[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference turris::fortran

Title:	Digital Fortran
Notice:	Read notes 1.* for important information
Moderator:	QUARK::LIONEL

Created:	Thu Jun 01 1995
Last Modified:	Fri Jun 06 1997
Last Successful Update:	Fri Jun 06 1997
Number of topics:	1333
Total number of notes:	6734

1232.0. "Visual Fortran - Floating Point Support" by MKOTS3::HAHN (SBU Americas Technical Support Group) Wed Mar 19 1997 10:39

A customer has a question about the new Visual Fortran product. He is 
currently using Microsoft's Fortran Powerstation, which he says "is optimized 
for Intel, and supports 80-bit internal floating point". He would like to
know if the same is true for our Visual Fortran product?

I explained the pieces from the Product Description re: support for 32-bit 
IEEE S_floating data type and 64-bit IEEE T_floating data type. But his reply 
was that they were "external" representations, and he was interested in
"internal".

Thanks,
 Dave

(If there is a new conference for Visual Fortran, please advise - I searched
through this conference, but didn't see a mention of a new conference).

T.R Title User Personal
Name Date Lines

1232.1 DVF is the same as FPS on this TLE::WHITLOCK Stan Whitlock Wed Mar 19 1997 15:38 24

T.R	Title	User	Personal Name	Date	Lines
1232.1	DVF is the same as FPS on this	TLE::WHITLOCK	Stan Whitlock	`Wed Mar 19 1997 15:38`	24
	>>currently using Microsoft's Fortran Powerstation, which he says "is optimized >>for Intel, and supports 80-bit internal floating point". He would like to >>know if the same is true for our Visual Fortran product? Digital Visual Fortran is the same as Fortran Powerstation on these 2 items: o DVF is optimized for Intel hardware o DVF "supports" 80-bit internal floating point to the same extent that Fortran Powerstation does. DVF runs the Intel hardware by default in 64-bit mode just like FPS does. So add, sub, mul, div, and sqrt all return 64-bit results to the registers. But sin, cos, etc return 80-bits - on both DVF and FPS. Neither DVF nor FPS has an explicit REAL type that lets the user store all 80 bits. There is an NT call {available from both DVF and FPS} so the user can set the hardware to run in 80-bit result mode... the user gets what he gets. >>(If there is a new conference for Visual Fortran, please advise - I searched >>through this conference, but didn't see a mention of a new conference). This conference is meant to handle all Fortran questions, f90 and f77, on all supported platforms, so this is the place for DVF questions too. /Stan

>>currently using Microsoft's Fortran Powerstation, which he says "is optimized 
>>for Intel, and supports 80-bit internal floating point". He would like to
>>know if the same is true for our Visual Fortran product?

Digital Visual Fortran is the same as Fortran Powerstation on these 2 items:

    o	DVF is optimized for Intel hardware

    o	DVF "supports" 80-bit internal floating point to the same extent that
	Fortran Powerstation does.  DVF runs the Intel hardware by default in
	64-bit mode just like FPS does.  So add, sub, mul, div, and sqrt all
	return 64-bit results to the registers.  But sin, cos, etc return
	80-bits - on both DVF and FPS.  Neither DVF nor FPS has an explicit
	REAL type that lets the user store all 80 bits.  There is an NT call
	{available from both DVF and FPS} so the user can set the hardware
	to run in 80-bit result mode... the user gets what he gets.

>>(If there is a new conference for Visual Fortran, please advise - I searched
>>through this conference, but didn't see a mention of a new conference).

This conference is meant to handle all Fortran questions, f90 and f77, on all
supported platforms, so this is the place for DVF questions too.

/Stan