Kalvan Posted March 3, 2010 Share Posted March 3, 2010 James D. Said that the changes my timeline works will have unpredictable side effects. Here's one of them: It is late a August Friday evening in 1984 in a timeline that had diverged from ours in 1972. By shearest coincidence, five then currently unemployed legends of the computer business, Jay Miner, Steve Wozniak, Phill Estridge, Gordon Campbell, and Dado Banato* are at the same bar at the same night on El Camino Real in Santa Clara. They begin talking chips, interfaces, assembly instruction sets, memory, and other such stuff. The more they talk through the evening, the more things each guy makes sense to the others (though that may be the beer, too). They decide to found a new computer company: Pentium Computers. The purpose of this project is to build the best possible computer using each guy's knowledge, experience and industry contacts to design and build the ultimate combination of computational power and price. 1: What would they name their new company? 2: Assuming Relatively minimal changes to computational technology compared to OTL, what sort of CPUs and chipsets would they use? What sort (and how much) of memory would they use? Whould they use hardware sprites, a blitter, or just bit bang the graphics? What sort of sound chip would they use? 3: What sort of OS would it run? What GUI, if any? 4: How much would this baby go for? Inquiring minds want to know! *Jay Miner quit Atari in this timeline after disagreements with my *Mary Sue over the direction to take with 16 and 32 bit computers and consoles. In particular, he thought that while RISC design procesors could get more done per clock cycle and run at higher clock speeds for the transistor count, they also bloated input code (and therefore program size) to an unacceptable degree for regular business and consumer computer use. Steve Wozniak didn't crash his plane in this timelinw, and he quit Apple after becoming disgusted at the degree of Steve Jobs' willingness to cheat Apple rank and file, squeeze the suppliers, and cheat consumers. The last straw was the way he had sabotaged the Apple IV Project by forcing Woz's team to use rediculously decolocked processors and hideously slow RAM, while actually pricing it within $500 of Jobs' team's Apple Nickajack which aside from CPU MIPS, was imferior to the Apple IV in every way. Phill Estridge was fired by IBM in this timeline after speaking out to management about the Surprise Gotcha clause in the IBM-Microsoft deal that IBM Sneaked in at the last second before signing the deal, the one that stated that IBM owned all x86 builds of MSDOS. He then went to Microsoft, explained everything, and together helped create the MSX1 standard, And specifically defining the limits of the memory map of the MSX Pro part of the standard as up to 12 MB system RAM and up to 4MB Video RAM. Gordon Campbell and Dado Benato didn't found Chips and Technologies in this timeline: instead, they helped Harry Fox launch Spectravideo and went on to help engineer the MSX Engine Chipset for Nishii Kazuhiko with Texas Instruments, Zilog, Yamaha, and Hitachi. Quote Link to comment Share on other sites More sharing options...
HiroProX Posted March 4, 2010 Share Posted March 4, 2010 James D. Said that the changes my timeline works will have unpredictable side effects. Here's one of them: It is late a August Friday evening in 1984 in a timeline that had diverged from ours in 1972. By shearest coincidence, five then currently unemployed legends of the computer business, Jay Miner, Steve Wozniak, Phill Estridge, Gordon Campbell, and Dado Banato* are at the same bar at the same night on El Camino Real in Santa Clara. They begin talking chips, interfaces, assembly instruction sets, memory, and other such stuff. The more they talk through the evening, the more things each guy makes sense to the others (though that may be the beer, too). They decide to found a new computer company: Pentium Computers. The purpose of this project is to build the best possible computer using each guy's knowledge, experience and industry contacts to design and build the ultimate combination of computational power and price. 1: What would they name their new company? 2: Assuming Relatively minimal changes to computational technology compared to OTL, what sort of CPUs and chipsets would they use? What sort (and how much) of memory would they use? Whould they use hardware sprites, a blitter, or just bit bang the graphics? What sort of sound chip would they use? 3: What sort of OS would it run? What GUI, if any? 4: How much would this baby go for? Inquiring minds want to know! *Jay Miner quit Atari in this timeline after disagreements with my *Mary Sue over the direction to take with 16 and 32 bit computers and consoles. In particular, he thought that while RISC design procesors could get more done per clock cycle and run at higher clock speeds for the transistor count, they also bloated input code (and therefore program size) to an unacceptable degree for regular business and consumer computer use. Steve Wozniak didn't crash his plane in this timelinw, and he quit Apple after becoming disgusted at the degree of Steve Jobs' willingness to cheat Apple rank and file, squeeze the suppliers, and cheat consumers. The last straw was the way he had sabotaged the Apple IV Project by