| Crates.io | const_soft_float |
| lib.rs | const_soft_float |
| version | 0.1.4 |
| created_at | 2023-02-07 15:03:26.329386+00 |
| updated_at | 2023-07-03 02:52:36.631129+00 |
| description | Const Soft Float Point |
| homepage | |
| repository | https://github.com/823984418/const_soft_float |
| max_upload_size | |
| id | 778820 |
| size | 179,261 |
Floating-point code is from compiler_builtins = "0.1.94" and libm = "0.2.6", and has been rewritten for use in a constant context.
Fuzzing of all operations is performed against the relevant reference code to ensure correctness of the port, but please open an issue if there is any inconsistent behavior.
Exported Soft Float Types:
SoftF32SoftF64Features:
no_std - Enabled by defaultconst_trait_impl - For const operator implementationsconst_mut_refs - For const operator with assignment implementationsOn stable:
const fn const_f32_add(a: f32, b: f32) -> f32 {
SoftF32(a).add(SoftF32(b)).to_f32()
}
On nightly with const_trait_impl usage:
const fn const_f32_add(a: f32, b: f32) -> f32 {
(SoftF32(a) + SoftF32(b)).to_f32()
}
On nightly with const_mut_refs usage:
const fn const_f32_add(a: f32, b: f32) -> f32 {
let mut x = SoftF32(a);
x += SoftF32(b);
x.to_f32()
}
Implemented const Functions:
from_(f32/f64)to_(f32/f64)from_bitsto_bitsaddsubmuldivcmpnegsqrtpowicopysigntruncroundfloorsincos