[plt-scheme] Scheme productivity data point

From: Matthias Felleisen (matthias at ccs.neu.edu)
Date: Sun Nov 8 16:28:21 EST 2009

Plus he may have needed the conversion once and only once. So who  
cares about the performance.

Here is a similar story from a scientific computation group at  
Cornell, working Boeing. (Zippel is my source.)

They developed a Lisp-based framework for transforming continuous  
mathematics for wings into discrete and then into Fortran programs.  
They reduced the development time for Cray Fortran programs from a  
year to a month. Then they ran the program, which took another month  
(and a bit less). Then they had the results and could throw away the  
program. Sounds like a winner? Sadly a month of Cray time was more  
expensive in 1992 than 11 month of Fortran programmer time. They had  
lost -- Matthias





On Nov 8, 2009, at 3:43 PM, Noel Welsh wrote:

> On Sun, Nov 8, 2009 at 5:38 PM, Jon Rafkind <rafkind at cs.utah.edu>  
> wrote:
>> The input files were HTML and the output files were something.. but  
>> does
>> that have anything to do with the fact that you used scheme? I  
>> don't see how
>> the engine is a factor here, unless I missed something.
>
> The issue is total time to complete the job. Neil is saying that w/  
> Scheme
>
> 1. The Scheme program was on ~2x slower than the time taken to just
> shift the bytes. I.e. runtime overhead was low
> 2. More importantly, development time was only 2 days. With C++ dev.
> time might well have been a week for very little performance
> improvement.
>
> N.
> _________________________________________________
>  For list-related administrative tasks:
>  http://list.cs.brown.edu/mailman/listinfo/plt-scheme



Posted on the users mailing list.