forcing Woz's team to use rediculously decolocked processors and hideously slow RAM, while actually pricing it within $500 of Jobs' team's Apple Nickajack which aside from CPU MIPS, was imferior to the Apple IV in every way. Phill Estridge was fired by IBM in this timeline after speaking out to management about the Surprise Gotcha clause in the IBM-Microsoft deal that IBM Sneaked in at the last second before signing the deal, the one that stated that IBM owned all x86 builds of MSDOS. He then went to Microsoft, explained everything, and together helped create the MSX1 standard, And specifically defining the limits of the memory map of the MSX Pro part of the standard as up to 12 MB system RAM and up to 4MB Video RAM. Gordon Campbell and Dado Benato didn't found Chips and Technologies in this timeline: instead, they helped Harry Fox launch Spectravideo and went on to help engineer the MSX Engine Chipset for Nishii Kazuhiko with Texas Instruments, Zilog, Yamaha, and Hitachi. Okay, first of all, a RISC computer circa 1984 is going to be a very expensive machine. Even in 1986 OTL the majority of the cost of the Atari 1040ST was its 1Mb of RAM. Also in the 1988-1991 timeframe you have the US protectionist tariffs against Japanese DRAM chips which caused an industrywide DRAM shortage. By 1984, the only options for RISC CPUs would be the Berkley RISC-1 or RISC-2 (which evolved into the Sun SPARC V7 in 1986), the Stanford MIPS design (first commercialized in the MIPS R2000 in 1985, followed by the R3000 in 1988, and R4000 in 1991), The Acorn RISC Machine ARM1 in 1985, followed by ARM2 in 1986, the AMD Am290x0 family in 1988, the Apollo PRISM in 1988 (later folded into PA-RISC when HP aquired Apollo), and the Intel i860 in 1989 followed by the i960 in 1991. So any delivery of CPUs to Atari for a RISC machine would not be at any remotely feasible price until 1987. In any case, this would delay next-gen RISC machines from Atari or Apple by several years and both companies would be making a major leap in system complexity as practically all of the early RISC CPUs were 32-bit. So in the interim, Apple may have had to release the Apple IV as if anything a stopgap until RISC CPUs were available. Likewise, Atari would have had to either go with Jay Miner's idea or keep extending the 8-Bits to try to at least maintain the existing customer base. As for the PC clones, your timeline doesn't even slow them down. Digital Research's continued development of CP/M-86 in OTL led to Gem DOS which then begat DR-DOS, a PC-DOS compatible OS which would have fueled the wave of PC clones in the absence of MS-DOS. Another side-effect would have probably been acceptance of GEM as the de facto standard GUI in PC clones, eventually causing a repeat of how MS-DOS in clones pretty much sabotaged OS/2. In the end, IBM would have PC-DOS, but would have to buy GEM from DR. About the only way to change this outcome may have been if IBM chose the MC68000 for the IBM PC 5150. This would have cut off Intel's PC cash cow, but given the dynamics at play at IBM, there was no compelling reason to select the 68000. Oddly, a possible outcome might be Apple and Atari both opting for WDC85816 CPUs for interim machines, and from the 65xx series, the ARM2 would be the most logical upgrade. So the realities of CPU availability may force Jobs to have to devote needed resources to the Apple IV (Apple IIgs analog) simply because of the limited availability of R2000 CPUs, and instead may cause a revision in the interim to use the R3000 instead. Atari would probably build a 65816 machine using Jay Miner's input incorporating many Amiga-like technologies, but with backwards compatibility to the 8-bit line. The follow-up ARM2 based Atari would probably use an as identical as possible sound and video chipset and have an optional "65816 cart" for backwards compatibility. Odd possibility, Atari licensing MS Windows as the new OS. Windows' driver system would allow Atari more latitude to improve the A/V hardware chipset without breaking software compatibility, except for the usual games written with gross violations of the programmer's guidelines. Another possibility is a way that Jack Tramiel's paranoia about an invasion of Japanese home computers can play out. Faced with a stronger MSX standard making his fear more realistic, the possibility of Commodore and Atari agreeing to a standard configuration for a next generation computer becomes a remote possibility. With MOS producing licensed ARM2 CPUs, and using a hardware independent OS allowing each company to come out with a machine characteristically their own, but software compatible. Eventually with a merging of R&D efforts, they create a standard RISC computer architecture that could be licensed and result in a competing standard against the PC and MSX. The main requirement to make that happen would be an Atari chief executive who was actually aware of the computer market, and was able to speak Jack Tramiel's language. Apple is pretty much going to be stuck in the "Boutique Technology" niche they've always occupied, but with Jobs in the picture that can't be helped. About the only way to impact apple would be a scenario where Jobs leaves, but Woz stays. But that opens up a huge can of worms, as Woz was no businessman. Quote Link to comment Share on other sites More sharing options...
