DTACK GROUNDED, The Journal of Simple 68000/16081 Systems
Issue #35 - September 1984 - Copyright Digital Acoustics, Inc
Our POTBOILER medium resolution graphics board production prototype is pictured below. One of five production prototypes, actually - we had a lot of confidence that the first version would work with the usual cuts and jumpers and it did. We modified the artwork to fix the cuts and jumpers and placed an order for the first fifty production boards on 22 Aug. With the usual lead time for circuit boards and our outside wave soldering, we should begin shipments around the middle of Oct.
The POTBOILER pictured below is a version configured for a certain CADD house. Their case requires a video connector at the upper left as shown. That connector configuration will not work with our no-longer-mythical case, so the DTACK hacker version is identical except for a 3/4 inch by 1-1/2 inch 'ear' on the left-hand side of the board, providing a video connector that IS compatible with our new case (and power supply).
Aside from the video connector, the CADD and DTACK versions are identical and come with identical prices - $550 - and identical documentation and software support - none! The POTBOILER was designed to be the most economical (all right, cheapest!) medium resolution graphics board possible while still using a 7220 graphics controller chip. We believe we have achieved our objective - see any other 7220-based RGB graphics boards for $550?
To get the lowest possible price we have unbundled software from the POTBOILER graphics board, which comes with just a brief memory map and a warranty. Software support will cost the usual $10 per disk; what we have on hand now (as of 25 Aug): is just one disk containing a memory test and a very few demo programs written (mostly) in assembly language, provided with sample source code.
Actually, we have a fair amount of software for such a new board and we do plan to provide high-level HALGOL support soon, like in the next two months. But if you need specific software support the safest thing to do would be to wait until such support is available - that way we (and you) avoid problems like PDQ's missing intelligent language laboratory or the Saybrook UNIX which will arrive in the spring any year now.
You may also notice that the POTBOILER pictured below has lots of empty real estate. We may build a version which populates some of that real estate. If so, the price will obviously be more than $550.
'POTBOILER' WORKING PRODUCTION PROTOTYPE BOARD
Oh, yes: you will also need a color graphics monitor AND connecting cable. Tandy will sell you an appropriate matching monitor and cable for $800. That cable is a specially-made one, so treat it nicely and take it back to Tandy if it gets sick. Digital Acoustics does not manufacture or repair specially-made cables either for free or for profit.
We have just explained to you all the 'gotchas' we can think of with respect to our forthcoming medium resolution graphics board. All the reasons you shouldn't buy one. Here is something else to think about:
Everybody who has seen that color monitor, whether driven from the Tandy 2000 or our prototype boards, LOVES it! A certain CADD house thinks they can sell a low-end CADD system built around it better than their existing 1024 X 796 (monochrome OR color) system. Your FNE is going to have one of those boards and a monitor at his personal Apple station for a very good reason - he is the boss! Our full-time HALGOL programmer is lobbying hard for the same equipment at HIS Apple station (so that he can support it with HALGOL, of course). Our project engineer is NOT lobbying for the equipment at HIS Apple station because he already has it! What we are trying to say is, we think that is going to be a very popular color graphics system for those who can afford the price of admission.
Again, both HALGOL implementers REALLY like that monitor. Draw your own conclusions. And now we think we have given you a picture of this new product which is accurate and balanced right now. (If not, give us credit for trying?)
As we reported on page 24 of the last issue, we recently discovered that NS16081D-6 chips made in 1984 did not work when placed in a socket formerly occupied with a NS16081D-6 made in 1983, even though the chip made in 1983 DID work in that socket. In fact, we and our customers went through about 20 of the 1983 chips and essentially all of them worked. We tried 6 of the 1984 chips and the University of Illinois (Urbana) tried 5 and NONE of them worked. And we couldn't get any help when we went to the factory.
In Feb this year there was a large turnover of Nat Semi technical personnel working with the 16081 and other advanced microprocessor stuff. We have recently learned from a reliable source that the turnover was even greater than we knew; almost everyone formerly associated with the project has left Nat Semi. The turnover was not limited to the line troops but extended up to and included the manager of the Nat Semi advanced microprocessor group (the new manager's name
is Col Rada). We have not been able to ascertain whether the departures were voluntary or involuntary or a combination thereof nor do we know the underlying reason for the massive turnover. Again, all or most of this turnover occurred around February.
