home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!gatech!prism!jm59
- From: jm59@prism.gatech.EDU (MILLS,JOHN M.)
- Newsgroups: comp.lang.ada
- Subject: Re: Ada for Transputers!
- Message-ID: <79333@hydra.gatech.EDU>
- Date: 7 Jan 93 14:05:10 GMT
- References: <3029.2b487bb9@vger.nsu.edu> <C0GuD4.5u@world.std.com>
- Organization: Georgia Institute of Technology
- Lines: 28
-
- In article <C0GuD4.5u@world.std.com> tne@world.std.com (Thomas N Erickson) writes:
- >g_harrison@vger.nsu.edu (George C. Harrison, Norfolk State University) writes:
- >
- >> We have been exploring using transputers for some undergraduate classes.
- >>Part of our plan was to use Ada to program the transputers, which come at
- >>very reasonable price; however, it appears that the Ada vendors that have
- >>compilers for the transputers charge at a "per node" basis.
- >> "Per Node" is fine
- >>for a network of various users, but a transputer-based platform is a single
- >>user system (in our case). Is there a compiler for transputers
- >>(on a PC) that
-
- If this is typical, I'ld say it just about drops Ada out of the massively-
- parallel computing market! What a bonanza for the compiler vendor: sell 1,000
- CPU licenses for one box! I'm _clearly_ in the wrong end of the business.
-
- Could you get _one_ vectorizing cross-compiler, and host it in something other
- than your transputer? Most licenses I've seen are predicated on the number
- of hosts, not the number of targets. Which compiler vendor(s) were treating
- each target node as a separate CPU for license purposes?
-
- Regards --jmm--
-
- --
- John M. Mills, SRE; Georgia Tech/GTRI/TSDL, Atlanta, GA 30332
- uucp: ...!{decvax,hplabs,ncar,purdue,rutgers}!gatech!prism!jm59
- Internet: jm59@prism.gatech.edu
- ... Not so fast -- I'm still thinking.
-