Kalvan Posted March 4, 2010 Author Share Posted March 4, 2010 (edited) Let me better explain myself What I meant to say is that My Atari is going to use the MIPS R2000 in this timeline's ST. However, until 1989, that computer will be mostly marketed as a business computer/bargain basement workstation (albeit one with incredibly lush graphics and sound), and Atari's home focus will be on the 65816 powered 1600, introduced with 128K of RAM. The MIPS mentioned with Apple only referred to Million Instructions Per Second. The Nickerjack is a Motorola 68000 computer platform jointly engineered with Tandy and Sharp (Each company creating its own OS {or in Tandy's case, licensing Microware's OS 10 68K} These OS's are binary incompatible down to the BIOS level, though)). The Apple version features beeper sound (like the Apples I-III) and a Motorola graphics chip roughly comparable to the VIC III in TTLs Commodore 128 (Except for no Hardware Sprites). Aside from having to use bog slow DRAM and a 2.8 Megahertz 65816 (rather than the Hickerjacks faster banks of SRAM and VRAM), the Apple IV (IIgs analogue) was a superior machine in every way. At the moment, Commodore's most potent upcoming computer is TTL's version of the 128 which ended up as something of a cross between a Commodore 65 and an MSX machine (Much the way the OTL Commodore 128 was a cross between a souped up 64 and a CP/M Machine). Basically, Commodore was forced to rewrite the MSX BIOS to fool it into thinking that SID and VIC III were the Texas Instruments/Yamaha MSX Engine chips. The Company is divided into two camps about where thry go from there: One side, lead by Irving Gould wants to embrace the upcoming MSX2 Pro standard with a souped-up reverse-engineered version of a Zilog 800 (Basically an ASCII R800 with all the z80's undocumented features, fully static core, and a 6510 mode) and SID and VIC III functionality built right into commodore's version of MSX sound chips. (Specifically, Irving plans to license the Yamaha YM2141 for the basis of the MSX2 Commodore's sound chip.) The other side, led by the Tramiels, wants to make OTL's 68000 based Atari ST and achieve a clean break from the Vic-20-Max Machine-64-128 generation of 8 bit computers. In adition to DR-DOS, Digital Research would also have to reverse engineer the IBM PS/2 Bus, which was integrated into the system in TTL with the PC 5150. The only PC Clones in TTL so far are the licensed NEC PC-81/83 (Japan and the Far East) and Phillips MK series (PAL Markets), and none of those were for sale in the US. This is not to say that IBM is the only major x86 computer maker in the US: We also have the Fujitsu Micro 16 (Running DRDOS, CP/M {out of the Zilog Z80 Sound Processor}, and OS 9 {Out of the Motorola 6809 Clock Chip/IO Conroller} all at the same time,) and TTL's DECStation, which ran a cut down version of VMS on the Intel 80286. One of the problems with taxing foreign memory is that a good number of MSX machies in TTL are from American companies, including not just Spectravideo, but also Kaypro, Compaq, Tandy (about to be dropped for that 68000 based Tandy 1000 mentioned above to provide an upgrade path for the TSR-90 {TTL's CoCo}), Texas Instruments (They didn't do any MSX Pro machines because it would conflict too much with the TI Professional/Scientific Business/Workstation series), and Hewlett-Packard. In short, only Atari will be running a RISC CPU for sure (With the MSX2 Commodore possibly being another contender), But I would be quite ammenable to any of the ther players doing the same. So, basically, I wanted to create a computer company completely out of left field, founded and run by essentially a team of legends. I know, we are dealing with a few big egos here, but from what I could find from looking through various history books and biographies, Wozniak, Estridge, and Banato seemed personable enough. As I said before, this is a tangent of my Atari Triumphant timeline: Right Here Edited March 4, 2010 by Kalvan Quote Link to comment Share on other sites More sharing options...
Kalvan Posted March 4, 2010 Author Share Posted March 4, 2010 Also, instead of the IBM-Apple-Motorola Power PC architecture, my plans eventually include TTL to eventually feature the SMART (Sun, Motorla, Apple, Remasis, Toshiba) series of SPARC chips. Quote Link to comment Share on other sites More sharing options...
HiroProX Posted March 4, 2010 Share Posted March 4, 2010 I'd still say ARM makes sense for Atari for the same reason it made sense for Acorn, aspects of its design were inspired by the 6502, a CPU Acorn and Atari have experience with. Quote Link to comment Share on other sites More sharing options...
JamesD Posted March 4, 2010 Share Posted March 4, 2010 I can't see the Woz twist. Without the crash and with the flop of the III, which was a Jobs induced mess... I think Woz would have had more pull on future development. I think the IIe would have been more capable from day one, with a high speed mode. I'd guess an intermediate step between the IIe and IIgs would have been introduced as well, possibly with a more 6516 like CPU. Do that and Apple gains even more dominance amongst the 8 bitters. But he's still going to fall in love and want to just go play rather than worry about making Apple dominant like Jobs. Woz wants to play, it's in his nature. Jay already had something in mind when he left Atari and I don't see that changing. RISC just didn't have the status quite yet and an affordable RISC CPU didn't exist. As far as ARM goes... Acorn may have been reluctant to have a competitor when they first released the ARM. They wanted to be the Apple of the UK and that was their advantage. So Amiga is probably still Amiga and ARM is still Acorn's baby. I don't see Phil as a player at all. After all... he died. Unless you can walk on water it's not changing. I think he was happy at IBM so his path was set. Why exactly are we obsessed with the MSX standard so much? It flopped. It might replace CP/M if you introduce it sooner but then CP/M quickly tanked with the intro of the PC and then all the business stuff you want to support starts becoming secondary to the game aspect... which is what MSX had going for it to begin with, and it couldn't compete with the 68K machines and game consoles. You end up back where you started. Spectravideo is based in Hong Kong anyway so I don't see them hiring these two Americans. They had the resources to do what you suggest without them. And didn't you already post this on soc.history.what-if? Quote Link to comment Share on other sites More sharing options...
