HFpShort}, ieee754::{DecDouble, FpDouble, FpHalf, FpQuadruple, FpSingle}, intel::Fp80, khronos::KhronosFp11, nvidia::Tf19, plot, Fp};
pub type F = Fp<u32, true, 8, 0, 23, 2>;
#[test]
fn it_works()
{
let mut x = [0.0; 8].map(|x| F::from(x));
x[0] = F::one() + F::one() + F::one();
x[1] = F::one() + F::one();
x[2] = F::one();
println!("{:?}", x);
let mut y = [Complex::zero(); 5];
x.real_fft(&mut y);
x.real_ifft(&y);
println!("{:?}", x);
}
pub fn ttable<F: Float>() -> Vec<F>
{
vec![
F::from(3.333333).unwrap(),
F::from(10.0).unwrap(),
F::from(16.0).unwrap(),
F::from(-2.2).unwrap(),
F::from(2.2).unwrap(),
F::one(),
F::one().tan(),
F::one() + F::epsilon(),
F::zero(),
F::from(0.5).unwrap(),
F::from(-0.5).unwrap(),
F::nan(),
F::infinity(),
F::neg_infinity(),
F::min_positive_value(),
]
}
pub fn test_op2(fn_name: &str, op1: impl Fn(f32, f32) -> f32, op2: impl Fn(F, F) -> F, d: Option<f32>)
{
for f0 in crate::tests::ttable()
{
for f1 in crate::tests::ttable()
{
let fp0 = F::from(f0);
let fp1 = F::from(f1);
//println!("{} ? {}", f0, f1);
let s = op1(f0, f1);
let sp: f32 = op2(fp0, fp1).into();
if match d
{
Some(d) => (s - sp).abs() > d,
None => s != sp
} && !(s.is_nan() && sp.is_nan())
{
if f0.is_subnormal()
{
println!("f is subnormal");
}
if s.is_subnormal()
{
println!("y is subnormal");
}
println!("{:e} ? {:e} == {:e} != {:e}", f0, f1, s, sp);
}
}
}
}
pub fn test_op1(fn_name: &str, op1: impl Fn(f32) -> f32, op2: impl Fn(F) -> F, d: Option<f32>, r: Option<Range<f32>>)
{
for f0 in ttable::<f32>()
{
let fp0 = F::from(f0);
let s = op1(f0);
let sp: f32 = op2(fp0).into();
if match d
{
Some(d) => (s - sp).abs() > d,
None => s != sp
} && !(s.is_nan() && sp.is_nan())
{
if f0.is_subnormal()
{
println!("f is subnormal");
}
if s.is_subnormal()
{
println!("y is subnormal");
}
println!("{:e} ? == {:e} != {:e}", f0, s, sp);
}
}
if let Some(r) = r
{
plot_approx(fn_name, r.clone(), &op1, |x| op2(Fp::from(x)).into());
plot_err(fn_name, r.clone(), &op1, |x| op2(Fp::from(x)).into());
if BENCH
{
plot_bench(fn_name, r, |x| {
let t0 = Instant::now();
for _ in 0..M
{
let _ = op1(x);
}
Instant::now().duration_since(t0).div_f64(M as f64).as_secs_f32()
}, |x| {
let x = Fp::from(x);
let t0 = Instant::now();
for _ in 0..M
{
let _ = op2(x);
}
Instant::now().duration_since(t0).div_f64(M as f64).as_secs_f32()
})
}
}
}
const M: usize = 64;
const N: usize = 1024;
const PLOT_TARGET: &str = "plots";
const BENCH: bool = false;
#[allow(unused)]
pub fn plot_err<R>(
fn_name: &str,
range: R,
func: impl Fn(f32) -> f32,
approx: impl Fn(f32) -> f32
)
where
R: RangeBounds<f32> + LinspaceArray<f32, N>
{
let x: [f32; N] = range.linspace_array();
let y_approx = x.map(approx);
let y = x.map(func);
let e = y_approx.sub_each(y);
let plot_title: &str = &format!("{fn_name}(x) error");
let plot_path: &str = &format!("{PLOT_TARGET}/error/{fn_name}_error.png");
plot::plot_curves(plot_title, plot_path, [x], [e])
.expect("Plot error");
/*let (avg_error, max_abs_error) = y.zip(y_approx)
.map(|(y, y_approx)| y - y_approx)
.map(|y| (y, y.abs()))
.reduce(|a, b| (a.0 + b.0, a.1.max(b.1)))
.map(|(sum_error, max_abs_error)| (sum_error/N as f32, max_abs_error))
.unwrap_or_default();
println!("Average Error: {}", avg_error);
println!("Max |Error|: {}", max_abs_error);*/
}
#[allow(unused)]
pub fn plot_approx<R>(
fn_name: &str,
range: R,
func: impl Fn(f32) -> f32,
approx: impl Fn(f32) -> f32
)
where
R: RangeBounds<f32> + LinspaceArray<f32, N>
{
let x: [f32; N] = range.linspace_array();
let y_approx = x.map(approx);
let y = x.map(func);
let plot_title: &str = &format!("{fn_name}(x)");
let plot_path: &str = &format!("{PLOT_TARGET}/{fn_name}.png");
plot::plot_curves(plot_title, plot_path, [x, x], [y, y_approx])
.expect("Plot error");
/*let (avg_error, max_abs_error) = y.zip(y_approx)
.map(|(y, y_approx)| y - y_approx)
.map(|y| (y, y.abs()))
.reduce(|a, b| (a.0 + b.0, a.1.max(b.1)))
.map(|(sum_error, max_abs_error)| (sum_error/N as f32, max_abs_error))
.unwrap_or_default();
println!("Average Error: {}", avg_error);
println!("Max |Error|: {}", max_abs_error);*/
}
#[allow(unused)]
pub fn plot_bench<R>(
fn_name: &str,
range: R,
func: impl Fn(f32) -> f32,
approx: impl Fn(f32) -> f32
)
where
R: RangeBounds<f32> + LinspaceArray<f32, N>
{
let x: [f32; N] = range.linspace_array();
let y_approx = x.map(approx);
let y = x.map(func);
let plot_title: &str = &format!("{fn_name}(x) benchmark");
let plot_path: &str = &format!("{PLOT_TARGET}/bench/{fn_name}_bench.png");
plot::plot_curves(plot_title, plot_path, [x, x], [y, y_approx])
.expect("Plot error");
/*let (avg_error, max_abs_error) = y.zip(y_approx)
.map(|(y, y_approx)| y - y_approx)
.map(|y| (y, y.abs()))
.reduce(|a, b| (a.0 + b.0, a.1.max(b.1)))
.map(|(sum_error, max_abs_error)| (sum_error/N as f32, max_abs_error))
.unwrap_or_default();
println!("Average Error: {}", avg_error);
println!("Max |Error|: {}", max_abs_error);*/
}
#[test]
fn int_bit()
{
let x = Fp80::from(2.0);
println!("{}", x.int_bits());
println!("{}", x);
println!("{:b}", x)
}
#[test]
fn test_convert()
{
test_op1("ident", |x| x, |x| x, None, Some(-5.0..20.0))
}
#[test]
fn test_to_int()
{
for n in u8::MIN..=u8::MAX
{
let f = F::from_uint(n);
assert_eq!(f.to_uint_wrapping::<u8>(), n);
assert_eq!(f.to_uint(), Some(n));
}
}
}