apfloat

Version history

New features in version 2.41:

64-bit version for x86-64 (AMD-64 / EM64T) gcc. Minor fixes.

New features in version 2.40:

Use new Standard Template Library (e.g. <iostream> instead of <iostream.h>). 64-bit version for IA-64 (Itanium) gcc. Factorial function. Renamed *.cc to *.cpp. Minor fixes and performance improvements.

New features in version 2.35:

Fixes for gcc 3.3 compatibility. Fix post-increment/decrement vs. pre-increment/decrement operators. Other minor fixes.

New features in version 2.34:

Minor bug fixes.

New features in version 2.33:

Fixes for gcc 3.0 compatibility.

New features in version 2.32:

Minor bug fixes.

New features in version 2.31:

64-bit versions for Borland C++ and Microsoft Visual C++. Minor bug fixes and perfomance improvements.

New features in version 2.30:

Parallel processing pi calculation program. Multithreading FNT algorithms for Win32 and Posix threads. Saving and loading numbers to/from disk using the swapto() and swapfrom() member functions. Environment variables for system settings as an alternative to those in the file apfloat.ini. Some bug fixes and small perfomance improvements.

New features in version 2.21:

Various bug fixes and small perfomance improvements. 64-bit version for Linux.

New features in version 2.20:

Stream input operators. Modulo power function with a sample RSA encryption application. Various bug fixes and some perfomance improvements.

New features in version 2.10:

Some perfomance improvements. Versions for Microsoft Visual C++. 64-bit version for djgpp. Some minor bug fixes.

New features in version 2.00:

Improved the multiplication of short numbers dramatically. Calculating pi is more than two times faster now with the Chudnovsky brothers' binsplit algorithm. Some minor bug fixes.

New features in version 1.51:

Some critical bug fixes.

New features in version 1.50:

Integer and rational arbitrary precision number types. More portable. The version using doubles is faster. A new Win32 version.

New features in version 1.40:

Transform length can have a factor of three, enabling a 25% speedup sometimes. fmod() and modf() functions. Slightly more portable.

New features in version 1.33:

Calling apinit() and apdeinit() is no longer required, all initializations are done automatically.

New features in version 1.32:

Fixed bug caused by the complex class being a template class in newer ANSI C++ draft implementations.

New features in version 1.31:

New transformed apfloat class. Speeds up multiplications by constant apfloats.

New features in version 1.30:

A lot faster. The Pentium version is up to 50% faster and the Alpha version up to 200% faster. Easier use of arbitrary bases. A new more portable 32-bit version.

New features in version 1.20:

Arbitrary bases (not just decimal). Complex numbers. Transcendental functions (exp(), log(), sin(), cos(), ...).

New features in version 1.10:

Different versions (32-bit, 64-bit, floating-point). floor() and ceil() functions. Slightly faster.
Back to the apfloat page.