JamesD Posted March 4, 2010 Share Posted March 4, 2010 BTW, if Phill Estridge gets upset with IBM he sends his resume to DEC, Cray, or aerospace companies... not a software company. Quote Link to comment Share on other sites More sharing options...
kool kitty89 Posted March 5, 2010 Share Posted March 5, 2010 I'd be more interested in a timeline where Ted Dabney and Wozniak managed to avoid joining up with such unscrupulous or problematic buisness partners, either with better partners or on their own. (or maybe even joining up) Quote Link to comment Share on other sites More sharing options...
JamesD Posted March 6, 2010 Share Posted March 6, 2010 (edited) If Woz hooks up with someone less headstrong and driven, how far beyond the kit computer do you think Apple would get? As far as hardware goes, the Apple II had way too many design hacks to compete on speed. If it weren't for the Apple "cult" built by Jobs... how long do you think Apple would have lasted? As far as Dabney goes... if he hooks up with someone that's better with finances then Atari might not need Warner. And as a result, maybe E.T. isn't a total piece of crap as a game. <edit> But if Apple has to compete solely on features then maybe they dump the hack for the display memory between the II and the II+, the machine isn't such a PITA to program games on and they are faster. But I think it was Jobs that pushed them into schools so maybe they sell half as many machines and we don't end up with Macs or the iCrap. Edited March 6, 2010 by JamesD Quote Link to comment Share on other sites More sharing options...
HiroProX Posted March 7, 2010 Share Posted March 7, 2010 Even then, MSX is going to be a primarily japanese-only platform, like in real life. The instant one MSX machine showed up on American shores with serious intent, Jack Tramiel and Commodore would do to them what they did to TI. Remember, Tramiel was absolutely paranoid about a japanese entry into the US computer market. ThE MSx2 and MSX2+ machines might have been contenders, but by the time they were brought out, the US market had already seen the introduction of 16-bit machines by Apple, Atari, and Commodore. MSX1 machines were no competition technically against either the Atari 8-bits or Commodore's C64. Even if they could be brought in against the VIC-20, they'd still be outclassed by the Atari 400 and 800. If they were to manage any degree of success, it'd be at the TI-99/4A's expense, and then they would meet TI's fate at the hands of Commodore. but even then, if a Z-80 machine didn't have an 80-column text display and run CP/M, it was toast. Quote Link to comment Share on other sites More sharing options...
kool kitty89 Posted March 9, 2010 Share Posted March 9, 2010 Even then, MSX is going to be a primarily japanese-only platform, like in real life. The instant one MSX machine showed up on American shores with serious intent, Jack Tramiel and Commodore would do to them what they did to TI. Remember, Tramiel was absolutely paranoid about a japanese entry into the US computer market. ThE MSx2 and MSX2+ machines might have been contenders, but by the time they were brought out, the US market had already seen the introduction of 16-bit machines by Apple, Atari, and Commodore. MSX1 machines were no competition technically against either the Atari 8-bits or Commodore's C64. Even if they could be brought in against the VIC-20, they'd still be outclassed by the Atari 400 and 800. If they were to manage any degree of success, it'd be at the TI-99/4A's expense, and then they would meet TI's fate at the hands of Commodore. Well, a couple things. If the MSX had been collaborated over with one or more American firms, that could have been a factor. Coleco would be the obvious possibility, opting to standardize over that rather than going with the Adam, though it would mean some modifications to the MSX design to match the ColecoVision's sound hardware and memory map (etc), but might have been a possibility. Plus, Jack was gone from Commodore by the beginning of 1984 (and if Atari was in good shape, TTL wouldn't go there either), so only a couple months after MSX was standardized, granted the C64 was already there, saturating the market, but with the CV to build on and a strong JP userbase (possibilities in EU as well), a Coleco-MSX may have had some chance on the market, though I hardly see it being a dominant player. (especially if Atari was handling their computers well) One thing on the MSX in general though; given that it was a Microsoft Initiative, it may have been advantageous to go a different route with that entirely, more towards a (semi)compatible IBM PC clone with reasonably balanced sound and video capabilities and price to fit somewhat into the home computer category, like the PCJr, or especially more successful (and very similar) Tandy-1000. Quote Link to comment Share on other sites More sharing options...
JamesD Posted March 9, 2010 Share Posted March 9, 2010 I'm still trying to figure out what this has to do with Atari being Triumphant. This is what the title implied. Quote Link to comment Share on other sites More sharing options...
