[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference ricks::dechips

Title:	Hudson VLSI
Notice:	For Digital Chip Data - CHIPBZ::PRODUCTION$:[DS_INFO...]
Moderator:	RICKS::PHIPPS

Created:	Wed Feb 12 1986
Last Modified:	Fri Jun 06 1997
Last Successful Update:	Fri Jun 06 1997
Number of topics:	701
Total number of notes:	4658

680.0. "Java and Alpha" by POLAR::MOKHTAR () Mon Apr 14 1997 16:12

    
    Hi,
    
    after reading a few articles on Java, it seems there is potential that 
    Java will pick up momentum. In such case future software will be not 
    be tied with Intel machine code and be totaly architecture independent.
    
    this seems like a boost to Alpha.
    
    is there any work within the company to prepare for this possibility ?
    for example work to write optimized Java interpreter for Alpha-NT
    
    maged

T.R	Title	User	Personal Name	Date	Lines
680.1		AXEL::FOLEY	http://axel.zko.dec.com	`Mon Apr 14 1997 16:23`	5
	Check with Marty Jack. mike
680.2		CXXC::REINIG	This too shall change	`Mon Apr 14 1997 16:39`	4
	See http://sdtad.zko.dec.com/pub/JavaGroup/ August
680.3		QUARK::LIONEL	Free advice is worth every cent	`Tue Apr 15 1997 10:54`	5
	Java is not "totally architecture independent" - it has a lot of implicit architectural assumptions, including a requirement for IEEE floating point arithmetic. Steve
680.4	Two definitions of "independent"	STEVEN::hobbs	Steven Hobbs	`Tue Apr 15 1997 12:40`	23
	Re: .3 > Java is not "totally architecture independent" Java is not architecture independent from a performance point of view. Requiring IEEE floating point will lead to slow performance on systems that use a hardware floating point format different from IEEE FP. However, requiring IEEE FP (and outlawing the results of other floating point formats) guarantees that all Java systems get the same floating point results. Indeed, the Java specification does not allow access to IEEE features which permit an implementation option. For example, Java does provide access the IEEE exception flags because the IEEE standard allows an implementor to choose one of 4 different definitions for setting the underflow flag. Also, the Intel x86 architecture is at a disadvantage because Java does not permit results with extended range and precision while arithmetic on the Intel chip is always done with an extended range option. Intel requires extra instructions after many floating point operations to guarantee the result remains in the range specified by "The Java(tm) Language Specification". The intent of the Java specification is that the results of a Java program are "total architecture independent".
680.5	What he meant to say	WIBBIN::NOYCE	Pulling weeds, pickin' stones	`Tue Apr 15 1997 14:18`	7
	Minor typo in .4... not > For example, Java does provide access the IEEE exception flags because the ^ > IEEE standard allows an implementor to choose one of 4 different > definitions for setting the underflow flag.
680.6		60675::nessus.cao.dec.com::Mayne	A wretched hive of scum and villainy	`Tue Apr 15 1997 18:28`	7
	If you hunt round some magazine articles, you'll see various comments about Java applications that run differently, or not at all, on different platforms. So much for platform independence. See also the rather interesting numeric thread in the Java conference. PJDM