gleicher@b.gp.cs.cmu.edu (Michael Gleicher) (01/22/89)
Has anyone successfully gotten SB-Prolog to run on an SGI Personal Iris? It refuses to compile it seems. Any hints would be greatly appreciated. Thanks, Mike -- Michael Lee Gleicher gleicher@cs.cmu.edu (-: Of course I believe Carnegie Mellon University 5610 Elmer St. Apt 10 (-: in miracles, Pittsburgh, PA 15232 (-: its my job. --
scotth@harlie.SGI.COM (Scott Henry) (01/24/89)
From article <4087@pt.cs.cmu.edu>, by gleicher@b.gp.cs.cmu.edu (Michael Gleicher): > Has anyone successfully gotten SB-Prolog to run on an SGI Personal > Iris? It refuses to compile it seems. > Any hints would be greatly appreciated. I have gotten SB-Prolog up and running on a 4D70G. Since there is no GL involved, it should work on all 4D's. The "standard" version of SB-Prolog will only work on byte-addressable architectures (eg: VAX), but Saumya Debray (debray@arizona.edu) had put together a version for alignment- restricted processors (such as SPARC), called SBP V3.0. I took the code and replaced the BSD4.3-isms by IRIX-isms, and sent the diffs back to him. I am still having some problems (core dumps and such), but I am currently learning prolog, and am not sure where the problem lies. Ideally, it should never core-dump, but it does so infrequently. Optimizing with -O4 shrunk the binary, and speeded up some benchmarks by 3-45% (depending). Compiled prolog seems to be (on one test, the seive of Eratosthenes) about 5-10x faster than interpreted. I would contact S. Debray about FTP-ing the V3 code. Or, If you have anonymous (incoming) FTP, I might be able to send it to you (we have no anonymous FTP). > Thanks, > Mike > > -- > Michael Lee Gleicher gleicher@cs.cmu.edu (-: Of course I believe > Carnegie Mellon University 5610 Elmer St. Apt 10 (-: in miracles, > Pittsburgh, PA 15232 (-: its my job. > -- Scott Henry <scotth@sgi.com> Silicon Graphics, Inc. #disclaimer: I did the port solely to support some of my own work, and SGI # makes no warranty about anything I say. This is my stuff only, # (etc, etc...). -- Scott Henry