Re: Compiler for "own" parallel language?

nmm1@cus.cam.ac.uk (Nick Maclaren)
29 Oct 2005 21:26:09 -0400

          From comp.compilers

Related articles
Compiler for "own" parallel language? GoodIntentions@lycos.de (Matrim) (2005-10-27)
Re: Compiler for "own" parallel language? nmm1@cus.cam.ac.uk (2005-10-29)
Re: Compiler for "own" parallel language? shreyas76@gmail.com (shrey) (2005-10-29)
| List of all articles for this month |

From: nmm1@cus.cam.ac.uk (Nick Maclaren)
Newsgroups: comp.compilers
Date: 29 Oct 2005 21:26:09 -0400
Organization: University of Cambridge, England
References: 05-10-191
Keywords: parallel

  Matrim <GoodIntentions@lycos.de> wrote:
I'm about to implement a compiler a parallel language. More specific
>its a (small) subset of the Fortran-Language. In my studies I came
>across the compiler tool Eli.
>
>So now my question: Is it possible with Eli to achieve that goal
>(implementing a compiler for that a parallel language)? Any hints to
>do so? Or is there a even better way/tool to do so?
>
>[It sort of depends on what you mean by a parallel langugage, but in
>general the tools automate the front end of a compiler, and the bits
>specific to parallelism tend to reside in the back end. -John]


Yes. The extreme example of that is a language like those used in
engineering or statistics (think APL or Matlab, but there were and are
thousands), where the basic objects in the language may be matrices.
Compiling such a language is very routine, and the only specialist
part is writing the service functions.


The more interesting languages bring parallelism into the language
proper, but I am afraid the question shows that the poster may be
heading for trouble. Parsing such languages is trivial; even code
generating them is not hard; the killer is analysing the program to
detect/avoid/handle aliasing.


Regards,
Nick Maclaren.


Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.