mirror of
https://codeberg.org/ziglang/zig.git
synced 2026-04-28 19:47:08 +03:00
Marc Tiehuis
2085a4af56
The previous float-parsing method was lacking in a lot of areas. This commit introduces a state-of-the art implementation that is both accurate and fast to std. Code is derived from working repo https://github.com/tiehuis/zig-parsefloat. This includes more test-cases and performance numbers that are present in this commit. * Accuracy The primary testing regime has been using test-data found at https://github.com/tiehuis/parse-number-fxx-test-data. This is a fork of upstream with support for f128 test-cases added. This data has been verified against other independent implementations and represents accurate round-to-even IEEE-754 floating point semantics. * Performance Compared to the existing parseFloat implementation there is ~5-10x performance improvement using the above corpus. (f128 parsing excluded in below measurements). ** Old $ time ./test_all_fxx_data 3520298/5296694 succeeded (1776396 fail) ________________________________________________________ Executed in 28.68 secs fish external usr time 28.48 secs 0.00 micros 28.48 secs sys time 0.08 secs 694.00 micros 0.08 secs ** This Implementation $ time ./test_all_fxx_data 5296693/5296694 succeeded (1 fail) ________________________________________________________ Executed in 4.54 secs fish external usr time 4.37 secs 515.00 micros 4.37 secs sys time 0.10 secs 171.00 micros 0.10 secs Further performance numbers can be seen using the https://github.com/tiehuis/simple_fastfloat_benchmark/ repository, which compares against some other well-known string-to-float conversion functions. A breakdown can be found here: https://github.com/tiehuis/zig-parsefloat/blob/0d9f020f1a37ca88bf889703b397c1c41779f090/PERFORMANCE.md#commit-b15406a0d2e18b50a4b62fceb5a6a3bb60ca5706 In summary, we are within 20% of the C++ reference implementation and have about ~600-700MB/s throughput on a Intel I5-6500 3.5Ghz. * F128 Support Finally, f128 is now completely supported with full accuracy. This does use a slower path which is possible to improve in future. * Behavioural Changes There are a few behavioural changes to note. - `parseHexFloat` is now redundant and these are now supported directly in `parseFloat`. - We implement round-to-even in all parsing routines. This is as specified by IEEE-754. Previous code used different rounding mechanisms (standard was round-to-zero, hex-parsing looked to use round-up) so there may be subtle differences. Closes #2207. Fixes #11169.
65 lines
2.0 KiB
Zig
65 lines
2.0 KiB
Zig
const std = @import("std");
|
|
const parse = @import("parse.zig");
|
|
const parseNumber = parse.parseNumber;
|
|
const parseInfOrNan = parse.parseInfOrNan;
|
|
const convertFast = @import("convert_fast.zig").convertFast;
|
|
const convertEiselLemire = @import("convert_eisel_lemire.zig").convertEiselLemire;
|
|
const convertSlow = @import("convert_slow.zig").convertSlow;
|
|
const convertHex = @import("convert_hex.zig").convertHex;
|
|
|
|
const optimize = true;
|
|
|
|
pub const ParseFloatError = error{
|
|
InvalidCharacter,
|
|
};
|
|
|
|
pub fn parseFloat(comptime T: type, s: []const u8) ParseFloatError!T {
|
|
if (s.len == 0) {
|
|
return error.InvalidCharacter;
|
|
}
|
|
|
|
var i: usize = 0;
|
|
const negative = s[i] == '-';
|
|
if (s[i] == '-' or s[i] == '+') {
|
|
i += 1;
|
|
}
|
|
if (s.len == i) {
|
|
return error.InvalidCharacter;
|
|
}
|
|
|
|
const n = parse.parseNumber(T, s[i..], negative) orelse {
|
|
return parse.parseInfOrNan(T, s[i..], negative) orelse error.InvalidCharacter;
|
|
};
|
|
|
|
if (n.hex) {
|
|
return convertHex(T, n);
|
|
}
|
|
|
|
if (optimize) {
|
|
if (convertFast(T, n)) |f| {
|
|
return f;
|
|
}
|
|
|
|
if (T == f16 or T == f32 or T == f64) {
|
|
// If significant digits were truncated, then we can have rounding error
|
|
// only if `mantissa + 1` produces a different result. We also avoid
|
|
// redundantly using the Eisel-Lemire algorithm if it was unable to
|
|
// correctly round on the first pass.
|
|
if (convertEiselLemire(T, n.exponent, n.mantissa)) |bf| {
|
|
if (!n.many_digits) {
|
|
return bf.toFloat(T, n.negative);
|
|
}
|
|
if (convertEiselLemire(T, n.exponent, n.mantissa + 1)) |bf2| {
|
|
if (bf.eql(bf2)) {
|
|
return bf.toFloat(T, n.negative);
|
|
}
|
|
}
|
|
}
|
|
}
|
|
}
|
|
|
|
// Unable to correctly round the float using the Eisel-Lemire algorithm.
|
|
// Fallback to a slower, but always correct algorithm.
|
|
return convertSlow(T, s[i..]).toFloat(T, negative);
|
|
}
|