HiroProX Posted March 9, 2010 Share Posted March 9, 2010 But the primary problem with the idea of an 'Atari Triumphant" timeline is that everything in the computer industry was headed for standardization. The closest possible thing would be three competing standards, with atari as a manufacturer of machines fitting one of those standards. Then again, I have trouble with the idea of Atari going to MIPS CPUs, as this would delay a replacement for the 8-bits by another year, when the 8-bits were getting hammered in the market by the C64. Likewise, a MIPS-based machine would be VERY expensive, like IBM PC/AT kind of expensive ($4000+). In that case, Apple could release a MC68000 machine for $2000 and be "the low price leader". If Apple had a hard time selling Macs OTL for $2000, Atari trying to sell a $4000+ machine would be hopeless. At least the IBM PC/AT had the PC's software library to use, and being an IBM product it could command a $4000+ price. Keep in mind that the major user of MIPS CPUs OTL was SGI, and a "cheap" SGI machine in the late 80s, the Personal Iris 4D/25 was a $10,000 machine. So I somehow doubt Atari could bring a MIPS based machine to market for under $4000. Quote Link to comment Share on other sites More sharing options...
JamesD Posted March 10, 2010 Share Posted March 10, 2010 Something else I found out that might be a problem. The MSX hardware design inserts additional wait states so the Z80 doesn't run at full speed. It was so they could use slower RAM chips if needed. That doesn't exactly make MSX competitive. Quote Link to comment Share on other sites More sharing options...
Kalvan Posted March 11, 2010 Author Share Posted March 11, 2010 (edited) But the primary problem with the idea of an 'Atari Triumphant" timeline is that everything in the computer industry was headed for standardization. The closest possible thing would be three competing standards, with atari as a manufacturer of machines fitting one of those standards. Then again, I have trouble with the idea of Atari going to MIPS CPUs, as this would delay a replacement for the 8-bits by another year, when the 8-bits were getting hammered in the market by the C64. Likewise, a MIPS-based machine would be VERY expensive, like IBM PC/AT kind of expensive ($4000+). In that case, Apple could release a MC68000 machine for $2000 and be "the low price leader". If Apple had a hard time selling Macs OTL for $2000, Atari trying to sell a $4000+ machine would be hopeless. At least the IBM PC/AT had the PC's software library to use, and being an IBM product it could command a $4000+ price. Keep in mind that the major user of MIPS CPUs OTL was SGI, and a "cheap" SGI machine in the late 80s, the Personal Iris 4D/25 was a $10,000 machine. So I somehow doubt Atari could bring a MIPS based machine to market for under $4000. I had not meant to discuss this on this thread, but... I've been looking up all I can on the MIPS and related architectures. Even for the time, it was a remarkably small chip (less than half the transistor count of the 286, for instance), and from what I could read, it broke no new ground in terms gate, switch, or transistor design (how they were arrangeded was of course a different matter, but that's less of a consequence once it gets to the fabs). It seems like the single biggest expense MIPS had was in the design phase, not in spinning the masks or printing and cutting the chips. I highly suspect that a well placed chunk of extra Level 2 Funding fron Sunnyvale might make MIPS a little responsive to a special bulk discount if offered the production volume Atari would have in mind... Edited March 11, 2010 by Kalvan Quote Link to comment Share on other sites More sharing options...
Kalvan Posted March 11, 2010 Author Share Posted March 11, 2010 Jay already had something in mind when he left Atari and I don't see that changing. RISC just didn't have the status quite yet and an affordable RISC CPU didn't exist. As far as ARM goes... Acorn may have been reluctant to have a competitor when they first released the ARM. They wanted to be the Apple of the UK and that was their advantage. So Amiga is probably still Amiga and ARM is still Acorn's baby. I said that he left Atari over it just like OTL, didn't believe in RISC. I also never mentioned ARM. As for the timeline on soc.history.what-if, that's back to the drawing board until Curt can give more details about Project GAZA (a motherboard layout would be a great starting point). Quote Link to comment Share on other sites More sharing options...
JamesD Posted March 11, 2010 Share Posted March 11, 2010 I had not meant to discuss this on this thread, but... I've been looking up all I can on the MIPS and related architectures. Even for the time, it was a remarkably small chip (less than half the transistor count of the 286, for instance), and from what I could read, it broke no new ground in terms gate, switch, or transistor design (how they were arrangeded was of course a different matter, but that's less of a consequence once it gets to the fabs). It seems like the single biggest expense MIPS had was in the design phase, not in spinning the masks or printing and cutting the chips. I highly suspect that a well placed chunk of extra Level 2 Funding fron Sunnyvale might make MIPS a little responsive to a special bulk discount if offered the production volume Atari would have in mind... Intel owned their own production facilities already. MIPS was probably using someone else to manufacture their chips. Intel also dedicated those facilities just to production of CPUs. If a facility has to change what chip they are fabricating it takes time. Huge difference in production cost. I worked on a contract for a company that produced their own chip... a very specialized chip. They made them by the dozen rather than thousands and it was ridiculously expensive. But they owned the patent and had a corner on the market. Sadly, I never got to see the actual chip production. I'm guessing the equipment was intended for prototypes rather than production. Basically, just because a chip is simple or has fewer transistors doesn't mean they were cheaper to produce. Quote Link to comment Share on other sites More sharing options...
