[gclist] Re: Naive Java question: Array index checking

Toon Moene toon@moene.indiv.nluug.nl
18 Apr 1997 15:48:57 GMT

[The following was posted by Toon Moene to comp.arch re array index checking
-- hbaker]

zalman@netcom.com (Zalman Stern) writes:
> That you don't see runtime checking used more in professional programming
> practice has more to do with attitudes of professional programers and
> priorities of software development tools makers than with hardware
> features.

Depends.  We've used run-time checking in running operational weather 
forecasting models.  The specific example is Digital's Fortran 77 compiler, 
but I'm sure this will extend to other vendors.  The overhead generated by 
using executables build (exclusively) from Fortran sources compiled with 
array bound checking on was indistinguisable from time differences between 
two subsequent runs of the executable (sans checking) with the same in- and 
output files.

To me, that _defines_ "zero overhead".

Not to mention that it pointed out to me that using a certain choice of 
advection scheme coupled with time step length and interaction type between 
sub-grid scale phenomena and large-scale physics was invalid:  The tables we 
use to derive saturation pressure dependent on temperature and pressure were 
addressed out-of-bounds for some choices (meaning that the local temperature 
went over 373 K, i.e. 100 degrees C, certainly unphysical).

Toon Moene (mailto:toon@moene.indiv.nluug.nl)
Saturnushof 14, 3738 XG  Maartensdijk, The Netherlands
Phone: +31 346 214290; Fax: +31 346 214286
g77 Support: mailto:fortran@gnu.ai.mit.edu; NWP: http://www.knmi.nl/hirlam