jonas@convex.csd.uwm.edu (Jonas K Manske) (04/25/91)
I everyone....
I am looking for people to let me know what they think about the NeXT..
1) Will it ever become popular? When people say, DECStations, you
what they mean. When you say, NeXT, people go what?
2) Does it have the capability to grow with the technology of the times?
If so, will it? Steve? ie, Will it become obsoleted?
I am just wondering,..... I am considering buying one, it has a few nice
features that interest me, but then again so does the Amiga. The Amiga
has a big problem in that it gets little coverage. I don't want a UNIX
workstation like that....
Any comments? Opinions? And yes, even flames....
I will summarize if response makes it worthwhile...
====
Jonaspeter@ficc.ferranti.com (peter da silva) (04/27/91)
In article <11399@uwm.edu>, jonas@convex.csd.uwm.edu (Jonas K Manske) writes: > I am just wondering,..... I am considering buying one, it has a few nice > features that interest me, but then again so does the Amiga. The Amiga > has a big problem in that it gets little coverage. I don't want a UNIX > workstation like that.... The NeXT and the Amiga 3000UX have the same problem here: they're 68000 based, so you won't expect any outstanding improvements in the next few years. Even if NeXT and Commodore make like Sun and switch to RISC processors your existing machine is likely to be left behind. Though the asynchronous bus on the Amiga means that a RISC coprocessor board would be quite a workable compromise. The NeXT bus is much slower, more designed as a peripheral bus instead of a general backplane like the Amiga's. -- Peter da Silva. `-_-' peter@ferranti.com +1 713 274 5180. 'U` "Have you hugged your wolf today?"
jtr@oakhill.sps.mot.com (Jim Reinhart) (04/29/91)
In article <JH_ABJF@xds13.ferranti.com> peter@ficc.ferranti.com (peter da silva) writes: >In article <11399@uwm.edu>, jonas@convex.csd.uwm.edu (Jonas K Manske) writes: >> I am just wondering,..... I am considering buying one, it has a few nice >> features that interest me, but then again so does the Amiga. The Amiga >> has a big problem in that it gets little coverage. I don't want a UNIX >> workstation like that.... > >The NeXT and the Amiga 3000UX have the same problem here: they're 68000 >based, so you won't expect any outstanding improvements in the next few >years. Even if NeXT and Commodore make like Sun and switch to RISC processors STUFF DELETED. >-- >Peter da Silva. `-_-' peter@ferranti.com >+1 713 274 5180. 'U` "Have you hugged your wolf today?" Gargage. What exactly qualifies as "outstanding" and what do you know about the price/performance curve of 68xxx machines to make this assertion? -- Regards, Jim Reinhart Motorola Microprocessor Products Group Austin, Texas
sef@kithrup.COM (Sean Eric Fagan) (04/29/91)
In article <1991Apr29.144421.19819@oakhill.sps.mot.com> jtr@oakhill.UUCP (Jim Reinhart) writes: >Gargage. What exactly qualifies as "outstanding" and what do you know about >the price/performance curve of 68xxx machines to make this assertion? A four- or five-fold performance increase at roughly the same price. The 68040 is about as close as it can get (oh, some more of the instructions can be sped up, and it can be made superscalar [although I am not convinced of this, nor even that a superscalar 68k chip would be that advantageous on current code], and, more likely, the clock rate can be bumped up, but all of these aren't going to happen soon, nor are they going to be *cheap*). Meanwhile, an R4000 machine, based on everything I've seen and heard, should run for about $10k or so (monochrome, I'm sure) and should be quite speedy. But you're right: we don't know what motorola is going to do in the immediate future. Want to tell us, or are you just trying to defend your company on generic grounds? -- Sean Eric Fagan | "I made the universe, but please don't blame me for it; sef@kithrup.COM | I had a bellyache at the time." -----------------+ -- The Turtle (Stephen King, _It_) Any opinions expressed are my own, and generally unpopular with others.
peter@ficc.ferranti.com (peter da silva) (04/30/91)
I made the effective claim that the 68000 (and thus the Amiga/NeXT) won't have any outstanding performance improvements in the next few years. In article <1991Apr29.144421.19819@oakhill.sps.mot.com>, jtr@oakhill.sps.mot.com (Jim Reinhart) writes: > Gargage. What exactly qualifies as "outstanding" and what do you know about > the price/performance curve of 68xxx machines to make this assertion? I have followed the 68000 family closely for the past decade, being a confirmed 68000 fan in the ongoing Motorola vs. Intel race. My primary computer is an Amiga 3000 with a Motorola 68030 processor. I also have an Amiga 1000, and I had an Atari ST, both of which use the earlier 68000. At the low end of the PC market, the 68000 has had the high-performance PC business pretty much to itself for most of the past decade, though with recent chips Intel has managed to catch up and twice briefly pass Motorola in performance. Neither the Intel nor Motorola CISC chips show any sign of regaining the performance edge from RISC, and it's only the existing base of 68000 and 8086 commodity PCs that has kept the system cost of 680x0 and 80x86 below the RISC chips. Both Intel and Motorola have their own RISC projects, with Motorola maintaining its traditional lead in general-purpose CPU performance in this field too. -- Peter da Silva. `-_-' peter@ferranti.com +1 713 274 5180. 'U` "Have you hugged your wolf today?"
2004ktz@ucsbuxa.ucsb.edu (David G. Koontz) (04/30/91)
In article <1991Apr29.144421.19819@oakhill.sps.mot.com> jtr@oakhill.sps.mot.com (Jim Reinhart) writes: >In article <JH_ABJF@xds13.ferranti.com> peter@ficc.ferranti.com (peter da silva) writes: >>In article <11399@uwm.edu>, jonas@convex.csd.uwm.edu (Jonas K Manske) writes: >>> I am just wondering,..... I am considering buying one, it has a few nice >>> features that interest me, but then again so does the Amiga. The Amiga >>> has a big problem in that it gets little coverage. I don't want a UNIX >>> workstation like that.... >> >>The NeXT and the Amiga 3000UX have the same problem here: they're 68000 >>based, so you won't expect any outstanding improvements in the next few >>years. Even if NeXT and Commodore make like Sun and switch to RISC processors How about a DEC 3100? Got a telegram from then saying the price was dropped to $5K. .
