notes@mcgill-vision.UUCP (01/20/86)
In trying to decide whether or not to use VMS or Ultrix on one of our new machines (a MicroVAX II), the politics has come down to the decision that if we can get a good FORTRAN compiler to run under Ultrix, we will go with that, otherwise it is VMS. Now, for reasons of systems management (I don't want to get into religious debates over the relative virtues of VMS and Unix) most of the programmers and system managers in the lab would like to continue with Unix, so... I'm looking for pointers, testimonials, addresses and basically recommendations from anyone out there who might have gone through this before and has something to recommend. As can probably be guessed from this, f77 is NOT a viable option. We are looking for something that produces EFFICIENT number-crunching code, and is hopefully compatible with FORTRAN IV, and, of course, it must run on VAXen. Please MAIL me any responses. Of course, if there is sufficient interest, I will post a summary to the net. Lee Iverson utcsri!mcgill-vision!leei Mcgill University, Montreal Computer Vision and Robotics Lab
ken@fear.UUCP (Ken Rauch) (01/27/86)
In article <348@mcgill-vision.UUCP>, notes@mcgill-vision.UUCP writes: > > In trying to decide whether or not to use VMS or Ultrix on one of our > new machines (a MicroVAX II), the politics has come down to the decision > that if we can get a good FORTRAN compiler to run under Ultrix, we will > go with that, otherwise it is VMS. Now, for reasons of systems management > (I don't want to get into religious debates over the relative virtues of > VMS and Unix) most of the programmers and system managers in the lab would > like to continue with Unix, so... > DEC has recently announced that it is offering the very same Fortran compiler under Ultrix that it has been offering under VMX so your choice should be easy.
sml@wdl1.UUCP (01/27/86)
<< In trying to decide whether or not to use VMS or Ultrix on one of our << new machines (a MicroVAX II), the politics has come down to the decision << that if we can get a good FORTRAN compiler to run under Ultrix, we will << go with that, otherwise it is VMS. The VMS Fortran compiler has been ported to Ultrix by Dec. You now have the SAME Fortran available on both VMS and Ultrix. 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
levy@ttrdc.UUCP (Daniel R. Levy) (01/30/86)
In article <961@wdl1.UUCP>, sml@wdl1.UUCP writes: >The VMS Fortran compiler has been ported to Ultrix by Dec. You now have >the SAME Fortran available on both VMS and Ultrix. > >Steve Lazarus (415) 852-4203 >Ford Aerospace ...fortune!wdl1!sml (USENET) I have seen a number of postings alluding to this port (which if true, is wonderful). Could someone out there who has actually USED this beast (on Ultrix) comment on whether its performance is really the equal of the VMS? In crunch mode I would imagine it would be, but doesn't its I/O have to go through the usual Unix mechanisms, that is, write() and read(), as well as the Unix idiosyncracies of its filesystem, with the (possible) degradation of performance or loss of features? For example, under VMS there are a zillion possible types of 'data files' with different record length, carriage control attributes, etc. How does this ported Fortran deal with open(unit=1,file='foobar.dat',form='unformatted',recl=80, + access='direct',status='old') for instance? Does it look like the same way the old f77 did such things? Would binaries written by the 'old' f77 be readable by the 'new' without any internal structure to the file? Or are headers prepended to binary files as was so strongly pushed a few months ago here on the net by some guy from DEC? Also carriage control. Sounds trivial enough to detect a tty on a logical unit and apply carriage control to IT, but how is carriage control handled on file output? Are there ways to get at the command line args and environment directly (a la f77's getarg() and getenv())? (Whew, I'm outta breath. Only reason I can't check this out myself is that I don't even HAVE Ultrix or BSD anything here.... just VMS and SysV.) -- ------------------------------- Disclaimer: The views contained herein are | dan levy | yvel nad | my own and are not at all those of my em- | an engihacker @ | ployer or the administrator of any computer | at&t computer systems division | upon which I may hack. | skokie, illinois | -------------------------------- Path: ..!{akgua,homxb,ihnp4,ltuxa,mvuxa, vax135}!ttrdc!levy
wyatt@cfa.UUCP (Bill Wyatt) (01/31/86)
> In article <961@wdl1.UUCP>, sml@wdl1.UUCP writes: > >The VMS Fortran compiler has been ported to Ultrix by Dec. You now have > >the SAME Fortran available on both VMS and Ultrix. > > > I have seen a number of postings alluding to this port (which if true, > is wonderful). Could someone out there who has actually USED this > beast (on Ultrix) comment on whether its performance is really the equal > of the VMS? In crunch mode I would imagine it would be, but doesn't its > I/O have to go through the usual Unix mechanisms, that is, write() and > read(), as well as the Unix idiosyncracies of its filesystem, with the > (possible) degradation of performance or loss of features? For example, > [ ... ] > open(unit=1,file='foobar.dat',form='unformatted',recl=80, > + access='direct',status='old') > > for instance? Does it look like the same way the old f77 did such things? > Would binaries written by the 'old' f77 be readable by the 'new' without > any internal structure to the file? Or are headers prepended to binary > files as was so strongly pushed a few months ago here on the net by some > guy from DEC? Also carriage control. [ ... ] VAX Fortran is available only under Ultrix 1.2 (and later), which isn't announced yet. Since 1.2 f77 has many, perhaps all of the bsd 4.3 f77 performance enhancements, I suspect the VAX Fortran will run very fast. The example cited above is standard f77 format, and since VAX Fortran is a superset of f77, the statement is the same. I would not expect the object modules to be compatible, but they are compatible with other languages, DBX, and AR. Sorry, but I don't know about carriage control issues. -- Bill UUCP: {seismo|ihnp4|cmcl2}!harvard!talcott!cfa!wyatt Wyatt ARPA: wyatt%cfa.UUCP@harvard.HARVARD.EDU