new amd64 architecture will indeed increase floating-point
precision. its use of SSE2 registers for floating-point which do
operations in IEEE 32-bit or 64-bit arithmetic instead of the x87's
default 80-bit IEEE arithmetic.
we have been waiting for these affordable amd64 bit. we have been eager to run fortran programs in a machine with more than 4gb memory and on a more precise floating-point operations.
we have been waiting for these affordable amd64 bit. we have been eager to run fortran programs in a machine with more than 4gb memory and on a more precise floating-point operations.
From: Rafael 'Dido' Sevilla <[EMAIL PROTECTED] >
Subject: Re: [plug] x86-64 not recommended?
To: [EMAIL PROTECTED], "The Main Philippine Linux Users' Group (PLUG)
Discussion List" <[email protected]>
Andy Sy wrote:
> c) need the precision for mathematical calculations
>
Need the precision for mathematical calculations? The last time I
checked, double precision IEEE-754 numbers are 64 bits wide, and every
PC FPU since the 8087 math coprocessor has been internally doing
floating point operations on 80-bit long doubles. I don't see how going
64-bit would make mathematical calculations more precise than you can
already do them. It would make them faster, certainly, but they'll be
just as accurate as they were on 32-bit processors.
--
The ability to change outcomes is not the ability to control outcomes.
http://stormwyrm.blogspot.com/
_________________________________________________ Philippine Linux Users' Group (PLUG) Mailing List [email protected] (#PLUG @ irc.free.net.ph) Read the Guidelines: http://linux.org.ph/lists Searchable Archives: http://archives.free.net.ph