davidsen@crdos1.crd.ge.COM (Wm E Davidsen Jr) (04/30/91)
In article <+=+A+N6@xds13.ferranti.com> peter@ficc.ferranti.com (peter da silva) writes: | Neither the Intel nor Motorola CISC chips show any sign of regaining the | performance edge from RISC, and it's only the existing base of 68000 and | 8086 commodity PCs that has kept the system cost of 680x0 and 80x86 below | the RISC chips. Perhaps integrating the cache, MMU, and FPU on the same chip has had some effect? A savings which I see is now being copied in RISC. Chips like the Intel 32 bit CISC offerings do make design a lot easier, and because of fewer support ships and connection smaller, cheaper, and more reliable. Not that I deny the ecconomies of scale, but there are technical advantages, too. And competition has driven the price down a lot, something we are seeing in the SPARC world. Multiple sources usually bring the price down. -- bill davidsen (davidsen@crdos1.crd.GE.COM -or- uunet!crdgw1!crdos1!davidsen) "Most of the VAX instructions are in microcode, but halt and no-op are in hardware for efficiency"
umh@vax5.cit.cornell.edu (05/01/91)
In article <1991Apr29.164102.11221@kithrup.COM>, sef@kithrup.COM (Sean Eric Fagan) writes: > A four- or five-fold performance increase at roughly the same price. > > Meanwhile, an R4000 machine, based on everything I've seen and heard, should > run for about $10k or so (monochrome, I'm sure) and should be quite speedy. A MIPS rep today told me to expect ACE machines (the compaq/microsoft etc group) at $2K to $5K in Q2 92.Of course these will be no cache except the 8K+8K on the R4000, and crippled in various other ways to be as cheap as possible, but will presumably still be far above PC/Mac/Amiga performance. Maynard Handley
src@scuzzy.in-berlin.de (Heiko Blume) (05/01/91)
2004ktz@ucsbuxa.ucsb.edu (David G. Koontz) writes: >In article <1991Apr29.144421.19819@oakhill.sps.mot.com> jtr@oakhill.sps.mot.com (Jim Reinhart) writes: >>>The NeXT and the Amiga 3000UX have the same problem here: they're 68000 >>>based, so you won't expect any outstanding improvements in the next few >>>years. Even if NeXT and Commodore make like Sun and switch to RISC processors >How about a DEC 3100? Got a telegram from then saying the price was >dropped to $5K. >. i'd guess that's the price without the keyboard, which probably costs $3000. seriously, the keyboard of my vt220 just died, and when i called DEC they told me that if i bring them the broken one and pay THREE HUNDRED AND FIFTY DOLLARS (aprox) i'll get a new keyboard!! considering that i can buy a brand new complete vt320 clone for about $50 more, that's what i call a RIPOFF. given that i wouldn't buy a ball pen from DEC, since it'll probably cost $50 to replace that ink thing inside. -- Heiko Blume <-+-> src@scuzzy.in-berlin.de <-+-> (+49 30) 691 88 93 [voice!] public UNIX source archive [HST V.42bis]: scuzzy Any ACU,f 38400 6919520 gin:--gin: nuucp sword: nuucp uucp scuzzy!/src/README /your/home
miklg@sono.uucp (Michael Goldman ) (05/01/91)
jtr@oakhill.sps.mot.com (Jim Reinhart) writes: >In article <JH_ABJF@xds13.ferranti.com> peter@ficc.ferranti.com (peter da silva) writes: >>In article <11399@uwm.edu>, jonas@convex.csd.uwm.edu (Jonas K Manske) writes: >>> I am just wondering,..... I am considering buying one, it has a few nice >>> features that interest me, but then again so does the Amiga. The Amiga >>> has a big problem in that it gets little coverage. I don't want a UNIX >>> workstation like that.... >> >>The NeXT and the Amiga 3000UX have the same problem here: they're 68000 >>based, so you won't expect any outstanding improvements in the next few >>years. Even if NeXT and Commodore make like Sun and switch to RISC processors >STUFF DELETED. >>-- >>Peter da Silva. `-_-' peter@ferranti.com >>+1 713 274 5180. 'U` "Have you hugged your wolf today?" >Gargage. What exactly qualifies as "outstanding" and what do you know about >the price/performance curve of 68xxx machines to make this assertion? >-- >Regards, >Jim Reinhart >Motorola Microprocessor Products Group >Austin, Texas I think the question is concerning market viability. No one knows the future but I heard from a friend at a software company that supports NeXT that the company got paid in NeXT boxes, not cash. He hears that NeXT will probably not make it, but then they said that about the Macintosh for its first 2 years so who knows? I think it depends on what you want your computer for. If its something that doesn't require constant upgrades, then your computer will last a lonnngg time. I've been content to let my company buy the latest, while I continue doing my home stuff on an 8088. As for CPU upgrades, I see the SPEC marks and other benchmarks showing CISC keeping up with RISC within reasonable limits. I don't see any reason why CISC can't keep performance up with a lag time of a year or 2, which is negiligible in terms of making a box, getting it to market, and getting the price down to reasonable levels. I've seen used SUNs on the market for the price of a new Mac or NeXT. Why not get a Sun. Regards, Michael Goldman (All opinions mine) "I do not know what I do not know" - Ludwig Wittgenstein
melling@cs.psu.edu (Michael D Mellinger) (05/02/91)
(Michael Goldman ) writes:
I think the question is concerning market viability. No one knows the future
but I heard from a friend at a software company that supports NeXT that
the company got paid in NeXT boxes, not cash. He hears that NeXT will probably
not make it, but then they said that about the Macintosh for its first 2 years
so who knows? I think it depends on what you want your computer for. If its
something that doesn't require constant upgrades, then your computer will
last a lonnngg time. I've been content to let my company buy the latest,
while I continue doing my home stuff on an 8088.
Are people at NeXT telling him this :-)? Wall Street types?
As for CPU upgrades, I see the SPEC marks and other benchmarks showing
CISC keeping up with RISC within reasonable limits. I don't see any reason
why CISC can't keep performance up with a lag time of a year or 2, which is
negiligible in terms of making a box, getting it to market, and getting
the price down to reasonable levels.
