Boost C++ Libraries of the most highly regarded and expertly designed C++ library projects in the world. Herb Sutter and Andrei Alexandrescu, C++ Coding Standards

This is the documentation for an old version of Boost. Click here to view this page for the latest version.


Appendix: History
Appendix: Rationale
Appendix: Implementation Notes
Appendix: FAQ
Appendix: Acknowledgements
Appendix: Future plans

New Features:

  • Enhance chrono I/O
    • #5980 Enhance chrono I/O with H. Hinnant proposal proposal which has the advantage to provide I/O for system clocks using the Gregorian Calendar.
    • #5981 Add i/o state savers for duration and time_point formatting state.
    • #7059 Add low level i/o facilities.


  • The chrono i/o version included in Boost.Chrono 1.2.x has been completly refactored in version 2.0.0
  • chrono I/O: The manipulators duration_short, duration_long are depreceated. You should use the parameterized form duration_fmt or the renamed manipulators __duration_symbol and __duration_prefix instead.
  • chrono I/O: The duration_punct<> facet is depreceated. You should use the get_duration_style free function to get the informations and use the duration_units facet for localization purposes.

When BOOST_CHRONO_VERSION==2 the preceding deprecated functions are not available.


  • #7381 C++11 compliance: unresolved symbol when assigning a constexpr duration to a non-const local variable.
  • #7479 Compiles fails with compilers supporting constexpr fails if the standard library doesn't provides the constexpr interface
  • #7493 compile fail on intel-linux- because of bug on explicit bool conversion
  • #7542 Missing -lpthread in chrono/io tester Sandia-clang-trunk

Would not fix:

  • #6871 chrono_io.hpp: operator<<(ostream& os, ...) modifies the state of os.
    • The neww io interface provided in version 2 solves this issue. You should move to the new version.

Known bugs not fixed yet:

  • #7525 Wrong clock_string<system_clock>::since() on Windows
[Warning] Warning

The time_point<system_clock,D> formatter/parser specializations don't work yet. It has been disable defining BOOST_CHRONO_DONT_PROVIDE_DATE_IO_FOR_SYSTEM_CLOCK_TIME_POINT until a fix for the following ticket is found:

- #7547 time_point<system_clock> input version 2 fails to compile

- #7546 time_point<system_clock> output version 2 fails to compile

In this case the io operators behave like any time_point as if the specialization was removed.

In order to enable this features on compilers working with, you will need to define BOOST_CHRONO_PROVIDES_DATE_IO_FOR_SYSTEM_CLOCK_TIME_POINT.


  • #6918 Boost Chrono compilation failure on HP uX due to undefined pthread_getcpuclockid.
  • #6241 boost::chrono compilation problems without std::wstring support.
  • #6987 Documentation & C++11.
  • #7041 time_point.hpp depends on Boost.System.
  • #7042 Avoiding time_point and duration dependency on time.h and CLOCK_REALTIME.
  • #7058 Make it work when BOOST_NO_EXCEPTIONS is defined.
  • #7069 Misspellings in clock_string<thread_clock>.
  • #7081 WinError.h capitalization in boost/detail/win/basic_types.hpp.


  • #6361 integer overflow in boost::chrono::process_real_cpu_clock::now() under Windows 32bits.
  • #6628 compiler warning in process_cpu_clocks.hpp.
  • #6666 thread_clock.hpp needs pthread.h.


  • #6092 Input from non integral durations makes the compiler fail.
  • #6093 [1/3]second fails as valid duration input.
  • #6113 duplicate symbol when BOOST_CHRONO_HEADER_ONLY is defined.
  • #6243 Sandia-pgi-11.9: more than one instance of overloaded function "min" matches.
  • #6257 process_cpu_clock::now() on linux gives time_points 1/1000 times.

New Features:

  • #5979 Added chrono rounding utilities as defined By Howard Hinnant here.
  • #5978 Added BOOST_CHRONO_HAS_PROCESS_CLOCKS to know if process clocks are available.
  • #5998 Make possible to don't provide hybrid error handling.
  • #5906 Take in account the constexpr as defined in the standard.
  • #5907 Take in account noexcept for compilers supporting it.


  • #2114 Enable visibility support (Boost.Chorno part)
  • #5669 Intel compiler failure to compile duration.hpp
  • #5909 process_cpu_clock::now() on MAC gives time_points 1/1000 times.
  • #5946 Process real cpu clock returns the system steady clock (windows).
  • #5974 Process real cpu clock should use clock() instead of times() in MAC which is twice faster and have better resolution.