Kalvan Posted March 11, 2010 Author Share Posted March 11, 2010 (edited) Intel owned their own production facilities already. MIPS was probably using someone else to manufacture their chips. Intel also dedicated those facilities just to production of CPUs. If a facility has to change what chip they are fabricating it takes time. Huge difference in production cost. Basically, just because a chip is simple or has fewer transistors doesn't mean they were cheaper to produce. Well, if that's the case (I haven't been able to tell whether MIPS was fabless or not), they can hitch a ride on the Atari Semiconductor Group's vendor chain. Surely the folks making GTIAs, POKEYs, AMYs, JANICEs, and the like for Atari would love to help MIPS turn the R2000 into a workstation chip with a mainstream price. After all, I'm quite certain that Atari, Apple, Nintendo, NEC, and Acorn (Among many, many others) didn't get their 6502s directly from MOS after the Commodore buyout. Edited March 11, 2010 by Kalvan Quote Link to comment Share on other sites More sharing options...
JamesD Posted March 11, 2010 Share Posted March 11, 2010 Well, if that's the case (I haven't been able to tell whether MIPS was fabless or not), they can hitch a ride on the Atari Semiconductor Group's vendor chain. Surely the folks making AMYs, JAIMEs, and the like for Atari would love to help MIPS turn the R2000 into a workstation chip with a mainstream price... But then they have to change over production between chips which adds to cost. It would definitely be cheaper than outsourcing... but Atari didn't own MIPS so that complicates things as well. Compare that to intel, where they have dedicated facilities and they only make changes when they change the CPU. I've participated in this discussion because some of it was interesting, but it's getting old. But what if, but what if, but what if... Atari failed... deal with it!!!!!!!!!! They were a video game company that gained success in the home game market. But when they tried to transition to computers and newer game consoles they just couldn't hit the market. Between that, poor money management and a bad economy, they died. The poor money management more than anything led to their demise and if you don't fix that they will still fail no matter what else you do. You want a what if? What if the boneheads in charge had kept closer track of what was going on at all levels of the company. What if they had kept expenses in line? What if they hadn't blown so much money during the Arcade cash cow years? What if they had actually paid some of the 2600 game programmers better so they wouldn't leave to form other companies? What if they had introduced a version of the computer chipset with 80 column text instead of just more graphics modes? What if they ponied up some money to put business apps on their computers early on instead of just games? What if they market the computers to a broader audience so they weren't seen as just an expensive game system? What if they introduced a version of the 5200 that was memory map compatible with the 800 in place of the 400 and used different controllers? What if they phased out the 2600 early and made a push to migrate all the old games to the 5200/800? Quote Link to comment Share on other sites More sharing options...
kool kitty89 Posted March 12, 2010 Share Posted March 12, 2010 But the primary problem with the idea of an 'Atari Triumphant" timeline is that everything in the computer industry was headed for standardization. The closest possible thing would be three competing standards, with atari as a manufacturer of machines fitting one of those standards. Then again, I have trouble with the idea of Atari going to MIPS CPUs, as this would delay a replacement for the 8-bits by another year, when the 8-bits were getting hammered in the market by the C64. Likewise, a MIPS-based machine would be VERY expensive, like IBM PC/AT kind of expensive ($4000+). In that case, Apple could release a MC68000 machine for $2000 and be "the low price leader". If Apple had a hard time selling Macs OTL for $2000, Atari trying to sell a $4000+ machine would be hopeless. At least the IBM PC/AT had the PC's software library to use, and being an IBM product it could command a $4000+ price. Keep in mind that the major user of MIPS CPUs OTL was SGI, and a "cheap" SGI machine in the late 80s, the Personal Iris 4D/25 was a $10,000 machine. So I somehow doubt Atari could bring a MIPS based machine to market for under $4000. From the original discussion:http://www.atariage.com/forums/topic/154910-reposed-from-general-chat/page__p__1897625__fromsearch__1#entry1897625 (also this related discussion) http://www.atariage.com/forums/topic/155306-one-tangent-of-my-atari-what-if/page__p__1903226__fromsearch__1#entry1903226 At least the way I was trying to push the discussion, I suggested MIPS, or ARM rather than Alpha which is what Kalvan had been sticking on... However, in the context I was referring to (favoring ARM personally), RISC would not come in until the tail end of the 1980s at earliest, rather, they'd stay with the 650x architecture prior to that. (starting with faster 6502s, 65C02s, then 65816s prior to going to RISC) The Idea was that at soem point, atari could get a licence for the 65816 core (relatively easy for that chip) and implement it in their own cusom, integrated chip to provide backwards compatibility on later models. I preferred ARM as the lower cost and more reasonable option compared to MIPS. (though that became much more reasonable with the expansion to the embedded market int he early 90s) ARM would probably be easier to license or obtain from alternate vendors than MIPS would have been, at least prior to the mid 90s. Something else I