NeXT is rumored to be working on an 88K machine. Motorola has made
some pretty big claims about the performance of future versions of the
chip. So, I think part of NeXT's future depends on how good the 88K
turns out to be. For now, though, NeXT computers are competitive with
the SPARCstations in their price range. Although, it looks like HP
has redefined price/performance.
I've seen used SUNs on the market for the price of a new Mac or NeXT. Why
not get a Sun.
Because the NeXT's are faster, have a better user interface, and all
in all are nicer machines.
-Mikegillies@m.cs.uiuc.edu (Don Gillies) (05/02/91)
Can someone make a educated guess as to the wholesale cost of the NeXT, not including manufacturing costs? I wonder what the markup is on a university-discounts NeXT, and on a $4995 NeXT. - Don Gillies --
jsh44765@uxa.cso.uiuc.edu (Jonathan S Hofmann) (05/02/91)
>>>>The NeXT and the Amiga 3000UX have the same problem here: they're 68000 >>>>based, so you won't expect any outstanding improvements in the next few >>>>years. Even if NeXT and Commodore make like Sun and switch to RISC processors >>How about a DEC 3100? Got a telegram from then saying the price was >>dropped to $5K. >>. Well, I've heard that NeXT is eyeing the Motorola 88k series for an expansion board for the Cube on comp.sys.next. Other than that, student price, I can get a slab (monochrome, 105Mb HD) for $3500-- about the STUDENT price of a Mac SE/30, for a full-blown UNIX workstation capable of X, among other equally impressive things. (The retail price is $5k). Cubes run $5k, student. Scott Hofmann hofmann@sumter.cso.uiuc.edu (NeXT) Independant NeXT developer
TROTH@RICEVM2.RICE.EDU (Rick Troth) (05/03/91)
NeXT survive? Do you mean the company or the machine? The company has introduced some slick new products, and they've changed their marketting strategy (picked a broader/better market). But if NeXT, Inc. went under, then the cubes (including the new boxes) would become "surplus", which would make me happy as a dog with a bone. :-I Being a ham/hobbyist, I'm more than willing to put in some time to keep-up what a vendor won't. If NeXT cubes were surplus, then I could afford to buy one for home and not just tell my wife stories about my workstation at the office. I think the cube will survive just fine. NeXTOS 2.0 is actually worth using. I don't even use X much anymore since I got 2.0 with it's new Terminal program (basically Stuart). Read comp.sys.next if you want the names of some real NeXT enthusiasts. THEY're the ones who will determine if the company and/or the machine will survive. THEY're the ones who will be supporting the machine if the company doesn't. Fundamentally, they did the right thing: they built the "pretty, GUI [pronounced 'gooey'?], click-your-rodent" environment on a real O/S base. Mach was an excellent choice. In truth, it's not UNIX, though many folks can't tell, and some folks don't need to know and don't care. You asked for opinions. I hope that helps. -- "The tomb is empty" Rick Troth <TROTH@RICEVM1.RICE.EDU> ------------- Rice ONCS VM Systems Support
troth@rio-grande.rice.edu (Richard M Troth) (05/03/91)
Organization: Rice University, Houston, TX. Date: Thursday, 2 May 1991 16:13:53 CDT From: Rick Troth <TROTH@RICEVM2.RICE.EDU> Message-ID: <91122.161353TROTH@RICEVM2.RICE.EDU> Newsgroups: comp.arch Subject: Re: Will NeXT survive? Grow with the times? References: <11399@uwm.edu> NeXT survive? Do you mean the company or the machine? The company has introduced some slick new products, and they've changed their marketting strategy (picked a broader/better market). But if NeXT, Inc. went under, then the cubes (including the new boxes) would become "surplus", which would make me happy as a dog with a bone. :-I Being a ham/hobbyist, I'm more than willing to put in some time to keep-up what a vendor won't. If NeXT cubes were surplus, then I could afford to buy one for home and not just tell my wife stories about my workstation at the office. I think the cube will survive just fine. NeXTOS 2.0 is actually worth using. I don't even use X much anymore since I got 2.0 with it's new Terminal program (basically Stuart). Read comp.sys.next if you want the names of some real NeXT enthusiasts. THEY're the ones who will determine if the company and/or the machine will survive. THEY're the ones who will be supporting the machine if the company doesn't. Fundamentally, they did the right thing: they built the "pretty, GUI [pronounced 'gooey'?], click-your-rodent" environment on a real O/S base. Mach was an excellent choice. In truth, it's not UNIX, though many folks can't tell, and some folks don't need to know and don't care. You asked for opinions. I hope that helps. -- "The tomb is empty" Rick Troth <TROTH@RICEVM1.RICE.EDU> ------------- Rice ONCS VM Systems Support
dmocsny@minerva.che.uc.edu (Daniel Mocsny) (05/03/91)
In article <1991May1.160128.1367@sono.uucp> miklg@sono.uucp (Michael Goldman ) writes: >As for CPU upgrades, I see the SPEC marks and other benchmarks showing >CISC keeping up with RISC within reasonable limits. I don't see any reason >why CISC can't keep performance up with a lag time of a year or 2, which is >negiligible in terms of making a box, getting it to market, and getting >the price down to reasonable levels. 2 years certainly is negligible in terms of developing a base of application software. RISC is 2 years ahead in hardware speed, but it is AT LEAST 2 years behind in applications base, according to the overwhelming majority of customers. If RISC gets 5 years ahead of CISC in hardware speed, that will be enough to counter its software lag, which is probably of the same order. (I.e., if present trends continue, the software available in 5 years for the current major RISC families, assuming they preserve binary compatibility, *might* resemble the software available NOW for the 80x86 family, in terms of range and price. However, leading-edge hardware seems hard-pressed to maintain binary compatibility very far backwards. This is *precisely* the software advantage and hardware disadvantage of CISC.) Do comp.arch pundits see RISC chips widening their gap over the CISC chips? If the performance gap stays at a factor of two or three, that represents a ~2 year