keithd@cadovax.UUCP (Keith Doyle) (03/17/87)
References: >>An Amiga-as-a-UNIX-box will never sell. SUN already does that. Apple will >>be there soon. >I disagree. A UNIX (tm) system at Amiga 1000 prices would sell like hotcakes >to universities, a huge market, and to all outside the university systems who >have had the chance to fall in love with the power and flexibiity of UNIX. You first have to produce a version of UNIX that would run in a 512k system with two floppy drives that would not be a total DOG. In my experience with Xenix on an AT, I found that with a 20mb hard disk and 640k, I was left with less than 10MB on my hard disk, and performance in 640k was intolerably slow. Only when I added an above-board RAM card could I even stand to use it. In 640k, just the loading of a medium size application took so long, the system probably swapping and re-swapping pages out of memory like crazy as it read the application into memory. I would think we can make a few statments about UNIX: 1) Requires HARD DISK of at least 15Mb. 2) Can run, but you won't want to use it with less than 1Mb RAM. In addition, you have to add to that AT&T license fees, though I suppose one might predict they will go down in time. So I would think, with this logic it wouldn't be too hard to make a statement something like: "For the next year or so, the CHEAPEST one could possibly sell a UNIX box that anyone would want to use would be somewhere in the range $xxxx-$yyyy" My guess is we're still talking $3000-$4000. I could be proven wrong, I'm sure, and I suppose one day this will change as everything gets cheaper. HOWEVER, it is certainly POSSIBLE to configure an Amiga with enough hardware to support UNIX. All you have to do is convince someone that it's worth spending money to do the port. Then you get what you want without penalizing the rest of us. I just can't see how it makes sense to provide UNIX on an Amiga as the plain vanilla OS that you get with every system. For one thing, not EVERYBODY wants or likes UNIX. I'd hate to pay the performance penalty, or the license fees myself. It works fine on the company VAX, and I do like the interface and features, but with Matt's shell, I can't tell the difference anyway. >Commodore had no choice but to do a quick and dirty OS release at the time, >they came very near bankruptcy, and really needed the cash flow the Amiga 1000 >has provided. The attempt to make a UNIX-like OS was 100% behind schedule, >and AmigaDOS got the product out the door and gave Commodore some breathing >room. Granted. Life in the big city. >However, AmigaDOS is the biggest sales weakness of an otherwise excellent >product, and now is the time for it to be either drastically upgraded, or >preferably replaced with an independently engineered UNIX clone. Not so much >for the end user, but simply because 3rd party software development in a >mixed BCPL and C/UNIX-clone environment is so much harder than a pure C/UNIX >environment that product releases have been slow. As Apple proved long ago >with the contrast in initial acceptance of the Apple II and the Macintosh, >it is third party hardware and software that makes or breaks a home computer >product, so anything Commodore can do to ease third party development is just >money in the bank for Commodore. Well, I pretty much agree with you here. Some people have been complaining about Pournelle's comments in the letters section of the latest BYTE about how "the Amiga is harder to port to than the Atari". Though I'm not to crazy about Pournelle's 'diplomacy' shall we say in the way he makes such a statement, he is basically right. And it's not because the Amiga is a multi-tasking machine, as UNIX and other systems are not as 'hard' as the Amiga to port to. It's the lack of resource managment/tracking, and memory management type features (whether provided in hardware or software). EVERY program has to be careful to clean up after itself. A UNIX program, while multi-tasking like the Amiga, dosen't have to do anything except exit(0); without fear his next task will be compromised because something was left un-done. A program on a single tasking machine such as the Atari just has to do a warm boot. But I don't know what the best path from here to success is, obviously either the addition of a MMU or perhaps a switch to 68010 and lotsa new software are among the choices. I suppose DO NOTHING is a choice too, but here's hoping they don't pick that one. Rumor has it that the original Amiga-destined OS is still out there and may get a shot at it. >Amen. As a Commodore stockholder, I weakly applaud the Amiga 2000's IBM >compatibility features (just because there are