home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.arch:10644 comp.lang.misc:3606
- Newsgroups: comp.arch,comp.lang.misc
- Path: sparky!uunet!mcsun!news.funet.fi!network.jyu.fi!sakkinen
- From: sakkinen@jyu.fi (Markku Sakkinen)
- Subject: Re: Hardware Support for Numeric Algorithms
- Message-ID: <1992Nov12.110605.3606@jyu.fi>
- Organization: University of Jyvaskyla, Finland
- References: <1992Nov5.202412.7266@linus.mitre.org> <1992Nov10.153705.27804@yrloc.ipsa.reuter.COM> <martin.721555689@bert>
- Date: Thu, 12 Nov 1992 11:06:05 GMT
- Lines: 41
-
- In article <martin.721555689@bert> martin@math.rwth-aachen.de ( Martin Schoenert) writes:
- > ...
- >bs@gauss.mitre.org (Robert D. Silverman) continued:
- >
- > Needless to say, there are people who have to worry about speed FIRST,
- > and that other considerations are (almost) irrelevent.
- >
- >rbe@yrloc.ipsa.reuter.COM (Robert Bernecky) replied:
- >
- > Are you saying that speed is more important than correct answers?
- > {I don't think so, but I though I had better ask.}
- >
- >No, I don't think that this is what Bob is talking about. However in the
- >area of integer factorization correctness is easy to achieve, because you
- >can check your answers so easily. But reasonable efficiency can be
- >*very* difficult to obtain.
-
- As somebody already remarked, that is what Silverman _wrote_ but probably
- not what he _meant_.
-
- Correctness must be important even for these special problems.
- Suppose that you get an answer after burning some hundreds of MIPS-years,
- and the easy check shows that the answer is wrong.
- If a correct but "inefficient" programme would have needed only twice
- as much resources for the computation, your choice to run the incorrect
- but "inefficient" programme was obviously bad -- if this is more or
- less a one-shot job.
-
- The economy becomes different if the software can be sufficiently
- tested on small problems, or if it is to be used without further
- modifications to solve a lot of huge problems, once it is correct.
-
- ----------------------------------------------------------------------
- Markku Sakkinen (sakkinen@jytko.jyu.fi)
- SAKKINEN@FINJYU.bitnet (alternative network address)
- Department of Computer Science and Information Systems
- University of Jyvaskyla (a's with umlauts)
- PL 35
- SF-40351 Jyvaskyla (umlauts again)
- Finland
- ----------------------------------------------------------------------
-