[multiprecision] New binary floating point type
Folks, One of the main feature requests during the multiprecision lib review was for a binary floating point type. There is now such a type in the sandbox under the "multiprecision.cpp_bin_float" directory - you will also need latest SVN Trunk to try this out. This new type is correctly rounded to nearest for all arithmetic operations plus binary-decimal conversions and sqrt. Minimal docs so far here: https://svn.boost.org/svn/boost/sandbox/multiprecision.cpp_bin_float/libs/mu... I'd welcome comments, and if folks would like to try and break it, that would be cool too ;-) Regards, John.
I'd welcome comments, and if folks would like to try and break it, that would be cool too ;-)
Awesome work! A minor nit: In the second paragraph of the preliminary docs, the back end being discussed is cpp_bin_float (binary: base-2). The typedefs mentioned, however, are for decimal types such as cpp_dec_float_50 (radix-10, base 10). A modification of the preliminary documentation seems to be in order. Awesome big numbers, John! I can't wait to get my grubby little hands on these... Sincerely, Chris. On Sunday, October 20, 2013 6:59 PM, John Maddock <john@johnmaddock.co.uk> wrote: Folks, One of the main feature requests during the multiprecision lib review was for a binary floating point type. There is now such a type in the sandbox under the "multiprecision.cpp_bin_float" directory - you will also need latest SVN Trunk to try this out. This new type is correctly rounded to nearest for all arithmetic operations plus binary-decimal conversions and sqrt. Minimal docs so far here: https://svn.boost.org/svn/boost/sandbox/multiprecision.cpp_bin_float/libs/mu... I'd welcome comments, and if folks would like to try and break it, that would be cool too ;-) Regards, John. _______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
The typedefs mentioned, however, are for decimal types such as cpp_dec_float_50 (radix-10, base 10). A modification of the preliminary documentation seems to be in order.
Done, thanks, John. PS thought you were supposed to be gone fishing?? ;-)
Yes, but I'm back now. I exaggerated the return date, just in case of any big troubles with travel or return to work. Caught some great scenery and king salmon up to 40 pounds. cpp_bin_float is looking good to me! And I will keep working with it and report anything significant. Sincerely, Chris. On Monday, October 21, 2013 2:27 PM, John Maddock <john@johnmaddock.co.uk> wrote:
The typedefs mentioned, however, are for decimal types such as cpp_dec_float_50 (radix-10, base 10). A modification of the preliminary documentation seems to be in order.
Done, thanks, John. PS thought you were supposed to be gone fishing?? ;-) _______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
Nitpicking again. In paragraph 4 of the cpp_bin_float tutorial, you astutely note that "providing an allocator as the final template parameter causes cpp_bin_float to dynamically allocate the memory it needs" But the allocator type is no longer the *final* template parameter, since exponent types and ranges have been added as template parameters. So... is it the 3rd parameter, the 4th-to-last parameter...? Whatever, but it's not the last. Groovy big numbers! Sincerely, Chris. On Monday, October 21, 2013 9:38 PM, Christopher Kormanyos <e_float@yahoo.com> wrote:
The typedefs mentioned, however, are for decimal types such as cpp_dec_float_50 (radix-10, base 10). A modification of the preliminary documentation seems to be in order.
Done, thanks, John. PS thought you were supposed to be gone fishing?? ;-)
Yes, but I'm back now. I exaggerated the return date, just in case of any big troubles with travel or return to work. Caught some great scenery and king salmon up to 40 pounds. cpp_bin_float is looking good to me! And I will keep working with it and report anything significant. Sincerely, Chris. On Monday, October 21, 2013 2:27 PM, John Maddock <john@johnmaddock.co.uk> wrote:
The typedefs mentioned, however, are for decimal types such as cpp_dec_float_50 (radix-10, base 10). A modification of the preliminary documentation seems to be in order.
Done, thanks, John. PS thought you were supposed to be gone fishing?? ;-) _______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
participants (2)
-
Christopher Kormanyos
-
John Maddock