a lot of people out there in >business land unsophisticated or conservative enough to think a computer has to >be an IBM clone to work at all, and they make buying decisions), although it >was silly to aim at the PC level performance, which is pretty out of date in >today's business community, where the XT is considered a minimum, and the AT >a preferable level of performance, for today's office. Well, I'm still kind of wondering: "what's the point". I mean, you can buy both an Amiga 1000 AND a PC clone for less than the cost of an A2000 with associated PC expansion hardware. And for now, you only get simple file transfer, which you can get with a cable and couple of public domain terminal programs. Commodore has made some hint of some features that will closer tie the two machines together, but I for one 1) am not going to hold my breath and 2) can't really imagine much they could do together that would make it a better option over any other co-processor which I kinda see as a bit of a frivolity anyway. The A2000 LOOKS more like a business machine, with it's slots, and I certainly think using the AT bus is a good idea, despite the fact that Commodore has effectively told all of the garage hardware shops "hey guys, you now have the entire PC marketplace as competition, hope you didn't decide on the Amiga cause it would give you a chance to be the only game in town for a while". But as far as I'm concerned, if you want a PC, BUY a PC. If you want an Amiga, BUY an Amiga. If you want BOTH, BUY BOTH. The only point I can see, is that it MAY give an excuse for certain ComputerLands, Nynex, and other such stores to have an Amiga there in the store next to the PCs and Macs. Then again it may not. Everybody knows the PC marketplace is flooded. It's tougher to compete there than just about anywhere else. What apparently keeps companies trying to edge in there, is the fact that it is the single largest micro marketplace (which spells G-R-E-E-D). The ONLY things that can save a company still trying to make it there are 1) totally cuthroat pricing, or 2) product differentiation. Certainly Amiga has #2, but the step of coming out with the A2000 does nothing new really to ADVANCE the Amiga along those lines. I guess we'll just have to wait for the A3000 and see. Keith Doyle # {ucbvax,ihnp4,decvax}!trwrb!cadovax!keithd # cadovax!keithd@ucla-locus.arpa
sean@ukma.ms.uky.csnet (Sean Casey) (03/18/87)
In article <1439@cadovax.UUCP> keithd@cadovax.UUCP (Keith Doyle) writes: >we can make a few statments about UNIX: > > 1) Requires HARD DISK of at least 15Mb. You couldn't GIVE me Unix unless I had a hard disk of at least 40 Megs. I'd rather have 80-120. Unix systems fill up their disks FAST, especially when they have access to Usenet. Unfortunately, I can buy a new Amiga for the price of a 40 meg hard disk, and a pretty well decked out Amiga to boot. This cost must be factored into the cost of having a Unix Amiga. If hard drives ever approach PC hard drive prices, having a Unix Amiga might be an affordable reality. > 2) Can run, but you won't want to use it with less than 1Mb RAM. I second this. Myself, I'd want about 4 megabytes. When you add up the kernal size, disk buffering, numerous background system processes, and stuff you occasionally run from cron, you run out of 1 meg pretty fast. Of course, if you have a 68020 and mmu, you can runs things virtual demand paged if you like. This brings me to another point. It might be possible to run Unix on a flat 68000. They've done it on IBM PC-XTs, so a 68000 is a step up. But without memory protection, every time the guru visits (assuming you can run amigados stuff too), you crash and have to fsck the disks. There's NO WAY IN HELL I am going to sit and wait twenty minutes for my file system to be repaired every single time a program crashes. This of course means that either you learn to like waiting or you pay for an '020, mmu, and memory fast enough to keep up (if you buy an 8 MHz 68020, I'll personally shoot you (do they make them that slow?)). The bottom line is: You have to pay Pay PAY for a feasable Unix system. Sean -- =========================================================================== Sean Casey UUCP: cbosgd!ukma!sean CSNET: sean@ms.uky.csnet ARPA: ukma!sean@anl-mcs.arpa BITNET: sean@UKMA.BITNET
mwm@eris.UUCP (03/20/87)
