paul@oddjob.UUCP (Paul Schinder) (03/15/85)
[For lack of better newsgroups to post to] We have an FPS-164 hanging off an Apollo workstation, and have had a great deal of trouble with the FPS. The biggest problem is that the Fortran compiler, when used with optimization levels greater than zero, does not produce good code. We have all sorts of bogus error messages, for example, error messages from library routines that aren't even called in the code. One of my codes, when compiled with opt=1 (local optimization only) produces error messages from the i/o libraries that don't occur when opt=0 (no optimization). The codes in question have run successfully on a great many different machines, and are written to be portable. There is no way we can optimize codes sufficiently to take advantage of the hardware, and so are getting speeds of only a few times a VAX 11/780 instead of a significant fraction of a Cray. We are thinking about getting rid of the FPS-164, and have gone so far as to find out that we should be able to get about 60% of our investment back by selling it used. What prompted this posting is that today we learned that FPS had the same trouble that we did with a code we sent to them when run on a Vax/VMS-FPS164 combination; when optimized, it produces errors that don't exist when it's not optimized. I was under the impression that the software for the Vax/VMS-FPS interface has been around for a number of years; I'm shocked that the Fortran compiler could be in such terrible shape. Have any of you had similar problems with the FPS-164? We would be interested in hearing what kinds of problems you've had and how well FPS handled them. -- Paul Schinder Astronomy and Astrophysics Center University of Chicago uucp: ..!ihnp4!oddjob!paul