silos@bench.sublink.ORG (Paolo Pennisi) (06/29/90)
In article <31093@cup.portal.com>, mmm@cup.portal.com (Mark Robert Thorson) writes: > A while back, I referred to pending announcements of prices cuts on > Motorola 68K-family chips. The info has now been published, so I feel > free to post it here: > > 68020 -- $47 > 68030 -- $95 > 68881 -- $38 > 68882 -- $44 > > All prices are in 1000-unit quantity at 16 MHz. These prices only apply > to the new packaging options, ceramic quad flat-packs for the CPU's and > PLCC's for the FPU's. [...] > I had earlier said it looked like the 68020 would be the world's cheapest > Unix engine. I was forgetting that it was the 68030 which has the on-chip > MMU. At $95, it's more expensive than the 386SX ($64 in 1000's at 16 MHz). > But when you factor in the price of the FPU, Motorola probably beats Intel. > Motorola should also be faster, because the 386SX has a 16-bit external bus. > BTW, there's a 12 MHz version of the 386SX for $55.50 in thousands. [...] Are we talking Unix engine or DOS ? How can you compare an handicapped version of a chip whose father and grandfather are crippled 16 bit segmented processor, a chip which spends a lot of its silicon to implement compatibility with those dark ages cpu to a real 32bit, semi Harvard, data and instruction cache processor, with a real 32 bit data bus a a real 32bit address bus, support for multipro cessing and more. Sure, if we want to build the smaller and cheapest Unix engine we could also use a Z8001 or a 80286 or whatever... maybe some crippled version III could run also on a Z80. The 68030 is definetly superior, more bangs per buck, even not including the FPU. The 68882 is surely better than a 387DX, not to mention the 387SX. And it costs definetly less! Paolo. PS I don't want to start another religion war... but the comparison with the 386SX is too much. -- (ARPA) silos@bench.sublink.ORG Paolo Pennisi (BANG) ...!deejay!bench!silos via Solari 19 (MISC) ppennisi on BIX & PTPOSTEL 20144 Milano ITALIA ----< S U B L I N K N E T W O R K : a new way to *NIX communications >-----
davidsen@crdos1.crd.ge.COM (Wm E Davidsen Jr) (07/09/90)
In article <457@bench.sublink.ORG> silos@bench.sublink.ORG (Paolo Pennisi) writes: | Are we talking Unix engine or DOS ? | How can you compare an handicapped version of a chip whose father and | grandfather are crippled 16 bit segmented processor, a chip which spends | a lot of its silicon to implement compatibility with those dark ages | cpu to a real 32bit, semi Harvard, data and instruction cache processor, | with a real 32 bit data bus a a real 32bit address bus, support for multipro | cessing and more. You're stuck in an infinite loop here. You can't compare the 68k with a Cray, either. The objective is to build a UNIX engine capable of doing useful work and running a 32 bit version of a modern O/S. The price point is that the color system with X will be competitive in price with a color X terminal. Since the only low prices UNIX around is V.3.2 (and maybe V.4 soon) for the 386, that is the choice. Since the cost of the DX chip is artificially high (I would bet the manufacturing and development cost is within 10% of the SX) the SX is the only practical choice. | Sure, if we want to build the smaller and cheapest Unix engine we could also | use a Z8001 or a 80286 or whatever... maybe some crippled version III could | run also on a Z80. Read the original article and the above. You're changing the objective. Those processors would cost a lot more because there is no mass production of systems and vendor competition to keep the price down. | The 68030 is definetly superior, more bangs per buck, even not including | the FPU. The 68882 is surely better than a 387DX, not to mention the 387SX. | And it costs definetly less! If you can deliver a system complete with color display and software for $3000 I will agree. Until then I think you're dreaming. This is not a perfect workstation, it's a color X terminal which is also a free standing UNIX system. Wish I could include the development set for $3k, but it doesn't look good. If anyone wants to help When I find vendors for all this stuff I intend to sell for cost (counting $100 as cost to assemble, burnin, and install UNIX). I will make the list of vendors available to anyone, in hopes that "competition" will drive the price down and put me out of business. Since I'm doing this to make a point, rather than make money, I'd be glad to have someone else make