craig@unicus.UUCP (Craig D. Hubley) (04/10/88)
*DISCLAIMER ON* OK, if you don't like arguing about vapor then don't read this. But to make an A3000 right, there are some things that Commodore simply has to do to satisfy it's customers and the market. I don't claim to know it all, but anyone on this net could have come up with several good solutions to the A500/A1000/A2000 expansion fiasco, even after the fact, and many of us have, *when it was too late.* So if they go ahead and screw up, fine, but I said my piece. *DISCLAIMER OFF* An A3000 with 68030/68882 running at 20 MHz is a very attractive product, the more so if Motorola actually adopts a SPARC-like Unix binary standard for 680x0 microcomputers: **Use it**, Commodore. Of course, it needs at least 1Kx800 (absolute minimum) noninterlaced output, since we can now buy $900 monitors to display this. It would be nice to have it pump out HDTV as well, if a standard exists by then. Letting all the slots at the video would be great, since the machine has shown itself a strong player in the video world. Don't sell it with a `special' monitor, especially not one with a built-in deinterlacer. If I want a deinterlacer, I'll buy it separately so that I can actually *do* things with it, like deinterlace my 9-foot TV. :-) There are excellent *cheap* monitors around that will do RGB analog at 1000x800 *and more*, and if I buy one I don't want a 640x400 image. Once you're away from NTSC as your maximum resolution, you're away from it. The *ideal* situation would be to get as close as possible, at least in terms of the size of one mode, to the impending HDTV standards, such as they are, so that you could easily support HDTV later the way you did NTSC, with only minor modifications to the machine when the standard is set. Imagine the massive sales to video companies, the networks, etc... I seem to remember the Japanese `standard' is a 5x3 aspect ratio, resolution about 1200x800. This won't be used in Europe or N.A., because it's too big on bandwidth, but the eventual standard will probably be about the same size, only with the signal compressed and decompressed by a smarter TV, including deinterlacing. Now, for my beef: The rumoured expansion bus. If you produce a Zorro III standard (32-bit, access to video, etc.), then it *must* somehow be able to support a Zorro II card, *in* the cage. In order of preference: Zorro III `wrappers' that fit on a Zorro II card, a Zorro II mini-cage that fits into several Zorro III slots, even a Zorro III card that supports a separate Zorro II cage might be acceptable. At the last resort, at least make the Zorro III cage big enough that someone else can make the necessary adapters. Consider the hardware vendors that gave up on the Amiga and Commodore, the users stuck with Zorro-I hardware and two kinds of SOTS boxes, Commodore and A2000 users waiting for Zorro II hardware to show up, the current hardware vendors trying to support three different products through FCC and onto the shelves... it's just stupid to do this again. There's a standard... extend it, don't abandon it! In article <249@sagpd1.UUCP> monty@sagpd1.UUCP (Monty Saine) writes: > That was the good news now fore the not so good news. >The Person giving this very abbreviated demo ( they had a plane to catch in >30 minuites) stated that the Amiga 3000 would contain an 68030, 68882 and >run at 20 Mhz. Sounds ok so far! But the 3000 will NOT be compatable with >2000 cards. Stop and think before you flame on. The 3000 will have a 32 bit >bus and the 2000 has a 16 bit bus. Stop. Think. It is a small thing, technically, to add a $30 piece of metal to your product line, to ensure that your new card cage can accomodate old cards. Commodore didn't do it the first time (Zorro I -> Zorro II), pleading near-insolvency. I accept no such excuses from a company in the black. >get overly excited. The machine appeared to be in an 2000 chassis but the This is a hopeful sign. *FLAME ON* If Commodore produces an A3000 that cannot take Zorro II cards, I will not only not buy it but will actively proselytize against the machine. I will tell everyone who will listen: ``Do not buy this machine. This company has proven beyond a shadow of a doubt that they do not care about their customers. When they changed Zorro I to Zorro II, they claimed `circumstance', when they changed A1000 expansion connector to A500 connector and Zorro II cage, they claimed `marketing'. Most of us listened to them, and bought Zorro II cages for their A500s and A1000s. Now they have