nljg0470@uxa.cso.uiuc.edu (11/11/89)
I hate to bother all of you apparent C gods out there with a question about something as trivial as FORTRAN, but the faculty here insist on using it. Question: In Think C, you can choose, via which library you link, whether your code will utilize SANE, the FPU, or a little bit of both. What I am wondering, since my prof asked me, is in Absoft FORTRAN/20 when you choose to use the 68881/68882 optiion, is SANE totally bypassed and the built in coprocessor functions are used exclussively or does it still use SANE in order to get the 96bit (I think it's 96 but I've been wrong before (yes, it's true)) precision. Also the same question applies to Language Systems FORTRAN. Do any FORTRAN compilers give you this control. Also, as long as you're here, does anyone know if SPAM can totally bypass SANE, and itself, it send all calcs to the FPU. I only recall it being for use in systems w/o the math chip. Finally, what about the Radius SANE patch, is there somewhere I can get a copy to play with w/o buying the whole damn board. (Perhaps someone could e-mail me a copy?) Thanks much Nick Jasper nljg0470@uxa.cso.uiuc.edu University of Illinois Aero/Astro Engineering
gilbertd@silver.bacs.indiana.edu (Don Gilbert) (11/13/89)
I think Language Systems fortran is excellent, and is well integrated
with MPW. It uses the same command line switches as MPW pascal to generate
68881/2 math chip calls. It also has an -extended switch to
force all fortran REAL types to Mac Extended type. This is what I use
to compile for MacII class machines:
Fortran loopcalc.f -3 -extended -u 0 -mc68020 -mc68881
Just leave out the -extended -mc... switches to compile for standard
SANE. BTW, LS Fortran (esp. with the 68881 direct calls) generates very
fast code.
-- Don
Don Gilbert biocomputing office / archive for
gilbertd@iubio.bio.indiana.edu / molecular & general biology
gilbertd@iubacs.bitnet /
biology dept., indiana univ., / ftp iubio.bio.indiana.edu
bloomington, in 47405, usa / (129.79.1.101) user anonymous siegel@endor.harvard.edu (Rich Siegel) (11/13/89)
In article <227700051@uxa.cso.uiuc.edu> nljg0470@uxa.cso.uiuc.edu writes: > > Question: In Think C, you can choose, via which library you link, >whether your code will utilize SANE, the FPU, or a little bit of both. What Actually, the choice is made by selecting the appropriate options from the Options... dialog; the choice of library is made simply to complement the choice of code generation, so that everything works correctly. The same holds true for other compilers: choose the code generation, and then link the libraries to match. >to use the 68881/68882 optiion, is SANE totally bypassed and the built in >coprocessor functions are used exclussively or does it still use SANE in order >to get the 96bit (I think it's 96 but I've been wrong before (yes, it's true)) As it turns out the 96-bit data type has the same precision as the 80-bit data type, but has an extra word of zeros after the exponent "to force longword alignment and for future expansion", according to the Motorola manuals. Generally, the compiler will do everything possible using the FPU, but some functions (such as string-to-binary conversions) are still done by SANE. R. ~~~~~~~~~~~~~~~ Rich Siegel Staff Software Developer Symantec Corporation, Language Products Group Internet: siegel@endor.harvard.edu UUCP: ..harvard!endor!siegel "There is no personal problem which cannot be solved by sufficient application of high explosives." ~~~~~~~~~~~~~~~