Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!usc!apple!fernwood!portal!cup.portal.com!Fletcher From: Fletcher@cup.portal.com (fletcher sullivan segall) Newsgroups: comp.sys.amiga.hardware Subject: Re: Amiga Custom Chips - why hasn't C= made them faster? Message-ID: <41008@cup.portal.com> Date: 7 Apr 91 06:07:02 GMT References: <1991Apr2.235710.13984@news.iastate.edu> Distribution: na Organization: The Portal System (TM) Lines: 68 >In article , cpetterb@glacier . >sim.es.com (Cary Petterborg) writes: >>The Amiga has been out for years now. There have been improvements >>made to the custom chips, Agnus in particular. But, I am amazed at >>the fact that their clock speed, etc. has remained the same. In an >>industry where last years chip runs twice as fast this year, C= sure >>has sat back on their b*tts as far as performance is concerned. Is >>it because they aren't willing to invest any more money into the >>technology because they can sell so many A500's as they are now? >>What gives? Certainly the technology exists to speed them up. > > Here is why Commodore has not improved the Amiga's custom chipset. The >following table lists the amounts that Commodore has invested in R&D since >1984, both as a raw figure and as a percentage of total sales (the better >figure for comparing companies): > >... > > As you can see, compared to SUN, Commodore has never invested much >in their own future, which is the reason why the custom chipset is so >ancient, and why improvements are nowhere in sight. This information >also gives us some insight into the success of the two companies. As >you can see from the first table, between 1984 and 1990, Commodore's >total sales steadily declined. The decline is actually much worse than >it appears, once you take inflation into account. > You are forgetting the basic history of the Amiga. Commodore didn't design the chips in the Amiga, and wouldn't be able to afford to design an equialent chip set from scratch in any case. The custom chips were hand optimized in NMOS for a home video game. As such they were designed to be as economical as possible on a very large scale production. The chip set is considerably more complex than the most advanced display cards available for the PC... ...but they are also essentially cast in stone. Redesigning in CMOS would require re-optimizing, or a loss of performance. Adding features has to be tested thoroughly because the chips are extremely interdependent, and any introduced bug could well affect the whole system. Expanding to 32 bits would require a hefty redesign of the chips to remain consistent with earlier versions. ... It should also be noted that Sun didn't put the SPARC chips in silicon, they only wrote the specifications for the chip, and allowed other manufacturers to build it. (Not impossible for commodore if they choose to give away the existing chip masks (so that they can be modified by other companies) but it isn't likely that commodore would want to release something that valuable for free.) Additionally, RISC processors are the easiest type of processor to implement in silicon, so spotting actual implementations after 4 years is quite reasonable. -F. Sullivan Segall _______________________________________________________________ /V\ E-Credibility: (n -- ME) The unguaranteed likelyhood that ' the electronic mail you are reading is genuine rather than someone's made up crap. _______________________________________________________________ Mail to: ...sun!portal!cup.portal.com!fletcher or fletcher@cup.portal.com