changed in again, claiming `32-bit bus'. Unlike the others, this was a purely technical matter and the solution based on obvious incompetence. The 68030 can retrieve 16-bit data, a 16-bit bus is just half of a 32-bit bus, nobody expected 32-bit performance out of a 16-bit card, but they wanted at least their older expansion gear to work while they bought new ones as they needed them, and they become available. But no, they provided a different bus, without providing a bus adapter to plug in Zorro II cards. Third parties are doing it, but without every dealer carrying them, many users will be tricked into paying big bucks for new 32-bit answering-machine cards and FAX cards, and modem cards and other things that don't have to be 32 bits. Don't buy a machine from this company.'' *FLAME OFF* > This is not the end of the world. The 3000 is targeted to compete >with *SUN* and is to be a parrallel development for the 2000 in a different >market. The 2000 (series) will still be a viable line with the 3000 targeted >at the workstation market. Oh yes the 3000 demo was running workbench 2.0 >and was said to have a non-interlaced display capability in the 1000x800 area. >This is a machine for a DIFFERENT MARKET!!!!! It is not the end of the world, but if it's true, it's the end of my patronage. This is an old IBM-trick, naming everything in the same line but keeping it so incompatible as to require new add-ons and even new software as you upgrade, and charging such outrageous prices for upgrades that you don't know whether they are adding a card or replacing the whole machine. Keeping the software model the same for the A500 to A3000 means much cheap, good, software. Look at the IBM PC line, where people pay big bucks for faster processors to run things in 8088 emulation mode! This is where the money is, folks, in having an upward-compatible line. Marketing is not an excuse for this idiocy. Don't expect the Zorro II card makers to have all their products available on Zorro III overnight, either... if true, this is the stupidest marketing- acceptance move ever seen by man. For a $30 piece of metal, and the common sense to make the cage just a little bigger than a Zorro II cage... sigh. > The key that we all need to remember before we scream is the more >(profitable) markets that Commadore is in, the better is the future growth >for ALL Amiga's. One more thing before I go, the sugggested price for this >machine was in the $4000 range with 2 Megabytes of 32 bit memory. There >was also mention of possible rebates (for those who wish to change market >branches). These guys were VERY non-commital. The display looked great >though. I don't give a shit what's good for Commodore. I'm not a stockholder. If I buy a 2000 and later buy a 3000, I see no reason to abandon my hardware. I can think of many 16-bit perhiperals that would be just as useful on a 32-bit machine. If I *must* abandon them, then I will, but then I'll go shopping for a new machine... After all, I have no hardware investment to protect, have I ? And if my software is Unix, I know it will work anywhere. Wise up, Commodore. That is, if you really sent this joker around telling people that Zorro II won't fit into Zorro III. I hope not. Craig Hubley, Unicus Corporation, Toronto, Ont. craig@Unicus.COM (Internet) {uunet!mnetor, utzoo!utcsri}!unicus!craig (dumb uucp) mnetor!unicus!craig@uunet.uu.net (dumb arpa)
rchampe@hubcap.UUCP (Richard Champeaux) (04/12/88)
In article <2503@unicus.UUCP>, craig@unicus.UUCP (Craig D. Hubley) writes: > > If you produce a Zorro III standard (32-bit, access to video, etc.), > then it *must* somehow be able to support a Zorro II card, *in* the cage. > > In article <249@sagpd1.UUCP> monty@sagpd1.UUCP (Monty Saine) writes: > >2000 cards. Stop and think before you flame on. The 3000 will have a 32 bit > >bus and the 2000 has a 16 bit bus. > > based on obvious incompetence. The 68030 can retrieve 16-bit > data, a 16-bit bus is just half of a 32-bit bus, nobody expected > 32-bit performance out of a 16-bit card, but they wanted at least > > Craig Hubley, Unicus Corporation, Toronto, Ont. Your right, if the 68030 is anything like the 68020, it is capable of dynamic memory port sizing, and would have no problem addressing a 16-bit card. However, if that card has DMA on it connected to some I/O device, then it would need to be able to address the 32-bit memory that is most likely to exist in an A3000. Getting a card that is connected only to D0-D15 of the address bus to access the odd addressed