Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!rutgers!ames!ucbcad!ucbvax!sdcsvax!sdcc6!loral!ian
From: ian@loral.UUCP
Newsgroups: comp.arch
Subject: Re: Will the Karp Problem Be Solved?
Message-ID: <1306@loral.UUCP>
Date: Tue, 9-Dec-86 16:27:55 EST
Article-I.D.: loral.1306
Posted: Tue Dec  9 16:27:55 1986
Date-Received: Sun, 14-Dec-86 02:01:26 EST
References: <3302@curly.ucla-cs.UCLA.EDU> <144@linus.UUCP>
Reply-To: ian@loral.UUCP (Ian Kaplan)
Organization: Loral Instrumentation, San Diego
Lines: 35

In article <144@linus.UUCP> bs@linus.UUCP (Robert D. Silverman) writes:
>
>
>At the Crypto '86 conference I presented a new variation of the
>Quadratic Sieve algorithm for factoring large integers. The variation
>has been programmed on a STAR configuation of SUN-3's using ethernet
>connections. 

  Although this many be an interesting parallel algorithm, it is far
  from meeting Karp's challenge.  The point of Karp's challenge is that
  you must have a _general_ purpose parallel programming system - at least
  general purpose in the domain of numeric computation.  The programs that
  Dr. Karp proposes are much more complex than the program mentioned above.

  I have written parallel sorts and matrix multiplies for the Loral
  dataflow processor, but these are toy programs compared to what Karp
  proposes.  One of the problems with Karp's challenge (as others have
  pointed out) is that it will not be met unless someone foots the bill for
  a rather prodigious software development effort.  For parallel machines
  of the scale that Karp's challenge covers (e.g., 100 or more processing
  elements) software development is harder than it is for sequential
  processors.  For example, it took me a week to get a parallel matrix
  multiply program written and debugged on the Loral dataflow system.
  On a sequential processor this would be a trivial task.

  I believe that better programming environments will emerge, but they are
  a few years away.  Karp's challenge will be met, if it is met at all, as
  a side effect of another project.

		     Ian Kaplan
		     Loral Dataflow Group
		     Loral Instrumentation
	     USENET: {ucbvax,decvax,ihnp4}!sdcsvax!loral!ian
	     ARPA:   sdcc6!loral!ian@UCSD
	     USPS:   8401 Aero Dr. San Diego, CA 92123