gregb15@pro-charlotte.cts.com (Greg Beckham) (06/25/88)
I have a suggestion for the new graphics chip (really obese agnus?)... To have a graphics mode with 256 colors on screen from a pallette of 4096. Interlace or Non-Interlace. Nuff' Said. Greg Beckham UUCP: [ ihnp4 cbosgd sdcsvax nosc ] !crash!pro-charlotte!gregb15 ARPA: crash!pro-charlotte!gregb15@nosc.mil INET: gregb15@pro-charlotte.cts.com
cmcmanis%pepper@Sun.COM (Chuck McManis) (06/27/88)
In article <3150@crash.cts.com> (Greg Beckham) writes: > I have a suggestion for the new graphics chip (really obese agnus?)... To >have a graphics mode with 256 colors on screen from a pallette of 4096. >Interlace or Non-Interlace. Nuff' Said. > Greg Beckham An excellent suggestion, except that 256 - 12 bit color registers is roughly 25,000 transistors, which in a 3 micron techonology is 225 square mils and that may not fit into Denise any more. There is a way however, and that involves taking all of the D/A stuff out of Denise and putting it into a separate RAMDAC. (Brooktree makes a really nice one, Apple uses it I believe). Then, one could make the video color resolution what ever they wanted. Just depends on how much they want to spend on the DAC. At the higher resolutions these things get pretty hot so we might need fins on the back of the case. The original suggestion though bears repeating, lets get more colors even if it is only in the 'high end' machines. --Chuck McManis uucp: {anywhere}!sun!cmcmanis BIX: cmcmanis ARPAnet: cmcmanis@sun.com These opinions are my own and no one elses, but you knew that didn't you.
oconnor@csb8.steinmetz (Dennis M. O'Connor) (06/29/88)
An article by cmcmanis@sun.UUCP (Chuck McManis) says:
] In article <3150@crash.cts.com> (Greg Beckham) writes:
] > I have a suggestion for the new graphics chip (really obese agnus?)... To
] >have a graphics mode with 256 colors on screen from a pallette of 4096.
] >Interlace or Non-Interlace. Nuff' Said.
] > Greg Beckham
]
] An excellent suggestion, except that 256 - 12 bit color registers is
] roughly 25,000 transistors, which in a 3 micron technology is 225
] square mils and that may not fit into Denise any more.
(please read as if said with "incredulous surprise", not "flaming")
"3 micron" technology ? Are you kidding ? GE's semi-custom ASICs
are available in 1.25 micron. Performance Semiconductor or
IDT ( Integrated Device Technologies ) are, I believe, offering
1 micron foundry services. Even MOSIS offers a "1.25 micron"
( really equivalent to 2 micron ) service.
Are the new Amiga chips really being done in a technology that's been
obsolete for at least three or four years ? I know Commodore has it's
own in-house fab lines it wants to use, but haven't they put any money
into improving their process lately ? 1.25 micron would allow 5 times
as many devices in the same area.
Smaller-dimensioned circuits are denser, cooler, and faster. I
can't tell you GE's yeilds, but its surprisingly high for
150,000 transistor chips ( that info is Company Confidential ).
And available to Commodore if they want to use it. I hope the
Amiga doesn't become a victim of the NIH syndrome.
Sorry if this seems like an add for GE Semiconductor : I didn't
mean it to be. But I can't see why the Amiga's new chips would
be limited by obsolete technology, when much better is available
to anyone who wants it. But of course, I have NO idea of the
factors involved in C-A's decision making, so this is NOT a flame.