words connected to D16-D31 would require extra hardware on each of the 32-bit memory cards that would allow word swapping. It might be possible to put such hardware on the system board, but that would make it really complicated, and probably slow up the bus. Another idea that was suggested by someone a few messages ago, was to include some 16-bit memory for the 16-bit cards to use, but to do that, you would have to require software wishing to use such devices, to place anything it wants DMA'ed in the 16-bit memory. Now your hardware is compatable, but your software doesn't work anymore. Anyways, if the A3000 is indeed aimed at the workstation market, and has the accompanying price of $4000 plus, then the A2000 will still be a very popular computer (probably more popular than the A3000). I hardly think that CA will drop, or that people will stop buying the A2000, so you won't have to worry about companies dropping their Zorro-II cards. Rich Champeaux Clemson University
craig@unicus.UUCP (Craig D. Hubley) (04/13/88)
In article <1359@hubcap.UUCP> rchampe@hubcap.UUCP (Richard Champeaux) writes: >In article <2503@unicus.UUCP>, I write: >> If you produce a Zorro III standard (32-bit, access to video, etc.), >> then it *must* somehow be able to support a Zorro II card, *in* the cage. I stand by this statement. > Your right, if the 68030 is anything like the 68020, it is capable of >dynamic memory port sizing, and would have no problem addressing a 16-bit >card. However, if that card has DMA on it connected to some I/O device, >then it would need to be able to address the 32-bit memory that is most likely >to exist in an A3000. Getting a card that is connected only to D0-D15 of the >address bus to access the odd addressed words connected to D16-D31 would >require extra hardware on each of the 32-bit memory cards that would allow >word swapping. It might be possible to put such hardware on the system board, >but that would make it really complicated, and probably slow up the bus. But you could always *bypass* this stuff when it wasn't being used, and not slow up the bus. I never said there wouldn't be a performance hit for this, but the bus and processors will be faster anyway. And you could always go my `third option' of supporting a separate cage (perhaps a 2000-and-1). >Another idea that was suggested by someone a few messages ago, was to include >some 16-bit memory for the 16-bit cards to use, but to do that, you would have >to require software wishing to use such devices, to place anything it wants >DMA'ed in the 16-bit memory. Now your hardware is compatable, but your >software doesn't work anymore. Remember that you probably already have a 16-bit memory card, in fact you definitely have one if you have expanded your machine at all. Why not just use *that* memory ? Anyway, an answer to the software question is to have FAST memory be 16-bit, by definition, and FASTER memory, or something, be 32-bit. You would want to recompile your software under a 32-bit environment anyway, so requiring such a change is no big deal, and everything else would work in the meantime. Since it's all written in C, it probably makes tons of assumptions of sizing anyway, and would need to be adjusted to work. The Amiga nicely supports 68000-020, but I'm not sure if it extends as far as a 32-bit bus, without needing recompilation. Anyone knowledgeable want to comment. You could always have a `give 16-bit memory only' mode for those incredibly sensitive applications, which could be adjusted to make the ] distinction later. I know this sounds Intel-ish, but look how many machines get sold with `this will work as is, no changes' 8088 emulation mode. The smart folks would bypass it with code that took advantage of the new goodies, and in the meantime the old stuff would all work, adn *be cheap*. > Anyways, if the A3000 is indeed aimed at the workstation market, and has >the accompanying price of $4000 plus, then the A2000 will still be a very >popular computer (probably more popular than the A3000). I hardly think that >CA will drop, or that people will stop buying the A2000, so you won't have >to worry about companies dropping their Zorro-II cards. Right. So now I must buy new Zorro III cards to replace all my old Zorro II cards that work just fine, taking a loss selling my old ones. This is still nasty. I'm not worried about dropping Zorro II, I'm worried about the wait as people decide whether or not to start up Zorro III design/production. Expect a year wait before the selection is as