rbc@houxu.UUCP (R.CONNAGHAN) (05/31/85)
Take a look at "Electronic Products June 3 page 23 for a product anouncement for thier 32 bit chip set. Quotes prices and availiblity. Mentioned are the: WE 32100 CPU (about $400) WE 32101 MMU (about $400) WE 32106 MAU (1 Mwhet) Other chips (DMA, DRAM Controller) are mentioned. Also mentione are the WE 321EB Evaluation Board (Benchmark Board) (About $5000) WE 321AP Analysis Board (In circuit Emulation) (? ) I am one of the designers of the WE321EB Evaluation Board. It does my heart good to see the product go commercial. Any questions or comments please respond to me directly. I will answer any questions I can. -- Robert Connaghan WE 32100 Microprocessor Group AT&T Information System - Holmdel, N.J. houxu!rbc
karn@petrus.UUCP (06/06/85)
> Mentioned are the: > WE 32100 CPU (about $400) > WE 32101 MMU (about $400) > WE 32106 MAU (1 Mwhet) At the last ham radio fleamarket ("hamfest") I was at, I could buy new 68000's for about $1.50 per megahertz (i.e, a 10 MHz part for $15)... Phil
res@ihuxn.UUCP (Rich Strebendt @ AT&T Information Systems - Indian Hill West; formerly) (06/07/85)
> > Mentioned are the: > > WE 32100 CPU (about $400) > > WE 32101 MMU (about $400) > > WE 32106 MAU (1 Mwhet) > > At the last ham radio fleamarket ("hamfest") I was at, I could buy new 68000's > for about $1.50 per megahertz (i.e, a 10 MHz part for $15)... And you could probably have bought brand new Z80's for about two bits a megahertz (4MHz for a dollar). PLEASE do not compare bananas and strawberries (I almost said apples and ... but lots of articles are posted comparing Apples to everything else :-). The WE32100 Microprocessor line is a line of full 32-bit (32-bit addressing and 32-bit data paths and registers) machines with a full complement of support chips. Do not confuse these chips with "32 bit" chips that need extra registers to address more than 64K !!! Rich Strebendt ...!ihnp4!iwsl6!res
karn@petrus.UUCP (06/08/85)
> The WE32100 Microprocessor line is a line of full 32-bit > (32-bit addressing and 32-bit data paths and registers) machines with > a full complement of support chips. Do not confuse these chips with > "32 bit" chips that need extra registers to address more than 64K !!! I really don't care much if a chip is "32-bit" or not. Touting the number of "bits" on a microprocessor is a bit reminiscent of the practice of advertising the number of tubes in old-time radios. If it doesn't improve the price/performance ratio of the product, it becomes a red herring. Anyway, I always look first at non-performance related issues like: 1. A large, flat address space 2. Support for virtual memory 3. Availability of reasonable system software and program development tools (e.g., 4.2BSD) 4. Second sources on the chips and support tools The 68K either equals or surpasses the BigMac32 (oops, WE32000) on each of these points. Only THEN do I look at how fast the thing runs C benchmarks (something that can be affected by the word size). Unless your $400 32-bit WE32000 can run C at least 26.67 times as fast as my $15 68K (which I seriously doubt), you get more bang for your buck with the 68K. Phil
gwyn@brl-tgr.ARPA (Doug Gwyn <gwyn>) (06/09/85)
Come on, guys, this arguing accomplishes nothing useful. Neither the $15 chip nor the $400 chip can do doodly-squat without a substantial investment in additional electronics. A better topic for discussion would be: How capable is a minimum system configuration built around the chip set, how capable is a maximal system using the chip set, and what are the probable manufacturing costs of such systems?
george@mnetor.UUCP (George Hart) (06/10/85)
> > > Mentioned are the: > > > WE 32100 CPU (about $400) > > > WE 32101 MMU (about $400) > > > WE 32106 MAU (1 Mwhet) > > > > At the last ham radio fleamarket ("hamfest") I was at, I could buy new 68000's > > for about $1.50 per megahertz (i.e, a 10 MHz part for $15)... > > And you could probably have bought brand new Z80's for about two bits a > megahertz (4MHz for a dollar). > > PLEASE do not compare bananas and strawberries (I almost said apples > and ... but lots of articles are posted comparing Apples to everything > else :-). The WE32100 Microprocessor line is a line of full 32-bit > (32-bit addressing and 32-bit data paths and registers) machines with > a full complement of support chips. Do not confuse these chips with > "32 bit" chips that need extra registers to address more than 64K !!! > > Rich Strebendt > ...!ihnp4!iwsl6!res I trust you're not talking about the 68xxx family? If you are, what are the extra registers you're referring to? -- Regards, George Hart, Computer X Canada Ltd. {cbosgd, decvax, harpo, ihnp4}!utcs!mnetor!george
mike@peregrine.UUCP (Mike Wexler) (06/11/85)
> > The WE32100 Microprocessor line is a line of full 32-bit > > (32-bit addressing and 32-bit data paths and registers) machines with > > a full complement of support chips. Do not confuse these chips with > > "32 bit" chips that need extra registers to address more than 64K !!! > > I really don't care much if a chip is "32-bit" or not. Touting the number of > "bits" on a microprocessor is a bit reminiscent of the practice of > advertising the number of tubes in old-time radios. If it doesn't improve > the price/performance ratio of the product, it becomes a red herring. Anyway, > I always look first at non-performance related issues like: > > 1. A large, flat address space > 2. Support for virtual memory > 3. Availability of reasonable system software and program development tools > (e.g., 4.2BSD) > 4. Second sources on the chips and support tools > > The 68K either equals or surpasses the BigMac32 (oops, WE32000) on each of > these points. > > Only THEN do I look at how fast the thing runs C benchmarks (something that > can be affected by the word size). Unless your $400 32-bit WE32000 can run C > at least 26.67 times as fast as my $15 68K (which I seriously doubt), you > get more bang for your buck with the 68K. > > Phil *** MESSAGE THIS LINE WITH YOUR REPLACE *** This comparison is not fair for two reasons. First of all it doesn't consider the "glue" chips it takes to make either of the processors work. Second, the 32100 is brand new. How much did the first 68000 cost? I agree with you that 4.2BSD is nicer for development than System V, but System V is catching up. It now has demand paging, sdb(a symbolic debugger and predecessor of dbx) and many others new features. -- -------------------------------------------------------------------------------- Mike Wexler(trwrb!pertec!peregrine!mike) | Send all flames to: 15530 Rockfield, Building C | trwrb!pertec!peregrine!nobody Irvine, Ca 92718 | They will then be given the (714)855-3923 | consideration they are due.
