Re: Comparing Compilers

28 Nov 2004 23:23:17 -0500

          From comp.compilers

Related articles
Comparing Compilers (G P) (2004-11-06)
Re: Comparing Compilers (TOUATI Sid) (2004-11-14)
Re: Comparing Compilers (G P) (2004-11-14)
Re: Comparing Compilers (G P) (2004-11-17)
Re: Comparing Compilers (A Pietu Pohjalainen) (2004-11-17)
Re: Comparing Compilers (Dick Weaver) (2004-11-17)
Re: Comparing Compilers (TOUATI Sid) (2004-11-28)
Re: Comparing Compilers (2004-12-13)
| List of all articles for this month |

From: TOUATI Sid <>
Newsgroups: comp.compilers
Date: 28 Nov 2004 23:23:17 -0500
Organization: Universite de Versailles Saint-Quentin-en-Yvelines
References: 04-11-015 04-11-033 04-11-050
Keywords: performance, practice
Posted-Date: 28 Nov 2004 23:23:17 EST

>>The same remark for code size : who cares if you generate a code of 4.9
>>Ko instead of 4 ko. However, I think that many people would care if you
>>generate 4.9 Go instead of 4 Go.
> People compiling Java to mobile phones (J2ME) actually do care on
> savings of one kilobyte in the size of the compiled code. This is
> because the phone manufacturers, especially on the low-end models
> limit the maximum size of the deployable program to e.g. 30kB.

They would care only if saving 0.9 kb of code size allows them to use
smaller memories. Since the sizes of the memories in the market are
mostly power of 2, building an application of 3 or 3.9 kb requires to
use a memory of 4kb in both cases.

In embedded systems, the problem is not to optimize an application given
an architecture, but to optimize an architecture given an application.
In other words, compiling for smallest codes wouldn't necessarily comes
up with smallest architectures.


Post a followup to this message

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