In article <6102@ukma.ms.uky.csnet> sean@ukma.ms.uky.csnet (Sean Casey) writes: >This brings me to another point. It might be possible to run Unix on a >flat 68000. They've done it on IBM PC-XTs, so a 68000 is a step up. It's not only possible, it's been done. The Sun-1's had 68000s in them. Several other workstations from that era also had 68000s. Most of them even ran 4BSD based Unix. >But without memory protection, every time the guru visits (assuming you can >run amigados stuff too), you crash and have to fsck the disks. Ah, but those systems had mmu's. The Stanford University MMU (usually called the SUN MMU) was used a lot. But what happend during a page fault is/was FM. The original point is still valid. Those systems didn't provide much more oomph than an Amiga, but cost as much as a Sun (natch). <mike -- But I'll survive, no you won't catch me, Mike Meyer I'll resist the urge that is tempting me, ucbvax!mwm I'll avert my eyes, keep you off my knee, mwm@berkeley.edu But it feels so good when you talk to me. mwm@ucbjade.BITNET
morenz@utai.UUCP (03/21/87)
Expires: Followup-To: [burp] Hmm, so you need 1Meg ram and >15M hard disk to run UNIX ... Quoting from my trusty "the UNIX prog. env." by K+Pike I see that "The UNIX operating system started on a cast-off DEC PDP-7 ..." with, I believe, a memory that an most computers today could misplace and never notice the loss of. Granted UNIX has grown slightly [:-)] since but, as Andrew Tannenbaum has recently shown us (MINIX), the essentials can still fit into a desk-sized box at studently attainable prices. I hope that my Amiga is an improvement over a PDP-7; so it seems to follow that some kind of UNIX could be implemented for it. I personally don't need it all but (and this is NOT a complaint) Matt's shell does leave a little to be desired if you've gotten really used to the environment. ----------------------------------------------------------------------------- OOPS - Daylight, time to go to bed... Gideon Sheps : occasionally found at gbs@gpu.utcs.UUCP unless they change the machine's name on me again. My opinions are, of course, mine... Your interpretation is, of course, yours. -----------------------------------------------------------------------------
mwm@eris.UUCP (03/31/87)
In article <3786@utai.UUCP> gbs@gpu.utcs.UUCP (Gideon B. Sheps) writes: >Hmm, so you need 1Meg ram and >15M hard disk to run UNIX ... >Quoting from my trusty "the UNIX prog. env." by K+Pike I see that Yeah, and it's gonna run real lousy, and you won't have much space for your own files. > "The UNIX operating system started on a cast-off DEC PDP-7 ..." Hand-written in assembler, without multitasking, and missing lots of other features. >with, I believe, a memory that an most computers today could misplace and >never notice the loss of. Granted UNIX has grown slightly [:-)] since >but, as Andrew Tannenbaum has recently shown us (MINIX), the essentials >can still fit into a desk-sized box at studently attainable prices. All true. But MINIX is a subset of V7, not 4BSD or SysV. And anyone who's seen OS/9 running on a 64K CoCo knows that you can put the essentials of Unix in a desktop box. >I hope that my Amiga is an improvement over a PDP-7; An Amiga is an improvement over a good-sized chunk of the PDP-11 line, much less a PDP-7. >so it seems to follow that some kind of UNIX could be implemented for it. Yeah, "some kind." Let me introduce you to Keith Bostic, who's spent the last few months going slowly crazy trying to squeeze 4.3BSD functionality into a micro PDP-11. This box is faster than an amiga, has about the same amount of memory, but suffers from the 11 16-bit address registers. V7 unix fits quite nicely on an 11. If you build a large system, you need to do things like overlaid kernels and shoving buffers into supervisor space to make it work. But a small system will run on an 11/23 without much trouble. But a SysV or 4BSD system? With 300K kernels on small systems? You wanna run _that_ on your Amiga? You wanna do fork() without an MMU? You crazy, mon. Maybe OS/9, which has fork/exec. Mabye a Minix port, if you can figure out how to do fork/exec. This has all been gone over in info-68k. Dig out the archives and see for yourself (actually, this is still going on there). <mike N.B. - in 4.3BSD, int is assumed to be a long. In 2.10BSD, int is assumed to be a short. You MANX people should _love_ the 2.10 utilities! -- Here's a song about absolutely nothing. Mike Meyer It's not about me, not about anyone else, ucbvax!mwm Not about love, not about being young. mwm@berkeley.edu Not about anything else, either. mwm@ucbjade.BITNET
keithd@cadovax.UUCP (04/03/87)
In article <2981@jade.BERKELEY.EDU> mwm@eris.BERKELEY.EDU (Mike Meyer) writes: >N.B. - in 4.3BSD, int is assumed to be a long. In 2.10BSD, int is >assumed to be a short. You MANX people should _love_ the 2.10 >utilities! I vote we move to ban 'int' altogether. Keith Doyle # {ucbvax,ihnp4,decvax}!trwrb!cadovax!keithd # cadovax!keithd@ucla-locus.arpa