the pittance. If you or your club or organization want to build some of these drop me a line and I'll send the vendor list when I get the final configuration. If you know if a good system which will run UNIX, or a good motherboard, or VGA display, hard disk, whatever, let me know! I want current production rather than remnants, so other people and clubs can build these systems. Please tell me about stuff which you have actually used, rather than ads. I have a pile of once used motherboards, disk controllers, and VGA cards, all of which are great for DOS and don't sun UNIX. If you can get a UNIX with NFS and X for less than Open DeskTop ($995) let me know. ESIX + NFS is slightly higher than that by the prices I got from Everex and ISC. -- bill davidsen (davidsen@crdos1.crd.GE.COM -or- uunet!crdgw1!crdos1!davidsen) "Stupidity, like virtue, is its own reward" -me
chip@tct.uucp (Chip Salzenberg) (07/10/90)
According to silos@bench.sublink.ORG (Paolo Pennisi): >How can you compare an handicapped version of a chip whose father and >grandfather are crippled 16 bit segmented processor, [...] to a real >32bit, semi Harvard, data and instruction cache processor, with a real >32 bit data bus a a real 32bit address bus, support for multiprocessing >and more. Because usable Unix is available for them both. Any more questions? -- Chip Salzenberg at ComDev/TCT <chip@tct.uucp>, <uunet!ateng!tct!chip>
yeung@eecs.cs.pdx.edu (Woodrow Yeung) (07/10/90)
In article <2329@crdos1.crd.ge.COM> davidsen@crdos1.crd.ge.com (billdavidsen) writes: > Since the only low prices UNIX around is V.3.2 (and maybe V.4 soon) >for the 386, that is the choice. Since the cost of the DX chip is >artificially high (I would bet the manufacturing and development cost is >within 10% of the SX) the SX is the only practical choice. I sure hope the SX is only 10% more expensive to make. It would be pretty silly for Intel to put any more engineering effort in making the 386SX than to disable some 16 line on the 386. Thus, the 386SX chip is a more expensive chip, Cost = 386 + lobotomy. Now if aluminum were as cheap as silicon, we'd be seeing 5.0L 4 cylinder Porsches that look identical to the 928's V8; the only difference would be 4 faulty fuel injectors that can't be fixed. Nah, that would be too obvious of a case of price discrimination. Hey any law students reading this? You should call up AMD, Harris or even NEC and launch an illegal price competition suit. Oh! I forgot, only Motorola gets nailed by lawsuits. ;-) yeung@reed yeung@eecs.cs.pdx.edu yeung!@lclark
seanf@sco.COM (Sean Fagan) (07/10/90)
In article <2329@crdos1.crd.ge.COM> davidsen@crdos1.crd.ge.com (bill davidsen) writes: >|we could also use a [...] 80286 >Those processors would cost a lot more because there is no >mass production of systems and vendor competition to keep the price >down. Huh? The '286 is produced by *several* people. Not just machines, but the *processor* (AMD, Harrisson, Intel). This is why Intel wants people to go to the 386SX so much. (Last I heard, AMD was in some arbitration with Intel about the rights to produce the '386 [this was after the court decision]; hopefully, they will be able to do so.)
davidsen@crdos1.crd.ge.COM (Wm E Davidsen Jr) (07/10/90)
In article <3115@psueea.UUCP> yeung@eecs.UUCP (Woodrow Yeung) writes: | I sure hope the SX is only 10% more expensive to make. It would be pretty | silly for Intel to put any more engineering effort in making the 386SX than | to disable some 16 line on the 386. Thus, the 386SX chip is a more expensive | chip, Cost = 386 + lobotomy. Since the logic is in the DX bus interface unit to build 32 bit fetches from 8 and 16 bit datapaths, I think all that's needed (more of less) is to set the 16 bit datapath line active. That much is true. However, note that most other chip manufacturers have not put logic in the BIU to work with narrow data paths or offset addresses. If you try to fetch 32 bits starting at a location not a multiple of four on a SPARC you get a bus error. This is a matter of design philosophy: should the chip do what I say, even if it isn't fast, or should the chip leave out logic to allow me to do what I want, so that there will not be slow instructions. Note that the 68k chip in a Sun3 also allows offset addressing, while several mainframe systems don't. This is not a CISC vs RISC issue, since some CISC systems restrict addressing. Since it's easier to not handle anything except bus access in a single read, I don't see where a chip which allows more complex bus access modes is inferior. -- bill davidsen (davidsen@crdos1.crd.GE.COM -or- uunet!crdgw1!crdos1!davidsen) "Stupidity, like virtue, is its own reward" -me