found out that might be a problem. The MSX hardware design inserts additional wait states so the Z80 doesn't run at full speed. It was so they could use slower RAM chips if needed. That doesn't exactly make MSX competitive. That's kind of possible an argument in favor of the Coleco/MSX partnership thing then, isn't it? Intel owned their own production facilities already. MIPS was probably using someone else to manufacture their chips. Intel also dedicated those facilities just to production of CPUs. If a facility has to change what chip they are fabricating it takes time. Huge difference in production cost. I worked on a contract for a company that produced their own chip... a very specialized chip. They made them by the dozen rather than thousands and it was ridiculously expensive. But they owned the patent and had a corner on the market. Sadly, I never got to see the actual chip production. I'm guessing the equipment was intended for prototypes rather than production. Basically, just because a chip is simple or has fewer transistors doesn't mean they were cheaper to produce. Yes, but if the fabless company uses a very large chip vendor, and produces a relatively high volume, that wouldn't be the case so much. Or, in particular, if the processor could have been licensed, and the licencee selected the vendor of choice (for a relatively high volume product), though this mightn't have been the case with MIPS, it might have with ARM (which I suggested, Kalvan seems to dislike ARM for soem reason), and WDC's cores should have been readily licencable as well. (hence that note in the context of the 650x and '816 machines) You want a what if? What if the boneheads in charge had kept closer track of what was going on at all levels of the company. What if they had kept expenses in line? What if they hadn't blown so much money during the Arcade cash cow years? What if they had actually paid some of the 2600 game programmers better so they wouldn't leave to form other companies? What if they had introduced a version of the computer chipset with 80 column text instead of just more graphics modes? What if they ponied up some money to put business apps on their computers early on instead of just games? What if they market the computers to a broader audience so they weren't seen as just an expensive game system? What if they introduced a version of the 5200 that was memory map compatible with the 800 in place of the 400 and used different controllers? What if they phased out the 2600 early and made a push to migrate all the old games to the 5200/800? Yeah, heh. Some of those would be pretty interesting though, but yeah. The problem I had from the start of Kalvan's alternate timelyine thread was that it was too broad and started jumping back and forth. With such a hypothetical situation starting that early and that broad, it really doesn't become a discussion so much as a story. Inevitably it got more segmented into individual propositions, but soem of those were dependent on other historical changes enabling them. It's not a simpler "what if" they'd don't better with one thing, or more specifically how could they have doen better with one area. (say the 5200, management issues, 8-bit computers, ST, Jaguar, etc) THat whole thing with the MSX really should have been a separate thread altogether. (that one didn't even need the original premise necessarily) In that line though, I still find the Coleco partnership idea a bit intriguing as well as the thing about a more PC-clone oriented machine instead of the MSX at all. (again kind of like a low-end Tandy 1000) Quote Link to comment Share on other sites More sharing options...
HiroProX Posted March 13, 2010 Share Posted March 13, 2010 But the primary problem with the idea of an 'Atari Triumphant" timeline is that everything in the computer industry was headed for standardization. The closest possible thing would be three competing standards, with atari as a manufacturer of machines fitting one of those standards. Then again, I have trouble with the idea of Atari going to MIPS CPUs, as this would delay a replacement for the 8-bits by another year, when the 8-bits were getting hammered in the market by the C64. Likewise, a MIPS-based machine would be VERY expensive, like IBM PC/AT kind of expensive ($4000+). In that case, Apple could release a MC68000 machine for $2000 and be "the low price leader". If Apple had a hard time selling Macs OTL for $2000, Atari trying to sell a $4000+ machine would be hopeless. At least the IBM PC/AT had the PC's software library to use, and being an IBM product it could command a $4000+ price. Keep in mind that the major user of MIPS CPUs OTL was SGI, and a "cheap" SGI machine in the late 80s, the Personal Iris 4D/25 was a $10,000 machine. So I somehow doubt Atari could bring a MIPS based machine to market for under $4000. From the original discussion:http://www.atariage.com/forums/topic/154910-reposed-from-general-chat/page__p__1897625__fromsearch__1#entry1897625 (also this related discussion) http://www.atariage.com/forums/topic/155306-one-tangent-of-my-atari-what-if/page__p__1903226__fromsearch__1#entry1903226 At least the way I was trying to push the discussion, I suggested MIPS, or ARM rather than Alpha which is what Kalvan had been sticking on... However, in the context I was referring to (favoring ARM personally), RISC would not come in until the tail end of the 1980s at earliest, rather, they'd stay with the 650x architecture prior to that. (starting with faster 6502s, 65C02s, then 65816s prior to going to RISC) The Idea was that at soem point, atari could get a licence for the 65816 core (relatively easy for that chip) and implement it in their own cusom, integrated chip to provide backwards