good as an A2000, and in the meantime the machine may die. I'd probably buy a SPARC machine instead, since I would have to replace my whole configuration anyway, might as well use what AT$T, $un, and Xerox use. After all,they write the software that I want to use. And eventually, companies *would* drop their Zorro II cards. More importantly, I'd be afraid to buy anything new for my Zorro II machine for fear that the replacement cost when I moved up to an A3000 (still running Unix) would be prohibitive. I'd save my money for the new one, and do without the capabilities in the meantime. Good for no one. My argument stands. >Rich Champeaux >Clemson University I still think this is a month's engineering effort for a design team, and maybe $50 of hardware and a few badckwards-compatible software tricks, to gain many potential customers and save everybody upgrading a year's wait. A smart company would do it, at least as an option. Craig Hubley, Unicus Corporation, Toronto, Ont. craig@Unicus.COM (Internet) {uunet!mnetor, utzoo!utcsri}!unicus!craig (dumb uucp) mnetor!unicus!craig@uunet.uu.net (dumb arpa)
peter@sugar.UUCP (Peter da Silva) (04/14/88)
I hope that Commodore isn't going to screw something up that even IBM got right. Look at the IBM AT bus. It's basically the XT bus with some extra lines. Look at the Zorro-2 bus. It's got room to put those extra lines in. They'd have to be left out of one of the bridge slots (you are going to have bridge slots in the 3000, aren't you?), but the other bridge slot and the Zorro-2 slots could have a secondary slot just like that ol' AT's. There's the extra slots you need for the 32-bit bus, and nobody gets screwed. And if you do it this way, someone will be able to build a 2000-and-3 card that gives you those lines and connects them to a magic connector on their 68030 card. On another point... have you seen the coverage of the new Sun 386 machine. We've been able to "cut and paste between MS-DOS and..." in this case AmigaDOS for months. Welcome to the future. -- -- Peter da Silva `-_-' ...!hoptoad!academ!uhnix1!sugar!peter -- "Have you hugged your U wolf today?" ...!bellcore!tness1!sugar!peter -- Disclaimer: These aren't mere opinions, these are *values*.
rchampe@hubcap.UUCP (Richard Champeaux) (04/14/88)
In article <2520@unicus.UUCP>, craig@unicus.UUCP (Craig D. Hubley) writes: > In article <1359@hubcap.UUCP> rchampe@hubcap.UUCP (Richard Champeaux) writes: [deleted argument and counter-arguement about how much effort it would require to allow the A3000 to use ZorroII cards ] > > > Anyways, if the A3000 is indeed aimed at the workstation market, and has > >the accompanying price of $4000 plus, then the A2000 will still be a very > >popular computer (probably more popular than the A3000). I hardly think that > >CA will drop, or that people will stop buying the A2000, so you won't have > >to worry about companies dropping their Zorro-II cards. > > Right. So now I must buy new Zorro III cards to replace all my old Zorro II > cards that work just fine, taking a loss selling my old ones. This is still > nasty. I'm not worried about dropping Zorro II, I'm worried about the wait > as people decide whether or not to start up Zorro III design/production. > Expect a year wait before the selection is as good as an A2000, and in the > meantime the machine may die. I'd probably buy a SPARC machine instead, > since I would have to replace my whole configuration anyway, might as well > use what AT$T, $un, and Xerox use. I said that in my opinion the A2000 would continue to be fully supported because the A3000 will probably be in a higher price range and therefore wouldn't be bought by the average home computer user. (After all, how many people do you know who have a workstation in their living room.) In other words, I expect CA to continue making the A2000 to sell in the home market and I would therefore expect other companies to continue making ZorroII cards. Assuming that to be true, then you would have no reason, other than to get higher performance, to upgrade to an A3000. In this case, upgrading to higher performance (even though your current machine will continue to be fully supported) will include the price of new ZorroIII boards. Now if CA dicides that the A3000 is going to be it's sole computer, and stops making the A2000, then your complaint has merit, because you'd be left with an unsupported machine with no possibility of upgrade. (This next example is exaggerated to make a point.) If CA came out with a Cray clone for $4000-$5000 yet still sold the A2000, would you complain that you couldn't use your ZorroII cards in it? > And eventually, companies *would* drop their Zorro II cards. Like I said, companies will drop their ZorroII cards only if CA stops selling and supporting the A2000, and in my opinion, they will continue to support it. > > Craig Hubley, Unicus Corporation, Toronto, Ont. > craig@Unicus.COM (Internet) > {uunet!mnetor, utzoo!utcsri}!unicus!craig (dumb uucp) > mnetor!unicus!craig@uunet.uu.net (dumb arpa) Rich Champeaux Clemson University
