km@emory.UUCP (Ken Mandelberg) (01/28/86)
I saw a DEC press release that says that the f77 in the
latest Ultrix release has been much improved, and is 3 times
faster than the BSD version.
Does anyone know what this means? Does the target code run
3 times faster, or the compiler itself?
--
Ken Mandelberg
Emory University
Dept of Math and CS
Atlanta, Ga 30322
{akgua,sb1,gatech,decvax}!emory!km USENET
km@emory CSNET
km.emory@csnet-relay ARPANETsml@wdl1.UUCP (01/31/86)
The Fortran on Ultrix is called VAX Fortran. It is a port of the VMS Fortran to Ultrix, not a rewrite of f77. Compilation speed is 2-3 times faster than f77. Excecution speed is 1.5x for the Whetstone benchmark and 2.4x for a Spice circuit simulation. Steve Lazarus (415) 852-4203 Ford Aerospace ...fortune!wdl1!sml (USENET) MS X-20 sml@ford-wdl1 (ARPA) 3939 Fabian Way Palo Alto, CA 94303
gwyn@brl-tgr.ARPA (Doug Gwyn <gwyn>) (02/01/86)
> I saw a DEC press release that says that the f77 in the > latest Ultrix release has been much improved, and is 3 times > faster than the BSD version. > > Does anyone know what this means? Does the target code run > 3 times faster, or the compiler itself? Almost certainly, the generated code (under favorable circumstances). The VMS Fortran compiler generates pretty decent VAX code, not surprisingly. It sounds very much like they ported that to Ultrix.
kenward@mdivax1.UUCP (kenward) (02/06/86)
> > >> I saw a DEC press release that says that the f77 in the >> latest Ultrix release has been much improved, and is 3 times >> faster than the BSD version. >> >> Does anyone know what this means? Does the target code run >> 3 times faster, or the compiler itself? > >Almost certainly, the generated code (under favorable circumstances). >The VMS Fortran compiler generates pretty decent VAX code, not >surprisingly. It sounds very much like they ported that to Ultrix. > Having used both compilers, I cannot agree. The Ultrix f77 still appears to be the same fortran to intermediate C code generator that bsd4.2 f77 is based upon. I would believe that DEC has tuned the code generation to improve execution (as well as fix some bugs). It would be nice if VMS Fortran were ported, it is much nicer to use (as in much more user friendly). -- Gary W. Kenward Mobile Data International Inc. Riverside Industrial Park Richmond, B.C. Canada V7A 4Z3 Plus ca change, plus c'est la meme chose! >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> SNAP! <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
wyatt@cfa.UUCP (Bill Wyatt) (02/12/86)
> >> I saw a DEC press release that says that the f77 in the > >> latest Ultrix release has been much improved, and is 3 times > >> faster than the BSD version. If it's a DEC press release, they are talking about their VAX FORTRAN for Ultrix 1.2, not the f77 compiler. > >> > >> Does anyone know what this means? Does the target code run > >> 3 times faster, or the compiler itself? > > > >Almost certainly, the generated code (under favorable circumstances). > >The VMS Fortran compiler generates pretty decent VAX code, not > >surprisingly. It sounds very much like they ported that to Ultrix. > > Having used both compilers, I cannot agree. The Ultrix f77 still appears > to be the same fortran to intermediate C code generator that bsd4.2 f77 > is based upon. Ditto. (see above comment). The VAX FORTRAN compiler *compiles* much faster, and runs about 20% faster (well, confession... it should, but I don't actually have it [yet]). HOWEVER, f77 under bsd 4.3 (and Ultrix 1.2) is very much improved, especially in its I/O efficiency, so some classes of programs might run 3x faster. > I would believe that DEC has tuned the code generation to improve > execution (as well as fix some bugs). It would be nice if VMS Fortran > were ported, it is much nicer to use (as in much more user friendly). I doubt DEC has tweaked anything in f77, but yes, they have ported the VMS Fortran compiler (under Ulrix 1.2 et.seq.). -- Bill UUCP: {seismo|ihnp4|cmcl2}!harvard!talcott!cfa!wyatt Wyatt ARPA: wyatt%cfa.UUCP@harvard.HARVARD.EDU
peters@cubsvax.UUCP (Peter S. Shenkin) (02/16/86)
In article <cfa.181> wyatt@cfa.UUCP (Bill Wyatt) writes: >> >> I saw a DEC press release that says that the f77 in the >> >> latest Ultrix release has been much improved, and is 3 times >> >> faster than the BSD version. > >If it's a DEC press release, they are talking about their VAX FORTRAN for >Ultrix 1.2, not the f77 compiler. > (etc., etc.....) My understanding -- which may be wrong -- is that VMS FORTRAN (ie, DEC's native FORTRAN compiler long available under VMS) has been ported to an ULTRIX environment; this is the FORTRAN that contains goodies such as DO...ENDDO, in-line comments starting with "!" and terminating with a newline, and so on. Though ported, this has definitely *not* been bundled with ULTRIX; its costs about the same to license as the same compiler would cost under VMS -- someplace in the $5-10K range for a VAX11-780 in a multiuser environment. In *addition*, I wouldn't be surprised to learn that the f77 ordinarily bundled with ULTRIX has been improved. The Berkeley 4.2 version which came with the first release of ULTRIX had many bugs, especially in the code optimiser, but not limited to it (for instance, it didn't properly pass a literal string to a subroutine). The Berkeley 4.3 f77 is said to be greatly improved, and I think (though don't know) that there is a VAX-specific (ie, non-ortable, ie, fast) version around. I have heard very good things about this compiler; for instance, that it compares well in execution speed with DEC's native (ie "VMS") FORTRAN. This, of course, is just a cleaned up version of the f77 that we know and love-hate, allowing strings to be delineated with "...", lines to begin with a tab, a statement number followed by a tab, an ampersand (indicating continuation) followed by a tab, and explicit recursion. If, in fact, the f77 that is bundled with ULTRIX has been improved, it has probably been replaced by the Berkeley 4.3 version. (Donn Seeley once told me that this version uses different intermediate code conventions than its predecissor, and so the C-compiler has probably been updated as well, if I am right. Hope I haven't taken your name in vain, Donn!) Well, if anyone can *categorically* confirm or deny the above mixture of revealed truth and speculation, the net would probably enjoy hearing about it.... you out there, DEC? Peter S. Shenkin Columbia Univ. Biol. Sciences {philabs,rna}!cubsvax!peters cubsvax!peters@columbia.ARPA