compatibility on later models. I preferred ARM as the lower cost and more reasonable option compared to MIPS. (though that became much more reasonable with the expansion to the embedded market int he early 90s) ARM would probably be easier to license or obtain from alternate vendors than MIPS would have been, at least prior to the mid 90s. This pretty much echos my thoughts. A movement to a fully incompatible RISC architecture would require more capable interim machines that would likely need to be backwards compatible with the previous 8-bit computers. And the insistence on the Alpha was kind of WTF? to me. I like Alphas, I owned some back then and still do, but they were expensive machines, and the architecture pretty much ran out of gas with the axp21264, as the axp21364 is a minimalist design meant to be used in massively parallel systems, not exactly a desktop box. Moving from more advanced 65xx designs tends to favor ARM more than MIPS. SGI went to MIPS CPUs from MC680x0 CPUs. Though it does become a possibility that if WDC can sell many more 65816s, that a "65832" might have happened, which would make a move to RISC more questionable, especially if the "WDC65832" offered performance comparable to the i386 or MC68020. Though I think that would result in a delayed move to RISC, probably when 64-bit RISC CPUs such as StrongARM, MIPS R4000, UltraSPARC, and DEC axp21064 became more available. But all of this would only be truely usable if Atari could move the 8-bits to an OS with an API so that major changes in the underlying hardware could be masked from programs. Otherwise, there would be massive compatibility problems with software that (as most A8 software did) directly access the hardware. In many ways, maintaining compatibility with the original A8s for Atari, or the C64 in the case of Commodore would have been next to impossible past 16-bit machines. PCs only managed this compatibility back to the 8088 by the use of several modes, Real Mode, 286 Protected Mode, 386 Protected Mode, and Virtual x86 Mode. Quote Link to comment Share on other sites More sharing options...
Kalvan Posted March 13, 2010 Author Share Posted March 13, 2010 It's not that I dislike ARM, it's just that Acorn won't license it for love or money beofre 1990, and for much the same reason Apple under Jobs will never liscense the Macintosh OS. Quote Link to comment Share on other sites More sharing options...
JamesD Posted March 13, 2010 Share Posted March 13, 2010 Something else I found out that might be a problem. The MSX hardware design inserts additional wait states so the Z80 doesn't run at full speed. It was so they could use slower RAM chips if needed. That doesn't exactly make MSX competitive. That's kind of possible an argument in favor of the Coleco/MSX partnership thing then, isn't it? Lets see... the MSX group intentionally crippled the MSX standard for the sake of cheaper machines. Coleco's design didn't. Does Coleco cripple their design to be SLOWER? I'm thinking no. Then they loose compatibility with their game system. Does the MSX group improve their standard? It was intentional for cost reasons... how would Coleco's involvement change that? The MSX group could have easily dropped the wait states from their design... but they had their eye on eventually taking on far cheaper machines and RAM was a significant cost at the time. So... no I don't think it's in favor of such a partnership. Quote Link to comment Share on other sites More sharing options...
kool kitty89 Posted March 14, 2010 Share Posted March 14, 2010 This pretty much echos my thoughts. A movement to a fully incompatible RISC architecture would require more capable interim machines that would likely need to be backwards compatible with the previous 8-bit computers. And the insistence on the Alpha was kind of WTF? to me. I like Alphas, I owned some back then and still do, but they were expensive machines, and the architecture pretty much ran out of gas with the axp21264, as the axp21364 is a minimalist design meant to be used in massively parallel systems, not exactly a desktop box. The weirder thing was suggesting it be used as the Jaguar's CPU later on... (when 68020s and MIPS R3000s were already quite compatible with the historical design and used int he arcade incarnations) Moving from more advanced 65xx designs tends to favor ARM more than MIPS. SGI went to MIPS CPUs from MC680x0 CPUs. Though it does become a possibility that if WDC can sell many more 65816s, that a "65832" might have happened, which would make a move to RISC more questionable, especially if the "WDC65832" offered performance comparable to the i386 or MC68020. The rough descriptions of any 65832 (or 65T32) seem to be using 16-bit data buses as well, though there doesn't seem to be much in the way of actual documentation of any developments. One thing I do wonder about, is what MOS might have done if they'd continued to develop the architecture rather than basically stopping after CBM buying them. But all of this would only be truely usable if Atari could move the 8-bits to an OS with an API so that major changes in the underlying hardware could be masked from programs. Otherwise, there would be massive compatibility problems with software that (as most A8 software did) directly access the hardware. In many ways, maintaining compatibility with the original A8s for Atari, or the C64 in the case of Commodore would have been next to impossible past 16-bit machines. PCs only managed this compatibility back to the 8088 by the use of several modes, Real Mode, 286 Protected Mode, 386 Protected Mode, and Virtual x86 Mode. Or the Apple IIgs route, with integrated hardware compatibility. Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.