hardware delay for CISC, which doesn't seem large compared to the ~5 year software lead for CISC. Also, consider that very few individual users are able to keep even a '386 machine busy all the time. To do even that is an intense, full-time job for anyone whose problem can't be solved in one "for" loop. I would guess that the average PC or Mac gets used a lot more than the average Sparcstation, primarily because the average PC or Mac user can afford more applications. The average useful program, if available under UNIX, costs 2--10 times as much as the DOS or Mac equivalent. This problem isn't going away anytime soon. Of course, since CISC and RISC are Turing-equivalent, there is no fundamental reason why the software lead of CISC should be immutable. In the real world, however, transparent portability across architectures appears elusive, for a variety of reasons, and most of them seem rather silly. -- Dan Mocsny Internet: dmocsny@minerva.che.uc.edu
dhesi%cirrusl@oliveb.ATC.olivetti.com (Rahul Dhesi) (05/04/91)
In <8283@uceng.UC.EDU> dmocsny@minerva.che.uc.edu (Daniel Mocsny) writes: >RISC is 2 years ahead in hardware speed, but >it is AT LEAST 2 years behind in applications base... Are vendors still writing software targeted to specific CPUs? Why? -- Rahul Dhesi <dhesi@cirrus.COM> UUCP: oliveb!cirrusl!dhesi
sef@kithrup.COM (Sean Eric Fagan) (05/04/91)
In article <3111@cirrusl.UUCP> Rahul Dhesi <dhesi@cirrus.COM> writes: >In <8283@uceng.UC.EDU> dmocsny@minerva.che.uc.edu (Daniel Mocsny) writes: >>RISC is 2 years ahead in hardware speed, but >>it is AT LEAST 2 years behind in applications base... >Are vendors still writing software targeted to specific CPUs? Why? I think it's more the case that everything is behind DOS. Also, various vendors add things to their systems which cause incompatabilities. For example, SCO XENIX and UNIX have some ioctl's on the console which don't exist under other vendors' versions. As a result, if an application was written under sco *nix, then it would take some (possibly non-trivial) work to port it to another version of *nix. But it's mostly a software issue, not hardware. More accurately, it's a *system* issue. -- Sean Eric Fagan | "I made the universe, but please don't blame me for it; sef@kithrup.COM | I had a bellyache at the time." -----------------+ -- The Turtle (Stephen King, _It_) Any opinions expressed are my own, and generally unpopular with others.
sjc@borland.com (Steve Correll) (05/04/91)
In article <8283@uceng.UC.EDU> dmocsny@minerva.che.uc.edu (Daniel Mocsny) writes: >Also, consider >that very few individual users are able to keep even a '386 machine >busy all the time. Gee, from my experience, X windows can keep any processor saturated without any action whatsoever on the part of the individual user. :-) Two serious observations: 1. Many NeXT users have said that the performance of the 68K is the main thing standing between them and nirvana. 2. It is an illusion that the 80x86 PC world has fewer compatibility problems than the RISC world. 80x86 PC software vendors expend enormous effort to preserve this illusion for the benefit of their users: they support multiple graphics options, multiple mice, 80386 real-versus-protected mode, various add-on memory managers, Microsoft Windows versus straight DOS versus OS/2 versus 32-bit DPMI, various 80x86 subroutine calling conventions, tiny versus small versus huge memory models (and on and on). Graphics, network, and mouse differences are often exposed to the application program rather than being hidden behind an operating system. QA compatibility testing is a big expense in the PC software world. This effort, if redirected, would be more than sufficient to port the applications to a single operating system on a RISC processor in short order. Before redirecting the effort, however, the prudent vendor must judge it profitable based on sales volume and market risks.
melling@cs.psu.edu (Michael D Mellinger) (05/04/91)
In article <1991May4.011456.25729@borland.com> sjc@borland.com (Steve Correll) writes:
Two serious observations:
1. Many NeXT users have said that the performance of the 68K is the main thing
standing between them and nirvana.
The 68040 NeXT comes pretty close. The machines with only 8MB of RAM
can drive you crazy when are running large programs, but $400 more
does solve that problem. With the educational prices running around
$3300, there isn't anything better(IMHO). As well as having nice
system software, the NeXT still gives a very competitive SPEC/$.
2. It is an illusion that the 80x86 PC world has fewer compatibility problems
than the RISC world. 80x86 PC software vendors expend enormous effort to
preserve this illusion for the benefit of their users: they support multiple
graphics options, multiple mice, 80386 real-versus-protected mode, various
add-on memory managers, Microsoft Windows versus straight DOS versus OS/2
versus 32-bit DPMI, various 80x86 subroutine calling conventions, tiny
versus small versus huge memory models (and on and on). Graphics, network,
and mouse differences are often exposed to the application program rather
than being hidden behind an operating system. QA compatibility testing is a
big expense in the PC software world.
Don't worry, Microsoft has it all figured out. OS/2 NT will be binary
compatible with everything :-). Actually, I hope IBM starts giving
OS/2 2.0 away to straigten out this mess.
-Mikeguy@auspex.auspex.com (Guy Harris) (05/05/91)
>However, leading-edge hardware seems hard-pressed to maintain binary >compatibility very far backwards. Examples, please? Yes, MIPS, SPARC, and HP-PA have all introduced additional user-mode-visible features to their architectures, but they haven't, as far as I know, invalidated any *old* stuff (other than blowing off "extended precision" in favor of "quad" floating point in SPARC, or something such as that, but I don't know that anybody was using the "extended precision" stuff, and don't know that anybody'd implemented it in hardware). Or are you thinking of changes that, while they don't require programs to be recompiled in order to run at all, require them to be recompiled in order to get the full performance boost from a new implementation (e.g., different instruction scheduling)?