--
Dennis O'Connor oconnor%sungod@steinmetz.UUCP ARPA: OCONNORDM@ge-crd.arpa
"Never confuse USENET with something that matters, like PIZZA."
daveh@cbmvax.UUCP (Dave Haynie) (07/02/88)
in article <11415@steinmetz.ge.com>, oconnor@csb8.steinmetz (Dennis M. O'Connor) says: > "3 micron" technology ? Are you kidding ? ... > Are the new Amiga chips really being done in a technology that's been > obsolete for at least three or four years ? I know Commodore has it's > own in-house fab lines it wants to use, but haven't they put any money > into improving their process lately ? 1.25 micron would allow 5 times > as many devices in the same area. The current Amiga chips were done in old NMOS. So based on that, you would have lots of trouble adding in many of the things that have been proposed here on the net. Then again, we don't do any new designs in 3 micon NMOS. However, you can't just take a complex NMOS design, at any level, and crank out the same thing in 1.25 micron CMOS. You might do better changing processes with ASICs, since most commercial ASICs are standard cell or gate arrays, and are basically process independent. But a full custom design is a mass of transistors and a by-hand layout (unless you've got a silicon compiler). Most layouts aren't scalable, though folks are moving in that direction. In any case, to put the Amiga chips, as is, into a smaller process, you'd have to do lost of new layout. If you're going to CMOS as well (and you would be), that's a complete circuit redesign. > Dennis O'Connor oconnor%sungod@steinmetz.UUCP ARPA: OCONNORDM@ge-crd.arpa > "Never confuse USENET with something that matters, like PIZZA." -- Dave Haynie "The 32 Bit Guy" Commodore-Amiga "The Crew That Never Rests" {ihnp4|uunet|rutgers}!cbmvax!daveh PLINK: D-DAVE H BIX: hazy "I can't relax, 'cause I'm a Boinger!"
tws@beach.cis.ufl.edu (Thomas Sarver) (07/02/88)
In article <3150@crash.cts.com> gregb15@pro-charlotte.cts.com (Greg Beckham) writes: > > I have a suggestion for the new graphics chip (really obese agnus?)... To >have a graphics mode with 256 colors on screen from a pallette of 4096. >Interlace or Non-Interlace. Nuff' Said. > > Greg Beckham > Make that a pallette of 16 million. Its only 24-bit pallette registers. I agree that 8 bit planes are a minimum for the 1990's. Gotta keep up with the Mac II, even if were already superior from the inside out. BTW, how about the setup the Mac II uses, have one card do multi-modes and just add memory when you want more bit-planes. Economical, let the consumer decide how much power he/she wants. +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ But hey, its the best country in the world! Thomas W. Sarver "The complexity of a system is proportional to the factorial of its atoms. One can only hope to minimize the complexity of the micro-system in which one finds oneself." -TWS Addendum: "... or migrate to a less complex micro-system."
doug-merritt@cup.portal.com (07/04/88)
Dave Haynie writes: >The current Amiga chips were done in old NMOS. [...] you can't just take a >complex NMOS design, at any level, and crank out the same thing in 1.25 >micron CMOS [...] a full custom design is a mass of transistors and a >by-hand layout (unless you've got a silicon compiler). Most layouts >aren't scalable ^^^^^^^^^^^^^^^^ ^^^^^^^^ This is comparable to the question of whether to write software in assembler or in C. It's taken quite some years, but by now most companies realize that by default, C should be first choice, with assembler used only when absolutely necessary. Similarly there has been quite a lot of resistence to using anything but by-hand random logic for silicon. There are many people who sneer at the mere notion of using even so much as (scalable!) Mead-Conway design rules. For some cases, this makes sense, just as assembler sometimes does... sometimes you need to squeeze as many gates as possible onto as little real estate as possible, etc. But not as often as some people think. In this day and age, it makes all kinds of sense to use a silicon compiler as first preference. And they *are* widely available. They can save vast amounts of development time. The resulting silicon tends to have many fewer errors, reducing the number of custom foundry runs needed to get a viable chip, and reducing the support costs otherwise incurred by chips in the field with subtle problems. And you can always hand-optimize the layout after you've got working silicon if you want to reduce the size or increase the speed or some such. I'm sure Dave knows all this; I just cringe when I hear things like "*if* you've got a silicon compiler". Everyone should have one! (Everyone? Well, hobbyists need a Heathkit ASIC Foundry, too, I suppose :-) Doug -- Doug Merritt ucbvax!sun.com!cup.portal.com!doug-merritt or ucbvax!eris!doug (doug@eris.berkeley.edu) or ucbvax!unisoft!certes!doug