craig@unicus.UUCP (Craig D. Hubley) (04/16/88)
In article <1401@hubcap.UUCP> rchampe@hubcap.UUCP (Richard Champeaux) writes: >words, I expect CA to continue making the A2000 to sell in the home market >and I would therefore expect other companies to continue making ZorroII cards. >Assuming that to be true, then you would have no reason, other than to get >higher performance, to upgrade to an A3000. In this case, upgrading to higher >performance (even though your current machine will continue to be fully >supported) will include the price of new ZorroIII boards. Now if CA decides Getting higher performance is the usual reason to upgrade, no surprise there. But the price of upgrading does not have to be artifically inflated by the price of buying new cards. This is pretty dumb from a corporate standpoint. After all, what reason do I have to upgrade to an `Amiga 3000' and not a `Mac II' or a `SPARCintosh' ? My software is Unix, my hardware is dead, I have to replace my whole system anyway... you see what I mean ? *This* would be yet another deep marketing fuckup. And the only way I see around it is to offer another tradein deal. That'll hardly be cheaper than the minor adjustments I and others have suggested. >that the A3000 is going to be it's sole computer, and stops making the A2000, >then your complaint has merit, because you'd be left with an unsupported >machine with no possibility of upgrade. That's possible, and it certainly happened to the A1000, but I don't expect it. Perhaps that's my naivete showing through, and a foolish trust of the company that made my first two computers. >(This next example is exaggerated to make a point.) >If CA came out with a Cray clone for $4000-$5000 yet still sold the A2000, >would you complain that you couldn't use your ZorroII cards in it? I would if they called it an Amiga. >Rich Champeaux >Clemson University Craig Hubley, Unicus Corporation, Toronto, Ont. craig@Unicus.COM (Internet) {uunet!mnetor, utzoo!utcsri}!unicus!craig (dumb uucp) mnetor!unicus!craig@uunet.uu.net (dumb arpa)
jdg@elmgate.UUCP (Jeff Gortatowsky) (04/17/88)
I hope CBM looks very closely at the market they *MIGHT* target for the A-3000. Many of the current ST/Amiga owners use workstations (SUN/APOLLO/MIPS/uVAX etc..) at work. I know that since my exposure to Sun's computers, my idea of a "Personal Computer" has changed quite a bit. You know the 'type'. Guy/Gal gets a Sun 3/160 (8meg, 1-280meg disk and 2 71meg disks) plopped in thier office as 'his/her' system. As I grew to accept this machine as MY system, my views of my A-1000 (2 floppies and 1meg RAM) became somewhat jaded. Before I don my aircraft fire fighting suit understand I like my A-1000 a lot. It just having been exposed to a system like the 3/160m 40+ hours a week I begin to *EXPECT* that as the computing environent *ALL* computers should support, including the computer at home. I think it's called being spoiled. I know I can't afford even a fire sale Sun 2/50 with a 71meg 'shoebox' for home. But I would give anything (you know what I mean) for a "home workstation". The question to CBM marketing is "Am I the norm? Or the exception?". Or better yet, "Am I becoming the norm? The norm for the near future?". If I'm "BECOMING" the norm then you (CBM) should look very hard at the price of any future machines. There may be more private people willing to pay $2500-$3000 for a color workstation from CBM than Fortune 1000 companies willing to pay $5000+ from CBM. Just a thought....... PS. The nebulous "workstation" I'm referring to means, UNIX or some multi-tasking protected VM OS, >800 x >600 color or mono rez display, some kind of card cage, and a memory sub-system that incurs 0 to 1 wait states even if not on the motherboard. Hardware graphics support (ie windowing) available if not standard. -- Jeff Gortatowsky .....allegra!rochester!kodak!elmgate!jdg Eastman Kodak Company These comments are mine alone and not Eastman Kodak's. How's that for a simple and complete disclaimer?
