vault/fastfloat: Fast and exact implementation of the C

fast_float number parsing library: 4x faster than strtod

The fast_float library provides fast header-only implementations for the C++ from_chars functions for float and double types as well as integer types. These functions convert ASCII strings representing decimal values (e.g., 1.3e10) into binary types. We provide exact rounding (including round to even). In our experience, these fast_float functions many times faster than comparable number-parsing functions from existing C++ standard libraries.

Specifically, fast_float provides the following two functions to parse floating-point numbers with a C++17-like syntax (the library itself only requires C++11):

from_chars_result from_chars(char const *first, char const *last, float &value, ...);
from_chars_result from_chars(char const *first, char const *last, double &value, ...);

You can also parse integer types:

from_chars_result from_chars(char const *first, char const *last, int &value, ...);
from_chars_result from_chars(char const *first, char const *last, unsigned &value, ...);

The return type (from_chars_result) is defined as the struct:

struct from_chars_result {
  char const *ptr;
  std::errc ec;
};

It parses the character sequence [first, last) for a number. It parses floating-point numbers expecting a locale-independent format equivalent to the C++17 from_chars function. The resulting floating-point value is the closest floating-point values (using either float or double), using the “round to even” convention for values that would otherwise fall right in-between two values. That is, we provide exact parsing according to the IEEE standard.

Given a successful parse, the pointer (ptr) in the returned value is set to point right after the parsed number, and the value referenced is set to the parsed value. In case of error, the returned ec contains a representative error, otherwise the default (std::errc()) value is stored.

The implementation does not throw and does not allocate memory (e.g., with new or malloc).

It will parse infinity and nan values.

Example:

#include "fast_float/fast_float.h"
#include <iostream>

int main() {
  std::string input = "3.1416 xyz ";
  double result;
  auto answer = fast_float::from_chars(input.data(), input.data() + input.size(), result);
  if (answer.ec != std::errc()) { std::cerr << "parsing failure\n"; return EXIT_FAILURE; }
  std::cout << "parsed the number " << result << std::endl;
  return EXIT_SUCCESS;
}

You can parse delimited numbers:

  std::string input = "234532.3426362,7869234.9823,324562.645";
  double result;
  auto answer = fast_float::from_chars(input.data(), input.data() + input.size(), result);
  if (answer.ec != std::errc()) {
    // check error
  }
  // we have result == 234532.3426362.
  if (answer.ptr[0] != ',') {
    // unexpected delimiter
  }
  answer = fast_float::from_chars(answer.ptr + 1, input.data() + input.size(), result);
  if (answer.ec != std::errc()) {
    // check error
  }
  // we have result == 7869234.9823.
  if (answer.ptr[0] != ',') {
    // unexpected delimiter
  }
  answer = fast_float::from_chars(answer.ptr + 1, input.data() + input.size(), result);
  if (answer.ec != std::errc()) {
    // check error
  }
  // we have result == 324562.645.

Like the C++17 standard, the fast_float::from_chars functions take an optional last argument of the type fast_float::chars_format. It is a bitset value: we check whether fmt & fast_float::chars_format::fixed and fmt & fast_float::chars_format::scientific are set to determine whether we allow the fixed point and scientific notation respectively. The default is fast_float::chars_format::general which allows both fixed and scientific.

The library seeks to follow the C++17 (see 28.2.3.(6.1)) specification.

The from_chars function does not skip leading white-space characters (unless fast_float::chars_format::chars_format is set).
A leading + sign is forbidden (unless fast_float::chars_format::skip_white_space is set).
It is generally impossible to represent a decimal value exactly as binary floating-point number (float and double types). We seek the nearest value. We round to an even mantissa when we are in-between two binary floating-point numbers.

Furthermore, we have the following restrictions:

We support float and double, but not long double. We also support fixed-width floating-point types such as std::float32_t and std::float64_t.
We only support the decimal format: we do not support hexadecimal strings.
For values that are either very large or very small (e.g., 1e9999), we represent it using the infinity or negative infinity value and the returned ec is set to std::errc::result_out_of_range.

