Date: Thu, 9 Aug 2001 12:45:59 +0930 From: Greg Lehey <grog@FreeBSD.org> To: Terry Lambert <tlambert2@mindspring.com> Cc: Mike Meyer <mwm@mired.org>, j mckitrick <jcm@FreeBSD-uk.eu.org>, freebsd-chat@FreeBSD.org Subject: Re: How did the MSFT monopoly start? Message-ID: <20010809124559.G73579@wantadilla.lemis.com> In-Reply-To: <3B7103A4.558B9B3B@mindspring.com>; from tlambert2@mindspring.com on Wed, Aug 08, 2001 at 02:17:24AM -0700 References: <20010806142544.A64348@dogma.freebsd-uk.eu.org> <15214.52633.581653.632317@guru.mired.org> <3B6F98D0.A3C22CC9@mindspring.com> <20010808160551.Q78395@wantadilla.lemis.com> <3B7103A4.558B9B3B@mindspring.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wednesday, 8 August 2001 at 2:17:24 -0700, Terry Lambert wrote: > Greg Lehey wrote: > > Obviously, like my off the list references to the livelock papers > I tried to send you, the direct email to you will bounce from your > overambitious "spam" bouncer, which insists I'm a spammer because > Earthlink bought my ISP and assignned me a mindspring address... > Oh well... Hmm. I've checked my bounce log. Is this you? Out: 220 wantadilla.lemis.com ESMTP Postfix In: EHLO smtp.netcabo.pt In: MAIL FROM:<nastyteen69@earthlink.net> SIZE=2352 Out: 450 Client host rejected: cannot find your hostname, [212.113.174.249] That's the only reference I can find to Earthlink. I reject this message, like many others, because it shows every sign of being spam. >>> The 86 was later. >> >> The 86 was earlier. 1976. The 8088 was just a low-cost 8086, with an >> 8 bit bus, enabling machines to be made with a lower chip count. The >> processor core was almost identical; I think the only difference was >> the pipeline length. I suspect that the part count was what really >> caused IBM to go with the 8088 and not the 68000; the former needed >> only 8 memory chips (1 bit wide), the latter would have needed 32. > > Lowe is quoted as wanting volume off the shelf parts; in fact, > Gates tried to steer him toward a 16 bit processor, but the IBM > confidentiality agreement urged him not to reveal confidential > information, so he didn't reveal what he knew about CP/M-86, so > the OS was not an issue: I don't know what makes you think the OS would have been an issue if he had known about CP/M 86. As I said, the processors were instruction set compatible, and the PC shipped with CP/M 86. > (Also from Cringely's book): > > Choosing a 16-bit processor was easy. Intel, Motorola, > and National Semiconductor were all shipping 16-bit > processors at the time. Intel had the 8086 and 8088 > processors, Motorola had the 68000, and National had its > 16032. The National processor was elegant and powerful; > the Motorola was powerful and easy to write software for; > the Intel 8086 was fairly powerful but had an awkward > memory architecture; the Intel 8088 was an 8086 without > the power. The National and Motorola processors were 32 bit. > Of course, IBM chose the 8088--the least attractive of > all the processors from a technical standpoint. In this > case, technical considerations took a back seat to IBM's > manufacturing and marketing concerns. The plan was to > build a computer without any custom components--just > off-the-shelf parts from major semiconductor makers. The > 8088 was the only 16-bit processor for which there was > available a full complement ot the support chips required > to build a computer. This is wrong. The 8086 and 8088 had the same support chips, specifically the 8284 clock generator, the 8228 system controller, the 8288 bus controller, the 8282 bus latch and the 8286 bus transceiver. The PC didn't use them; instead, it used perfectly acceptable and probably considerably cheaper TTL logic which would have worked just as well on the 8086 (and indeed did on my SCP board). The Intel chips on the original PC were the 8088 (processor), 8087 (optional FPU), 8259 (PIC), 8255 (parallel port), 8237 (DMA controller) and 8253 (CTC). They would have worked just as well with the 8086. We still have their descendents on modern 32 bit machines today. > Motorola and National were still working on their 16-bit > support chips, I think this is wrong. I saw complete computers running 68000s in 1980. I'm pretty sure they were using off-the-shelf components. > as was Intel for the 8086. > But the 8088 was a 16-bit processor in an > 8-bit body, since it used an 8-bit data bus--sending and > receiving data 8 bits at a time and then processing them > in 16-bit mode. This 8-bit bus is what made the 8088 > less powerful than the other contenders, Correct. > but it also > made it possible for the 8088 to use support chips > intended for the earlier 8080 family of Intel 8-bit > processors. Since the 8088 was the only processor that > could be used without developing custom support chips, > it was the only processor that fit IBM's needs. This is all wrong. See above. > [ ... DEC and Tandy can't make a serial port work to save their life ... ] > >> I'm pretty sure that it wasn't the UART which killed these machines. >> Was that the Z-80 SIO? > > It was a Zilog UART. But I think you are maybe thinking of > the Z80 based serial processor in the Tandy-16 and retrofit > Tandy 6000, which had 8 inch floppies and could have 14 inch > 5M hard drives added, for ungodly cost... No, I was thinking of the Zilog UART. It was called the Z-80 SIO, and it was terrible. IIRC it didn't have a status register. You had to take an interrupt when it came, or lose it. >>>> FWIW, Gates sold IBM a product he didn't have. He then went out and >>>> bought QDOS - the Quick and Dirty OS - from SCC, which had written >>>> it for their 8086 S-100 boxes because Digital Research kept >>>> delaying CP/M-86. >> >> Ah, I missed this before. Yes, this is almost exactly correct. The >> company was Seattle Computer Products, SCP. The rest is exactly >> correct. > > The price tag for the rights was $50,000. Gates also knew about > CP/M-86, but didn't disclose it, He didn't need to. You've said yourself that they had been in touch with Digital Research. > even though QDOS had code taken line-for-line from CP/M. That's new to me. How could they do that? And how would he have known? 86/DOS did come with some source, but CP/M never did. >>> IBM attempted several times to contact Digital Research about >>> licensing CP/M, but they never returned IBM's calls, >> >> So how come the PC was released with optional CP/M? > > That was optional later. Originally, it wasn't. Yes, it was. Read my quote in another message. That's from a contemporary document. > Mostly, it was because there were tools that would run under it with > a simple cross-assembler. Huh? Greg -- See complete headers for address and phone numbers To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-chat" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20010809124559.G73579>