morgan@brambo.UUCP (Morgan W. Jones) (04/20/88)
In article <913@elmgate.UUCP> jdg@aurora.UUCP (Jeff Gortatowsky) writes: >PS. The nebulous "workstation" I'm referring to means, UNIX or some >multi-tasking protected VM OS, >800 x >600 color or mono rez display, >some kind of card cage, and a memory sub-system that incurs 0 to 1 >wait states even if not on the motherboard. Hardware graphics support >(ie windowing) available if not standard. Holy deja vu, Batman! The Joker's missed the biggest scandal of the quarter. C-A has announced the 2620, a 68020 with MMU, and Unix to run on it. Rumour has it that 1.4 will handle VM. The new monocrome monitor will support 1000x800 in 4 scale and doesn't require special hardware. A2000 has a card cage. 2620 will use static ram (2 or 4 M). (I think, but I'm not sure) The 1000 has hardware grahics support for windowing and graphics. >Jeff Gortatowsky .....allegra!rochester!kodak!elmgate!jdg -- Morgan Jones - Bramalea Software Inc. morgan@brambo.UUCP ...!{uunet!mnetor!lsuc!ncrcan, utgpu!telly}!brambo!morgan "These might not even be my opinions, let alone anyone else's."
daveh@cbmvax.UUCP (Dave Haynie) (04/21/88)
in article <1359@hubcap.UUCP>, rchampe@hubcap.UUCP (Richard Champeaux) says: > Summary: Is it as easy as you claim? > You're right, if the 68030 is anything like the 68020, it is capable of > dynamic memory port sizing, and would have no problem addressing a 16-bit > card. However, if that card has DMA on it connected to some I/O device, > then it would need to be able to address the 32-bit memory that is most likely > to exist in an A3000. Getting a card that is connected only to D0-D15 of the > address bus to access the odd addressed words connected to D16-D31 would > require extra hardware on each of the 32-bit memory cards that would allow > word swapping. It's not all that complicated. We do it on the A2620 card. When a 16 bit DMA occurs to 32 bit memory, odd words go to one half of the data bus, even words go to the other half. It required exactly two extra 8 bit buffer chips and one additional buffer control line. It might, however, get more complex when such a setup is running on an expansion bus. On the A2620, we have the advantage that all transfers are happening through our controlled buffers. On a bus that shares 16 and 32 bit devices, this is no longer the case. A 16 bit device may want to address a 32 bit device. That necessarily places the burden of word shifting on the 32 bit device instead of the motherboard system, as long as the shared bus is actually sharing data lines. So there's some additional logic required, probably the same buffers I use on the A2620, but they'd be required on every 32 bit card unless a better scheme can be discovered. > Rich Champeaux > Clemson University -- Dave Haynie "The B2000 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!"