To: bob who wrote (5726 ) 10/8/1998 6:58:00 PM From: bob Read Replies (1) | Respond to of 8581
All, How about this one from USENET : Re: Infoworld: "Sun scrapping Java chip plans, analyst says " more options Author: needleman Email: needleman@my-dejanews.com Date: 1998/10/08 Forums: comp.os.os2.advocacy more headers author profile view thread -------------------------------------------------------------------------------- In article <zbvyrwgrevozarg.f0fvhh0.pminews@news3.ibm.net>, "Moises Lejter" <Moi_Lejter@ibm.net> wrote: > On 6 Oct 1998 23:14:54 GMT, Robato Yao wrote: > > > I have just been waiting for this to happen. A chip running the > > pseudocode of a high level language is not going to be efficient. > > Projects from Intel and Japan trying to do so on the ADA and Prolog > > languages have all fallen on the way side. It would have simply been > > better and faster to just compile the stuff for RISC chips, like > > MicroSparc. > > Well - not quite "the pseudo-code of a high-level language", perhaps - > but I would agree... History had shown before (I was thinking of the > Common Lisp processors on the Symbolics line...) that cpus designed > for a particular language will not "out-process" a mainstream processor > for long - advances on CPU/compiler optimization appear on mainstream > processors much faster, and volume manufacturing drops prices much > faster, than the special-purpose processors can beat... > > But I wonder why the picoJava design is so "big"... I thought I heard of > a company that implemented a Java VM on top of a Forth interpreter, > and it was pretty small/fast.... Anyone have a name for this? > > Moises > > Moises, The PSC1000 processor from Patriot Scientific Corp. (www.ptsc.com) is a FORTH machine that is both stack and register based. It is small, fast, inexpensive (<$10 in volume) and runs java at native speeds. It is a 32-bit RISC processor with an internal clock of 150MHz in the .35m version. Needleman