dmocsny@minerva.che.uc.edu (Daniel Mocsny) (05/05/91)
In article <1991May4.011456.25729@borland.com> sjc@borland.com (Steve Correll) writes: >In article <8283@uceng.UC.EDU> dmocsny@minerva.che.uc.edu (Daniel Mocsny) writes: >>Also, consider >>that very few individual users are able to keep even a '386 machine >>busy all the time. >Gee, from my experience, X windows can keep any processor saturated without any >action whatsoever on the part of the individual user. :-) Out of curiosity, how many individual users can set up and maintain *their own* computer running X? (The ideal answer should be: all of them.) That was my point, anyway. Some of my best friends are RISC users, you know. :-) Must I now pay with my life, my fortune, and my sacred (?) net.honor for having egregiously blown the shibboleth of comp.arch? (I.e., by daring to speculate that a 386 could in any sense suffice for anyone.) >2. It is an illusion that the 80x86 PC world has fewer compatibility problems > than the RISC world. 80x86 PC software vendors expend enormous effort to > preserve this illusion for the benefit of their users: they support multiple > graphics options, multiple mice, 80386 real-versus-protected mode, various > add-on memory managers, Microsoft Windows versus straight DOS versus OS/2 > versus 32-bit DPMI, various 80x86 subroutine calling conventions, tiny > versus small versus huge memory models (and on and on). Yes, but they can do this (in principle) from one box, and they can propagate their success to other 80x86 box programmers in the form of libraries that mask most of the fragmentation. Illusion or not, it's led to the largest pile of software that customers want to run, so badly they buy lots of inferior hardware to run it. The only incompatibilities that matter are those which the *application programmer* can't hide from. That's where the productivity evaporates, when the application coders have to get sidetracked by all this garbage. -- Dan Mocsny Internet: dmocsny@minerva.che.uc.edu
dmocsny@minerva.che.uc.edu (Daniel Mocsny) (05/05/91)
In article <7628@auspex.auspex.com> guy@auspex.auspex.com (Guy Harris) writes: I wrote: >>However, leading-edge hardware seems hard-pressed to maintain binary >>compatibility very far backwards. > >Examples, please? Yes, MIPS, SPARC, and HP-PA have all introduced >additional user-mode-visible features to their architectures, but they >haven't, as far as I know, invalidated any *old* stuff (other than Will MIPS, SPARC, or HP-PA CPU run any binary compiled before the year 1985? That is what I meant by "very far backwards". (See that I qualified my claim with "seems": I don't claim to have expertise here.) Here is another question for the comp.arch pundits: Are we likely to see the fastest CPU in year X being able to run, without change, a binary program more than 5 years old? (Does HP-PA do this right now? If so, I am very impressed. I would be much more impressed if it could also run the large existing libraries of CISC binaries at full speed, but that would be asking quite a bit :-) Already, I have seen claims (possibly biased) that SPARC is now sacrificing performance to maintain compatibility, and will therefore not reach the top of the RISC pile again. Is this inescapable? Must advancing hardware technology always obsolete software? Would someone like to plot each leading CPU's performance vs. the age of the oldest compiled binary each CPU can run (at all)? Express CPU performance both as absolute, and normalized by clock rate, if possible. My hunch would be that the chips with the shortest history of binary compatibility tend to be the fastest, especially when normalized by clock rate. -- Dan Mocsny Internet: dmocsny@minerva.che.uc.edu
peter@ficc.ferranti.com (peter da silva) (05/06/91)
In article <3111@cirrusl.UUCP>, dhesi%cirrusl@oliveb.ATC.olivetti.com (Rahul Dhesi) writes: > In <8283@uceng.UC.EDU> dmocsny@minerva.che.uc.edu (Daniel Mocsny) writes: > >RISC is 2 years ahead in hardware speed, but > >it is AT LEAST 2 years behind in applications base... > Are vendors still writing software targeted to specific CPUs? Why? Can't teach an old dog new tricks. Check out some of the PC groups... even when you have a 68030 and 2 MB RAM minimum to play with, there are people claiming that they need to use 68030-specific assembly for things like text editors! Yes, I mean version-specific assembly. Now maybe for the inner loop of a ray tracer I could see it, but a text editor? But, yes, even when the O/S is good enough and the chips fast enough you don't ever have to write assembly code, they keep doing it. And applications programmers are bad enough... you should see some of the things game programmers do. Blow away the O/S and make direct calls to known locations in ROM (jump tables are too slow). Then they wonder why their programs die with new releases of the software. -- Peter da Silva. `-_-' peter@ferranti.com +1 713 274 5180. 'U` "Have you hugged your wolf today?"
peter@ficc.ferranti.com (peter da silva) (05/06/91)
In article <z!7G1fw=1@cs.psu.edu>, melling@cs.psu.edu (Michael D Mellinger) writes: > The 68040 NeXT comes pretty close. The machines with only 8MB of RAM > can drive you crazy when are running large programs, but $400 more > does solve that problem. With the educational prices running around > $3300, there isn't anything better(IMHO). Hi Mike! Here you are with those educational prices again... I told you I'd take you into the loony bin with me if you did that again. :-> The NeXTstation costs 5 grand. I can practically buy a new car for that. The competitors for the NeXT in the real world cost O(2 grand). Oh sure, it's pretty fast (right at the low end of RISC chips), but what's the market? From my copy of the NeXT software catalog, it's business and engineering. These people already have Macs and DOS machines (and the 486 and 68040 are pretty close, so you get the same Specmarks), or if they're using workstations they're going for the high end of the RISC world... not a machine only a little slower than a Sparc. -- Peter da Silva. `-_-' peter@ferranti.com +1 713 274 5180. 'U` "Have you hugged your wolf today?"