  • #5975 Reduce the combinations of header-only, shared, static link to reduce test time by 50%.
  • #5976 chrono_accuracy_test is not deterministic and should be removed from the regression tests
  • #5977 Remove old files from Beman's version. Some old files included in the Beman's version and not documented in the reviewed version that have been definitely removed from the repository as
    • boost/chrono/timer.hpp,
    • boost/chrono/process_times.hpp
    • boost/chrono/detail/process_clock.hpp,
    • boost/chrono/detail/mac/process_clock.hpp,
    • boost/chrono/detail/posix/process_clock.hpp,
    • boost/chrono/detail/win/process_clock.hpp,
    • boost/chrono/detail/run_timer.hpp,
    • boost/chrono/detail/run_timer_static.hpp,

New Features:

  • #???? Added time_point unary operators +,-,++,-- and binary operators +,- with Rep al RHS.
  • #5323 Add Associated type difference_type for chrono::time_point.


  • #5322 Explicit default constructed chrono::durations are uninitialized
  • Moved chrono to trunk taking in account the review remarks.
  • Documentation revision.


  • Boost_Chrono is now a configurable header-only library version (that also allows the user to choose if the windows.h file is included or not).
  • Added clock_string<> traits.
  • Define chrono-io for all the clocks.
  • Add input of process_times representation.


  • Use of detail/win files to avoid the use of windows.h file.
  • Completed the error_code handling.


  • Fix some warnings.
  • Fix original errors on Mac
  • Don't fix the link with boost_system to static.


  • Added test on process and thread clocks.
  • Moved to lightweight_test.hpp.
  • Able to test multiple configurations.


  • Removed some not useful parts as the test and the tickets.

See N2661 - A Foundation to Sleep On which is very informative and provides motivation for key design decisions. This section contains some extracts from this document.

Why duration needs operator%

This operator is convenient for computing where in a time frame a given duration lies. A motivating example is converting a duration into a "broken-down" time duration such as hours::minutes::seconds:

class ClockTime
    typedef boost::chrono::hours hours;
    typedef boost::chrono::minutes minutes;
    typedef boost::chrono::seconds seconds;
    hours hours_;
    minutes minutes_;
    seconds seconds_;

    template <class Rep, class Period>
      explicit ClockTime(const boost::chrono::duration<Rep, Period>& d)
        : hours_  (boost::chrono::duration_cast<hours>  (d)),
          minutes_(boost::chrono::duration_cast<minutes>(d % hours(1))),
          seconds_(boost::chrono::duration_cast<seconds>(d % minutes(1)))
Which APIs have been chosen to implement each clock on each platform?

The following table presents a resume of which API is uused for each clock on each platform

Table 4.3. Clock API correspondence


Windows Platform

Posix Platform

Mac Platform



clock_gettime( CLOCK_REALTIME)



QueryPerformanceCounter and QueryPerformanceFrequency

clock_gettime( CLOCK_STEADY)






















Why does process_cpu_clock sometimes give more cpu seconds than real seconds?

Ask your operating system supplier. The results have been inspected with a debugger, and both for Windows and Linux, that's what the OS appears to be reporting at times.

Are integer overflows in the duration arithmetic detected and reported?

Boost.Ratio avoids all kind of overflow that could result of arithmetic operation and that can be simplified. The typedefs durations don't detect overflow. You will need a duration representation that handles overflow.

Which clocks should be used to benchmarking?

Each clock has his own features. It depends on what do you need to benchmark. Most of the time, you could be interested in using a thread clock, but if you need to measure code subject to synchronization a process clock would be better. If you have a multi-process application, a system-wide clock could be needed.

Which clocks should be used for watching?

For trace purposes, it is probably best to use a system-wide clock.

The library's code was derived from Howard Hinnant's time2_demo prototype. Many thanks to Howard for making his code available under the Boost license. The original code was modified by Beman Dawes to conform to Boost conventions.

time2_demo contained this comment:

Much thanks to Andrei Alexandrescu, Walter Brown, Peter Dimov, Jeff Garland, Terry Golubiewski, Daniel Krugler, Anthony Williams.

The file <boost/chrono_io.hpp> has been adapted from the experimental header <chrono_io> from Howard Hinnant. Thanks for all Howard.

Howard Hinnant, who is the real author of the library, has provided valuable feedback and suggestions during the development of the library. In particular, The chrono_io_io.hpp source has been adapted from the experimental header <chrono_io> from Howard Hinnant.

The acceptance review of Boost.Ratio took place between November 5th and 15th 2010. Many thanks to Anthony Williams, the review manager, and to all the reviewers: David Deakins, John Bytheway, Roland Bock and Paul A. Bristow.

Thanks to Ronald Bock, Andrew Chinoff, Paul A. Bristow and John Bytheway for his help polishing the documentation.

Thanks to Tom Tan for reporting some compiler issues with MSVC V10 beta and MinGW-gcc-4.4.0 and for the many pushing for an homogeneous process_cpu_clock clock.

Thanks to Ronald Bock for reporting Valgind issues and for the many suggestions he made concerning the documentation.

For later releases
  • Include Stopwatches.
  • Add User defined literals for some durations.
  • Include chrono::date as defined by Howard Hinnant here.