We support Visual Studio, macOS, Linux, freeBSD. We support big and little endian. We support 32-bit and 64-bit systems.

We assume that the rounding mode is set to nearest (std::fegetround() == FE_TONEAREST).

Integer types

You can also parse integer types using different bases (e.g., 2, 10, 16). The following code will print the number 22250738585072012 three times:

#include "fast_float/fast_float.h"
#include <iostream>

int main() {
  uint64_t i;
  std::string str = "22250738585072012";
  auto answer = fast_float::from_chars(str.data(), str.data() + str.size(), i);
  if (answer.ec != std::errc()) {
    std::cerr << "parsing failure\n";
    return EXIT_FAILURE;
  }
  std::cout << "parsed the number " << i << std::endl;

  std::string binstr = "1001111000011001110110111001001010110100111000110001100";

  answer = fast_float::from_chars(binstr.data(), binstr.data() + binstr.size(), i, 2);
  if (answer.ec != std::errc()) {
    std::cerr << "parsing failure\n";
    return EXIT_FAILURE;
  }
  std::cout << "parsed the number " << i << std::endl;

  std::string hexstr = "4f0cedc95a718c";

  answer = fast_float::from_chars(hexstr.data(), hexstr.data() + hexstr.size(), i, 16);
  if (answer.ec != std::errc()) {
    std::cerr << "parsing failure\n";
    return EXIT_FAILURE;
  }
  std::cout << "parsed the number " << i << std::endl;
  return EXIT_SUCCESS;
}

Behavior of result_out_of_range

When parsing floating-point values, the numbers can sometimes be too small (e.g., 1e-1000) or too large (e.g., 1e1000). The C language established the precedent that these small values are out of range. In such cases, it is customary to parse small values to zero and large values to infinity. That is the behaviour of the C language (e.g., stdtod). That is the behaviour followed by the fast_float library.

Specifically, we follow Jonathan Wakely’s interpretation of the standard:

In any case, the resulting value is one of at most two floating-point values closest to the value of the string matching the pattern.

It is also the approach taken by the Microsoft C++ library.

Hence, we have the following examples:

  double result = -1;
  std::string str = "3e-1000";
  auto r = fast_float::from_chars(str.data(), str.data() + str.size(), result);
  // r.ec == std::errc::result_out_of_range
  // r.ptr == str.data() + 7
  // result == 0

  double result = -1;
  std::string str = "3e1000";
  auto r = fast_float::from_chars(str.data(), str.data() + str.size(), result);
  // r.ec == std::errc::result_out_of_range
  // r.ptr == str.data() + 6
  // result == std::numeric_limits<double>::infinity()

Users who wish for the value to be left unmodified given std::errc::result_out_of_range may do so by adding two lines of code:

  double old_result = result; // make copy
  auto r = fast_float::from_chars(start, end, result);
  if (r.ec == std::errc::result_out_of_range) { result = old_result; }

C++20: compile-time evaluation (constexpr)

In C++20, you may use fast_float::from_chars to parse strings at compile-time, as in the following example:

// consteval forces compile-time evaluation of the function in C++20.
consteval double parse(std::string_view input) {
  double result;
  auto answer = fast_float::from_chars(input.data(), input.data() + input.size(), result);
  if (answer.ec != std::errc()) { return -1.0; }
  return result;
}

// This function should compile to a function which
// merely returns 3.1415.
constexpr double constexptest() {
  return parse("3.1415 input");
}

C++23: Fixed width floating-point types

The library also supports fixed-width floating-point types such as std::float32_t and std::float64_t. E.g., you can write:

std::float32_t result;
auto answer = fast_float::from_chars(f.data(), f.data() + f.size(), result);

Non-ASCII Inputs

We also support UTF-16 and UTF-32 inputs, as well as ASCII/UTF-8, as in the following example:

#include "fast_float/fast_float.h"
#include <iostream>

int main() {
  std::u16string input = u"3.1416 xyz ";
  double result;
  auto answer = fast_float::from_chars(input.data(), input.data() + input.size(), result);
  if (answer.ec != std::errc()) { std::cerr << "parsing failure\n"; return EXIT_FAILURE; }
  std::cout << "parsed the number " << result << std::endl;
  return EXIT_SUCCESS;
}

Advanced options: using commas as decimal separator, JSON and Fortran

The C++ standard stipulate that from_chars has to be locale-independent. In particular, the decimal separator has to be the period (.). However, some users still want to use the fast_float library with in a locale-dependent manner. Using a separate function called from_chars_advanced, we allow the users to pass a parse_options instance which contains a custom decimal separator (e.g., the comma). You may use it as follows.

#include "fast_float/fast_float.h"
#include <iostream>

int main() {
  std::string input = "3,1416 xyz ";
  double result;
  fast_float::parse_options options{fast_float::chars_format::general, ','};
  auto answer = fast_float::from_chars_advanced(input.data(), input.data() + input.size(), result, options);
  if ((answer.ec != std::errc()) || ((result != 3.1416))) { std::cerr << "parsing failure\n"; return EXIT_FAILURE; }
  std::cout << "parsed the number " << result << std::endl;
  return EXIT_SUCCESS;
}

You can also parse Fortran-like inputs

#include "fast_float/fast_float.h"
#include <iostream>

int main() {
  std::string input = "1d+4";
  double result;
  fast_float::parse_options options{fast_float::chars_format::fortran};
  auto answer = fast_float::from_chars_advanced(input.data(), input.data() + input.size(), result, options);
  if ((answer.ec != std::errc()) || ((result != 10000))) { std::cerr << "parsing failure\n"; return EXIT_FAILURE; }
  std::cout << "parsed the number " << result << std::endl;
  return EXIT_SUCCESS;
}

You may also enforce the JSON format (RFC 8259)

#include "fast_float/fast_float.h"
#include <iostream>

int main() {
  std::string input = "+.1"; // not valid
  double result;
  fast_float::parse_options options{fast_float::chars_format::json};
  auto answer = fast_float::from_chars_advanced(input.data(), input.data() + input.size(), result, options);
  if (answer.ec == std::errc()) { std::cerr << "should have failed\n"; return EXIT_FAILURE; }
  return EXIT_SUCCESS;
}

By default the JSON format does not allow inf:

#include "fast_float/fast_float.h"
#include <iostream>

int main() {
  std::string input = "inf"; // not valid in JSON
  double result;
  fast_float::parse_options options{fast_float::chars_format::json};
  auto answer = fast_float::from_chars_advanced(input.data(), input.data() + input.size(), result, options);
  if (answer.ec == std::errc()) { std::cerr << "should have failed\n"; return EXIT_FAILURE; }
  return EXIT_SUCCESS;
}

You can allow it with a non-standard json_or_infnan variant:

#include "fast_float/fast_float.h"
#include <iostream>

int main() {
  std::string input = "inf"; // not valid in JSON but we allow it with json_or_infnan
  double result;
  fast_float::parse_options options{fast_float::chars_format::json_or_infnan};
  auto answer = fast_float::from_chars_advanced(input.data(), input.data() + input.size(), result, options);
  if (answer.ec != std::errc() || (!std::isinf(result))) { std::cerr << "should have parsed infinity\n"; return EXIT_FAILURE; }
  return EXIT_SUCCESS;
}

Users and Related Work

The fast_float library is part of:

GCC (as of version 12): the from_chars function in GCC relies on fast_float,
Chromium, the engine behind Google Chrome, Microsoft Edge, and Opera,
WebKit, the engine behind Safari (Apple’s web browser),
DuckDB,
Redis,
Apache Arrow where it multiplied the number parsing speed by two or three times,
Google Jsonnet,
ClickHouse.