kls30@duts.ccc.amdahl.com (Kent L Shephard) (05/06/91)
In article <91122.161353TROTH@RICEVM2.RICE.EDU> TROTH@RICEVM2.RICE.EDU (Rick Troth) writes: > >Fundamentally, they did the right thing: they built the "pretty, GUI >[pronounced 'gooey'?], click-your-rodent" environment on a real O/S base. >Mach was an excellent choice. In truth, it's not UNIX, though many ~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~ No, No, No. The NeXT does run UNIX. I runs a Mach kernal under UNIX. >folks can't tell, and some folks don't need to know and don't care. IMHO - NeXT will survive. But my opinion is very biased. You see, I own one and have to think they would survive. I'm not in the habit of wishing my tools become unsupported. >-- > "The tomb is empty" > Rick Troth <TROTH@RICEVM1.RICE.EDU> ------------- Rice ONCS VM Systems Support KeNT -- /* -The opinions expressed are my own, not my employers. */ /* For I can only express my own opinions. */ /* */ /* Kent L. Shephard : email - kls30@DUTS.ccc.amdahl.com */
sjc@borland.com (Steve Correll) (05/07/91)
In article <8323@uceng.UC.EDU> dmocsny@minerva.che.uc.edu (Daniel Mocsny) writes: >Out of curiosity, how many individual users can set up and >maintain *their own* computer running X? I watched five users here install Windows 3.0 with a particular variety of 1024x768 super-VGA card and handler. Only three succeeded, and one of those by using a text editor to put a magic character into the system.ini file (he didn't know exactly why that worked). Symptoms differed (setup.exe locked the machine, or prompted for a diskette which it then rejected, or prompted for a diskette whose name it couldn't display) and the magic character worked only in one case. Nobody seemed particularly surprised. >The only incompatibilities that matter are those which the *application >programmer* can't hide from. That's where the productivity evaporates, >when the application coders have to get sidetracked by all this garbage. I think the 80x86 environment is difficult and productivity-sapping for both application programmers (who certainly do have to deal with "all this garbage", lest users complain "Your application doesn't work when my input file is on the network" or "Your application doesn't work when I run it from a DOS shell if the DOS shell is running inside Windows" or "Your application dies mysteriously when I install the handler for the new mouse I bought over the weekend" or "Your application claims I'm out of memory even though Quarterdeck says I have lots left") and for users, who must memorize a complicated matrix of allowed combinations. Fortunately we amortize this tremendous cost over many units, and fortunately users share arcane knowledge by word of mouth. If the following fairly summarizes your viewpoint, it seems true to me: The 80x86 was there first; lots of software is available for it; and lots of customers already own both the software and hardware: therefore, they are loathe to migrate to a different computing environment unless it is X times as fast and offers a set of Y applications. Anyone who knows for sure the values of X and Y will have a lucrative career in market forecasting.
melling@cs.psu.edu (Michael D Mellinger) (05/07/91)
In article <-U3BJU9@xds13.ferranti.com> peter@ficc.ferranti.com (peter da silva) writes:
The NeXTstation costs 5 grand. I can practically buy a new car for that.
Check out the price of an SE/30 two years ago. Guess what it costs
$5000. How much does a 386 computer from IBM or Compaq cost today?
And a 486? If I'm not mistaken a Model 90 costs around 10K. You
could get a fancy car for that much :-).
The competitors for the NeXT in the real world cost O(2 grand). Oh sure,
it's pretty fast (right at the low end of RISC chips), but what's the
market? From my copy of the NeXT software catalog, it's business and
engineering. These people already have Macs and DOS machines (and the 486
and 68040 are pretty close, so you get the same Specmarks), or if they're
using workstations they're going for the high end of the RISC world... not
a machine only a little slower than a Sparc.
Which computers can you buy from IBM, Compaq, Apple, Sun, Dec, etc,
that cost $2000? The NeXT is still competitive in SPEC/$, something
that not many of the PC makers can claim. A SSII goes for $14K, so
it's not in the same market as the NeXT. In short, you are comparing
computers without regard to how they fit into the market.
Something else you aren't considering, a 486 machine running DOS +
Windows is not going to be as versatile as the NeXT. Well, perhaps I
should wait for DOS 5.0's imminent release before I say such a nasty
thing. Maybe Microsoft will surprise me.
-Mikerthomson@mesa.dsd.es.com (Rich Thomson) (05/07/91)
In article <1991May4.011456.25729@borland.com> sjc@borland.com (Steve Correll) writes: Gee, from my experience, X windows can keep any processor saturated without any action whatsoever on the part of the individual user. :-) In article <8323@uceng.UC.EDU> dmocsny@minerva.che.uc.edu (Daniel Mocsny) writes: >Out of curiosity, how many individual users can set up and >maintain *their own* computer running X? (The ideal answer should >be: all of them.) Setup? Open up the boxes, plug in the cables and turn on the power. Maintain? backups Seriously, setting up the average *commerically produced* X workstation isn't any harder than setting up a PC. Maintenance isn't any more difficult either. (Note: most people don't properly maintain their PC until they lose their first important chunk of data). I have had to setup both, and I would actually say that setting up an X workstation is easier than setting up a PC. In PC land everything has a list of supported printers, display adapaters, etc. and you have to make sure that the support set of the software you buy intersects the set of hardware you have. This is because everybody codes directly to the hardware instead of some abstract interface because the latter is "too slow". -- Rich -- ``Read my MIPS -- no new VAXes!!'' -- George Bush after sniffing freon Disclaimer: I speak for myself, except as noted. UUCP: ...!uunet!dsd.es.com!rthomson Rich Thomson ARPA: rthomson@dsd.es.com PEXt Programmer
kevinh@cmi.com (Kevin Hegg) (05/07/91)
References:<1991Apr29.144421.19819@oakhill.sps.mot.com> <z!7G1fw=1@cs.psu.edu> <-U3BJU9@xds13.ferranti.com> This is probably better discussed in comp.sys.next but here goes anyway. peter@ficc.ferranti.com (peter da silva) writes > Hi Mike! Here you are with those educational prices again... I told you > I'd take you into the loony bin with me if you did that again. :-> > > The NeXTstation costs 5 grand. I can practically buy a new car for that. I think you have these mixed up. If you buy any computer for list then maybe you belong in the loony bin. :-) Seriously, if you want a Next and do not qualify for educational discounts then you can become a registered developer. Then you can purchase Next's at about a 30% discount. > The competitors for the NeXT in the real world cost O(2 grand). Where? If you list all of the hardware and software included in a Next you will find nothing on the market comparable for $2K. There are some in the same ball park as the Next. Kevin Hegg, EDS Corp - Center for Machine Intelligence 2001 Commonwealth Blvd., Ann Arbor, Michigan 48105 Phone: (313) 995-0900 Internet: kevinh@cmi.com Applelink: D5990
peter@ficc.ferranti.com (peter da silva) (05/07/91)
In article <rs4G8kw&1@cs.psu.edu>, melling@cs.psu.edu (Michael D Mellinger) writes: > In article <-U3BJU9@xds13.ferranti.com> peter@ficc.ferranti.com (peter da silva) writes: > The NeXTstation costs 5 grand. I can practically buy a new car for that. > Check out the price of an SE/30 two years ago. Guess what it costs > $5000. Check out the price of Macs today... they've gone down quite a bit. > How much does a 386 computer from IBM or Compaq cost today? I can get a 386SX clone with 1 MB of RAM for $875, and it runs Windows 3.0 just as well as those IBMs and Compaqs. Oh, sure, it costs a bit more to get a good VGA, a decent sized hard disk, etc... but 2 grand is about right. > Which computers can you buy from IBM, Compaq, Apple, Sun, Dec, etc, > that cost $2000? The NeXT is still competitive in SPEC/$, something > that not many of the PC makers can claim. What does SPEC/$ do for J. Random Suit? Judging by the NeXT software catalog, that's who they want to sell to. > A SSII goes for $14K, so > it's not in the same market as the NeXT. Never said it was. > In short, you are comparing > computers without regard to how they fit into the market. No, I'm looking at the market and I'll be damned if I can figure out where the NeXT fits in. > Something else you aren't considering, a 486 machine running DOS + > Windows is not going to be as versatile as the NeXT. True. It's not going to be as versatile as a Mac or Amiga either. But people keep buying the damn things... > Maybe Microsoft will surprise me. What? And give up show biz? -- Peter da Silva. `-_-' peter@ferranti.com +1 713 274 5180. 'U` "Have you hugged your wolf today?"