jona@clyde.UUCP (Jon Allingham) (06/12/85)
> ... > Unless your $400 32-bit WE32000 can run C > at least 26.67 times as fast as my $15 68K (which I seriously doubt), you > get more bang for your buck with the 68K. > > Phil It was shown a long time ago that performance ( ie speed here ) vs cost is NOT linear. I for one would rather have the increased speed. -- Jon M. Allingham (201)386-3466 AT&T Bell Laboratories-WH "Beam me up Scotty, no intelligent life down here!"
guy@sun.uucp (Guy Harris) (06/13/85)
> > > The WE32100 Microprocessor line is a line of full 32-bit > > > (32-bit addressing and 32-bit data paths and registers) machines with > > > a full complement of support chips. Do not confuse these chips with > > > "32 bit" chips that need extra registers to address more than 64K !!! > > The 68K... > This comparison is not fair for two reasons. The original comparison was inaccurate for one reason; the 68K doesn't require extra registers to address more than 64K. > I agree with you that 4.2BSD is nicer for development than System V, but > System V is catching up. It now has demand paging, After 4.xBSD has had it for a couple of years and after several groups have independently implemented it in Systems III and V... > sdb(a symbolic debugger and predecessor of dbx) "sdb" came out with 4.1BSD and System III. The 4.1BSD version is a later release than the System III version, but both were ferociously bug-ridden. It's an ugly implementation of an ugly debugger. "dbx" owes nothing to "sdb" and actually has a syntax that isn't rife with ambiguities. > and many others new features. But *still* no sane networking support! Hint to AT&T: if your releases had more of the facilities needed by vendors and users and didn't wait for a year or two until everybody else already did their own implementations, there would be a lot less variation in UNIX systems out there... (providing that the AT&T implementations aren't as silly as shell layers, for instance). I'd like to see that happen, but I don't see it happening for at least another year or so; meanwhile, the best approach would probably be a System V Interface Definition-compatible system with a full set of 4.2BSD features (not impossible - the S5ID doesn't say what file systems or directories look like). Then again, it's irrelevant to a <your_favorite_microprocessor> vs. WE31x00 debate - the WE chips will *not* melt if you build a machine out of them and run 4.2BSD on it.... Guy Harris
keithd@cadovax.UUCP (Keith Doyle) (06/18/85)
> PLEASE do not compare bananas and strawberries (I almost said apples > and ... but lots of articles are posted comparing Apples to everything > else :-). The WE32100 Microprocessor line is a line of full 32-bit > (32-bit addressing and 32-bit data paths and registers) machines with > a full complement of support chips. Do not confuse these chips with > "32 bit" chips that need extra registers to address more than 64K !!! > > Rich Strebendt > ...!ihnp4!iwsl6!res How did you figure that out? I've been trying to get information (through previously mentioned channels, the 800 number etc.) on this chip, and so far have only received a skimpy little brochure that dosen't say squat about architecture. Has ANYONE gotten any information about some specifics? Such as: number of registers, addressing modes, etc. and all that kind of good stuff? Maybe its a secret huh? (maybe it's still 'vaporware' huh?) Keith Doyle # {ucbvax,ihnp4,decvax}!trwrb!cadovax!keithd
keithd@cadovax.UUCP (Keith Doyle) (06/19/85)
[.......]
WOOPS! Just after I wrote a previous article complaining about lack
of information on the new WE 32100 lo and behold, I receive just the
specs I was lacking in the mail. All kinds of neat stuff. Pinouts,
registers, addressing modes, etc. including specs on a MMU. Have to
peruse it more carefully yet. First impression-- adressing modes
kind of limited (read rudimentary) lots of displacement nnn(rn)
modes, but seems like little else (I miss auto increment/decrement for
multiple stacks). None of the exotic addressing modes of the NS32000 or
68020. Maybe they're taking a somewhat RISC approach? (maybe you can't
make use of that stuff from UNIX anyway?)
Warrants further research. Anyone else seen these specs yet? What do
you think? I would guess it was designed with UNIX in mind, but not being
a kernel wiz, I have no idea.
Keith Doyle
# {ucbvax,ihnp4,decvax}!trwrb!cadovax!keithd