The fastfloat algorithm is part of the LLVM standard libraries. There is a derived implementation part of AdaCore.

The fast_float library provides a performance similar to that of the fast_double_parser library but using an updated algorithm reworked from the ground up, and while offering an API more in line with the expectations of C++ programmers. The fast_double_parser library is part of the Microsoft LightGBM machine-learning framework.

References

Daniel Lemire, Number Parsing at a Gigabyte per Second, Software: Practice and Experience 51 (8), 2021.
Noble Mushtak, Daniel Lemire, Fast Number Parsing Without Fallback, Software: Practice and Experience 53 (7), 2023.

Other programming languages

There is an R binding called rcppfastfloat.
There is a Rust port of the fast_float library called fast-float-rust.
There is a Java port of the fast_float library called FastDoubleParser. It used for important systems such as Jackson.
There is a C# port of the fast_float library called csFastFloat.

How fast is it?

It can parse random floating-point numbers at a speed of 1 GB/s on some systems. We find that it is often twice as fast as the best available competitor, and many times faster than many standard-library implementations.

fast_float is many times faster than many standard-library
implementations

$ ./build/benchmarks/benchmark
# parsing random integers in the range [0,1)
volume = 2.09808 MB
netlib                                  :   271.18 MB/s (+/- 1.2 %)    12.93 Mfloat/s
doubleconversion                        :   225.35 MB/s (+/- 1.2 %)    10.74 Mfloat/s
strtod                                  :   190.94 MB/s (+/- 1.6 %)     9.10 Mfloat/s
abseil                                  :   430.45 MB/s (+/- 2.2 %)    20.52 Mfloat/s
fastfloat                               :  1042.38 MB/s (+/- 9.9 %)    49.68 Mfloat/s

See https://github.com/lemire/simple_fastfloat_benchmark for our benchmarking code.

Video

Using as a CMake dependency

This library is header-only by design. The CMake file provides the fast_float target which is merely a pointer to the include directory.

If you drop the fast_float repository in your CMake project, you should be able to use it in this manner:

add_subdirectory(fast_float)
target_link_libraries(myprogram PUBLIC fast_float)

Or you may want to retrieve the dependency automatically if you have a sufficiently recent version of CMake (3.11 or better at least):

FetchContent_Declare(
  fast_float
  GIT_REPOSITORY https://github.com/fastfloat/fast_float.git
  GIT_TAG tags/v6.1.6
  GIT_SHALLOW TRUE)

FetchContent_MakeAvailable(fast_float)
target_link_libraries(myprogram PUBLIC fast_float)

You should change the GIT_TAG line so that you recover the version you wish to use.

You may also use CPM, like so:

CPMAddPackage(
  NAME fast_float
  GITHUB_REPOSITORY "fastfloat/fast_float"
  GIT_TAG v6.1.6)

Using as single header

The script script/amalgamate.py may be used to generate a single header version of the library if so desired. Just run the script from the root directory of this repository. You can customize the license type and output file if desired as described in the command line help.

You may directly download automatically generated single-header files:

https://github.com/fastfloat/fast_float/releases/download/v7.0.0/fast_float.h

Packages

The fast_float library is part of the Conan package manager.
It is part of the brew package manager.
Some Linux distribution like Fedora include fast_float (e.g., as fast_float-devel).

Credit

Though this work is inspired by many different people, this work benefited especially from exchanges with Michael Eisel, who motivated the original research with his key insights, and with Nigel Tao who provided invaluable feedback. Rémy Oudompheng first implemented a fast path we use in the case of long digits.

The library includes code adapted from Google Wuffs (written by Nigel Tao) which was originally published under the Apache 2.0 license.

License

^{Licensed under either of Apache License, Version 2.0 or MIT license or BOOST license.}

_{Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in this repository by you, as defined in the Apache-2.0 license, shall be triple licensed as above, without any additional terms or conditions.}