peter@ficc.ferranti.com (peter da silva) (05/07/91)
In article <1991May6.232136.17257@dsd.es.com>, rthomson@mesa.dsd.es.com (Rich Thomson) writes: > the set of hardware you have. This is because everybody codes > directly to the hardware instead of some abstract interface because > the latter is "too slow". Remove the quotes. MS-DOS' "device handling" is so abysmal that you really can't go through it, or even the BIOS, and get decent speed. For example, I wrote a terminal emulator for MS-DOS, for the 8088 PCs. Going through DOS or the BIOS, the fastest I could reliably get characters from the serial device was < 300 baud. I had to install a replacement driver, and of course my program didn't run on the AT. For screen handling, I went through the BIOS and by doing heavy curses-like heuristics I could generally keep up at 2400 baud. "cu" under IBM-Xenix 1.0 on the same machine could keep up just about to 9600, and that's running separate processes for input and output. MS-DOS and the IBM ROM Bios really *is* too slow. CP/M was better. -- Peter da Silva. `-_-' peter@ferranti.com +1 713 274 5180. 'U` "Have you hugged your wolf today?"
jhagen@talos.npri.com (Jarom Hagen) (05/07/91)
melling@cs.psu.edu (Michael D Mellinger) writes: >In article <-U3BJU9@xds13.ferranti.com> peter@ficc.ferranti.com (peter da silva) writes: >Check out the price of an SE/30 two years ago. Guess what it costs >$5000. What was that, the suggested retail price? My SE/30 cost about $3,700 2 years ago. >Which computers can you buy from IBM, Compaq, Apple, Sun, Dec, etc, >that cost $2000? Apple and IBM offer computers for less than $2,000. They don't run Unix and they aren't that fast, but they do enough fast enough for many people who would rather have the $3,000 than the latest and greatest computer. Jarom -- ------------------------------------------------------------------------------- *Not paid for and/or endorsed by National Political Resources Incorporated. 602 Cameron St, Alexandria VA 22314 (UUCP: ...uunet!uupsi!npri6!jhagen)
eachus@largo.mitre.org (Robert I. Eachus) (05/09/91)
Right now, you can get a 16 MHz Amiga 3000 for $1850, and a 25
MHz version for $2250. (True, this is a "trade-in offer" from the
manufacturer, but you can buy a machine to trade-in for less than
$200, and give it to the kids, since all you trade in is the cover of
the users manual and the serial number.) I expect that the list will
be coming down when this offer ends.
As usual, choose your own poison on MIPS numbers, but in my
experience the 25 MHz machine is faster than a Mac IIfx or a VAX
8700...
Robert I. Eachus
with STANDARD_DISCLAIMER;
use STANDARD_DISCLAIMER;
function MESSAGE (TEXT: in CLEVER_IDEAS) return BETTER_IDEAS is...
--
Robert I. Eachus
with STANDARD_DISCLAIMER;
use STANDARD_DISCLAIMER;
function MESSAGE (TEXT: in CLEVER_IDEAS) return BETTER_IDEAS is...lm@slovax.Eng.Sun.COM (Larry McVoy) (05/10/91)
umh@vax5.cit.cornell.edu writes: > A MIPS rep today told me to expect ACE machines (the compaq/microsoft etc > group) at $2K to $5K in Q2 92.Of course these will be no cache except the 8K+8K > on the R4000, and crippled in various other ways to be as cheap as possible, > but will presumably still be far above PC/Mac/Amiga performance. I don't normally get into this sort of thing and probably shouldn't now, but I find this statement a little hard to stomach. Are R4000's even taped out yet? Maybe I missed something, but I thought that the R4000 was a future chip. Let's suppose that I could get one in my grubby little paws today. I still have to build a motherboard, a case, all that glob. Call it 9 months of a very agressive schedule. I still have to have an OS that runs on the thing. Call it another 9 months on an extremely aggresive schedule. Q2 '92? I suppose it could happen. I'll be very impressed. --- Larry McVoy, Sun Microsystems (415) 336-7627 ...!sun!lm or lm@sun.com
davecb@yunexus.YorkU.CA (David Collier-Brown) (05/10/91)
umh@vax5.cit.cornell.edu writes: | A MIPS rep today told me to expect ACE machines (the compaq/microsoft etc | group) at $2K to $5K in Q2 92.Of course these will be no cache except the 8K+8K | on the R4000, and crippled in various other ways to be as cheap as possible, | but will presumably still be far above PC/Mac/Amiga performance. lm@slovax.Eng.Sun.COM (Larry McVoy) writes: | I don't normally get into this sort of thing and probably shouldn't now, | but I find this statement a little hard to stomach. Are R4000's even taped | out yet? Maybe I missed something, but I thought that the R4000 was a | future chip. So did I... Before anyone jumps on Larry for his time estimates, realize that Sun has been having to work **very hard** to get their time-to-first-system down, and that the discussion of those concerns and overheads has spread widely within sun, to the extent that people on the faaaaaaar outside get to hear about them [Howard Lee spoke about this at University of Toronto last November] Now the criticisms (:-)) | Let's suppose that I could get one in my grubby little paws today. | I still have to build a motherboard, a case, all that glob. Call it | 9 months of a very aggressive schedule. | I still have to have an OS that runs on the thing. Call it another 9 | months on an extremely aggressive schedule. Ok, lets postulate that one elects to run ahead a while before announcing: This in principle allows us to do a new case-and-peripherals design before we start on the motherboard, and plan to use a shipped OS with a very ``spare'' set of device drivers. That leaves us with a motherboard to design and rush into production. It's risky (if you make a mistake you lose credibility as well as sales, and the announcement bottlenecks your existing small-system sales), but you can get this baby out the door in nine months if you put enough mothers on it (:-)) --dave | Q2 '92? I suppose it could happen. I'll be very impressed. I would be too, and I wouldn't like to try it myself! --dave (does this have anything to do with architecture-in-the-large?) c-b -- David Collier-Brown, | davecb@Nexus.YorkU.CA | lethe!dave 72 Abitibi Ave., | Willowdale, Ontario, | Today's featured dish: CANADA. 416-223-8968 | Sun-dried alligator.