silos@bench.sublink.ORG (Paolo Pennisi) (07/14/90)
In article <2698CE85.50A3@tct.uucp>, chip@tct.uucp (Chip Salzenberg) writes: > According to silos@bench.sublink.ORG (Paolo Pennisi): > >How can you compare an handicapped version of a chip whose father and > >grandfather are crippled 16 bit segmented processor, [...] to a real > >32bit, semi Harvard, data and instruction cache processor, with a real > >32 bit data bus a a real 32bit address bus, support for multiprocessing > >and more. > > Because usable Unix is available for them both. > > Any more questions? > -- > Chip Salzenberg at ComDev/TCT <chip@tct.uucp>, <uunet!ateng!tct!chip> Yes, Xenix 286 on a 286 clone is usable, I'm using it right now, posting this news, but... what about: segments, 64k segment limits, near & far pointer, large model compiles, 16bit integers, stacks overflow, hours spent on PD software to catch a bug, which with a normal 32bit processor would'nt hard more than a warning... (pointers (far:32bit) to int (16but). Yes, it is usable, to receive and send e-mail, to read and post news (if u get the bug i mentioned or u have the patched version). Yes, with a 386SX, all these problems are gone... but SX at clones are slower than 286 to run DOS, OS/2... let's think of Unix... I can hardly stand a comparison with a 386DX... the SX is far away, the 286 doesn't exist (wot... my computer is vanishing....acc!). Let's have a look of the prices again... 68030-16Mhz, sold in quantity (1k) for 95$, 68882-16Mhz for just 44$. These are clone prices... Let's start to think why Intel keeps it prices for inferior silicon so high? Well, I gone away from the main path, it is my flame fever... I'm sorry. No more questions! Paolo. -- (ARPA) silos@bench.sublink.ORG Paolo Pennisi (BANG) ...!deejay!bench!silos via Solari 19 (MISC) ppennisi on BIX & PTPOSTEL 20144 Milano ITALIA ----< S U B L I N K N E T W O R K : a new way to *NIX communications >-----
wbeebe@bilver.UUCP (Bill Beebe) (07/15/90)
In article <1990Jul10.105047.9817@sco.COM> seanf@sco.COM (Sean Fagan) writes: >In article <2329@crdos1.crd.ge.COM> davidsen@crdos1.crd.ge.com (bill davidsen) writes: >>|we could also use a [...] 80286 >>Those processors would cost a lot more because there is no >>mass production of systems and vendor competition to keep the price >>down. > >Huh? The '286 is produced by *several* people. Not just machines, but the >*processor* (AMD, Harrisson, Intel). This is why Intel wants people to go >to the 386SX so much. (Last I heard, AMD was in some arbitration with Intel >about the rights to produce the '386 [this was after the court decision]; >hopefully, they will be able to do so.) In the last quarter for which such statistics are available (I believe Q1 90), some 10 million 80286 devices were shipped by AMD, Harris, and Intel, with the largest share by AMD. Reasons for the continuing upsurge in shipments are the lower price for 16-, 20-, and 25Mhz 80286 versus the equivalent 80386SX. Intel wants about $65 for the SX in 1000s at 16Mhz, while the going price for the 80286 in 1000s is between $20 and $30. These 80286 chips at 16Mhz are true CMOS (Intel can't make anything faster than 12Mhz NMOS). Another equally compelling reason for the use of the 80286 is the current allocation of the 386 family, including the SX. Oh yes, Intel shipped only 1 million 80386SX chips in the same period, and because of very high demand, the spot-market (broker) price for the 80386SX is about $200. The difference in units shipped does not bother Intel, since they claim they made about as much profit shipping the 1 million SX chips as everyone else made shipping the 10 million 80286s.
mmm@cup.portal.com (Mark Robert Thorson) (07/20/90)
yeung@eecs.cs.pdx.edu (Woodrow Yeung) says: > I sure hope the SX is only 10% more expensive to make. It would be pretty > silly for Intel to put any more engineering effort in making the 386SX than > to disable some 16 line on the 386. Thus, the 386SX chip is a more expensive > chip, Cost = 386 + lobotomy. Not quite. The SX comes in a plastic flat-pack. The other MS-DOS CPU's come in ceramic, so the SX should be significantly cheaper, percentage-wise, to manufacture. Of course, cost-of-production has nothing to do with the sale price, other than to place a lower bound on it.