dhinds@elaine18.Stanford.EDU (David Hinds) (05/10/91)
In article <576@appserv.Eng.Sun.COM> lm@slovax.Eng.Sun.COM (Larry McVoy) writes: >umh@vax5.cit.cornell.edu writes: >> A MIPS rep today told me to expect ACE machines (the compaq/microsoft etc >> group) at $2K to $5K in Q2 92.Of course these will be no cache except the 8K+8K >> on the R4000, and crippled in various other ways to be as cheap as possible, >> but will presumably still be far above PC/Mac/Amiga performance. > >I don't normally get into this sort of thing and probably shouldn't now, >but I find this statement a little hard to stomach. Are R4000's even taped >out yet? Maybe I missed something, but I thought that the R4000 was a >future chip. Someone told me that yes, the R4000 is taped out. And they have it running on simulators, sort of. There are timing problems, like the 1MHz test mode doesn't work, but it does work at full speed. I don't know enough about the way chips are engineered to say how far from market this should be. -David Hinds dhinds@cb-iris.stanford.edu
rolande@kuling.UUCP (Roland Eriksson) (05/11/91)
Hi! There has been a debate going on about the NeXT computer, if it will be able to keep up with the competition because of the CISC nature of the Motorola 68040. I am no expert on this subject, so please do not flame me if this posting is all wrong. IMHO the main advantage that RISC arcitechtures has over CISC is that register operations are faster than memory operations and that the compiler writer do not know good ways to use the more complex ones of the instructions of a CISC processor. This is NOW. Whe have "only" about 1 miljon transistors in the most complex processors today (RISC or CISC). But we will soon have fabrication techniques to produce chips with 10 miljon transistors. Then there will be room for (on a single chip) a really complex CISC processor (with tons of microcode), a huge amount of registers (register to register operations ARE faster), a big memory cache (with some very complex associative control logic) and maybe even some interesting "extras" (math processor with vector operations, MMU, DMA, multiple CPU:s, you name it). If this processor is built AND if compiler technology keeps up, i can *NOT* see any technical reasons why this machine will not make any RISC look like some kind of primitive *AND* slow joke. With complex instructions there will be less bus bandwidth wasted on instructions. Am i mistaken? Is compiler technology on its absolute height of evolution? Please explain this matters to me, and is there any good books on this subject? No flames please, i am no expert and i will never be one if no one tells me. Thanx, Roland. rolande@yyc.docs.uu.se
ingoldsb@ctycal.UUCP (Terry Ingoldsby) (05/25/91)
In article <2050@kuling.UUCP>, rolande@kuling.UUCP (Roland Eriksson) writes: ... > IMHO the main advantage that RISC arcitechtures has over CISC > is that register operations are faster than memory operations > and that the compiler writer do not know good ways to use the > more complex ones of the instructions of a CISC processor. This > is NOW. Whe have "only" about 1 miljon transistors in the most > complex processors today (RISC or CISC). But we will soon have > fabrication techniques to produce chips with 10 miljon transistors. Debating RISC vs CISC on this basis misses one of the most important factors governing RISC/CISC performance. One of the original features of RISC technology was the speed with which a design could be created and implemented. Since the implementation technology has to be frozen at some point in the design cycle (I presume fairly early on) this means that performance of the implementation is inversely proportional to the time it took to produce it. CISC machines take longer than RISC machines, so currently available CISC chips necessarily use older technology. At present, hardware improves at a dizzying pace. This means that RISC chips have an inherent advantage. Unfortunately exponential growth never continues indefinitely. Eventually RISC will lose the implementation advantage over CISC. At that point, my guess is that architectures will end up sort of in the middle; i.e. more features than RISC, less than current CISC. -- Terry Ingoldsby ingoldsb%ctycal@cpsc.ucalgary.ca Land Information Services or The City of Calgary ...{alberta,ubc-cs,utai}!calgary!ctycal!ingoldsb
gary@proa.sv.dg.com (Gary Bridgewater) (05/29/91)
In article <658@ctycal.UUCP> ingoldsb@ctycal.UUCP (Terry Ingoldsby) writes: >... One of the original features >of RISC technology was the speed with which a design could be created >and implemented. Since the implementation technology has to be frozen >at some point in the design cycle (I presume fairly early on) this >means that performance of the implementation is inversely proportional >to the time it took to produce it. CISC machines take longer than >RISC machines, so currently available CISC chips necessarily use older >technology. CISC system designers work with process designers. Process designers generally have some idea where they are going next. Aggressivly designed CISC (or RISC, for that matter) will be designed more-or-less in parallel with the target process. Pure in-house implementations can be even more aggressive in that the two can have the end product as their common goal. So there is no technical reason that a CISC chip released in Quarter X should be any less "State of the Art" than a RISC chip released in Quarter X with respect to technology (modulo the FABs). If the CISC implementation cycle is longer than the one for RISC then that means the CISC designers have to peer a bit deeper into their process designers' minds but they can also adjust to changes as they go. And, if you watch closely, you will sometimes see either flavor having a rev 1.0 release at X Mhz followed, some months later, by an X++ MHz version 1.1 reflecting process and implementation fine tuning (some do this the other way around - announcing the X++ version, shipping "early" versions at X and then shipping the X++ version later). -- Gary Bridgewater, Data General Corporation, Sunnyvale California gary@sv.dg.com or {amdahl,aeras,amdcad}!dgcad!gary "I am a pizza. I am a pizza. ..."