This turnover was undoubtedly the reason we couldn't get any help from the Nat Semi folk. The new-hires simply didn't know much about the math chip. It was obvious from the evidence at hand that some change had been made to the chip design but nobody (at least nobody we could reach) knew WHAT change. Finally, we were invited us to send one of the new chips back to the factory for evaluation, which we did.
In the middle of August, two months later, we finally heard from the Nat Semi folks over that evaluation. It seems they discovered that the new ('84) chip worked fine - AND SO DID OLD ('83) CHIPS! IN THE SAME TEST SETUP! This result contradicted OUR experience, of course. So our project engineer (NOT Eloi) took a hard look at the software we were driving the chip with and discovered THAT was the problem! With a simple software change (involving not addressing a particular address) both '83 and '84 chips work very nicely in the experimental math board.
In fact, ALL FOUR 10MHz SAMPLES WE HAVE (on loan from Rockwell) WORK JUST FINE IN OUR MATH BOARD WHEN PLUGGED INTO A 10MHz STATIC DTACK BOARD! Was that FIRST 10MHz chip we maligned a good one after all? We dunno - it's gone. It did run AWFULLY cool.
Those of you who are tired of waiting for the Nat Semi 68000/16081 application note, which is intended to show how to couple a 10MHz 68000 to a 10MHz 16081, can check out the schematic on the back page of this newsletter. That's the schematic of (less the clock divider for 6MHz math chips) our experimental math board and like we just said, it works!
The software change is very simple - see p.24 of our issue #26, for instance. If you change line 24, address $2616 from $0EE0 to $0EC0, the code will run! What that is doing is changing the FOP (Floating point OPerand) from address select $E0 to $C0 - a very simple change. (We do not know whether the unannounced mask changes which trashed our old software have eliminated the very minor two-item bug list.)
We DO know that Nat Semi now claims that the 16081 math chip is in production. However, the distributor price differential between quantity 1 (still $200 at 6MHz) and quantity 100 is not the traditional 1.5 to 1 for true production parts. In fact, the quantity 'discount' is very small. [On the other hand, we must keep in mind that Nat Semi spokesperson R.M. tried to tell us in Feb this year that the 16081 math chip had
been in production for a year! Official Nat Semi pronouncements do not always accord with the facts...]
[Aug 29: the 16081 pricing structure is in a state of flux. 10MHz 16081s are available ($300) but 10MHz 16032s ain't - while 8MHz ARE.]
P-system based Sage Computer is not really in direct competition with DTACK but prexy Rod Coleman, in his guise as a sometimes-author of a newsletter, IS. SAGE NEWS, Aug '84 has a lot of very practical comments about the sad-sack low-tech IBM PC and some very accurate predictions of when customers will be able to get their hands on new stuff like 256K chips and 68020s. In fact, you would almost think that Rod and SAGE had built a machine or two and had learned some real-world stuff via the usual school of hard knocks.
We read his Aug newsletter and cruised right along (ignoring the double misspelling "Nicklas Writh" for Niklaus Wirth) until we came to the end of page 2 where Rod asserts that multi-user systems have a bright future with - get this - CONSUMERS! Not a bright future with business honchos who want to chain a bunch of peons to a cost-effective multi-terminal computer mind you, but CONSUMERS!
We indignantly reached for our handy Webster's Ninth New Collegiate Dictionary to use as a reference to devastate Rod's ridiculous assertion and discovered that the dictionary definition of 'consumer' isn't much help! Put it this way - we think it is ridiculous to assert that consumers are going to cotton to multi-user systems but we can't prove it. Hence, in true demagogue fashion, the following red herring:
Coleman and SAGE are guilty of the worst kind of nepotism! On page three there is a software review by SHERI Coleman! Now, either Sheri is improperly using her family affiliation with prexy Rod to unfairly bypass others on the promotion ladder and attain a high-paying executive position OR Rod and SAGE are unfairly using low-paid labor in violation of Social Security and other federal laws to gain an advantage over other newsletters such as this one. (Did you notice we got 'em coming and going?) Not only that, but Sheri has not yet filed a complete financial disclosure for herself and her family for the past 99 years. Did Rod pay corporate income taxes for his lemonaide stand at age 6 or newspaper route at age 12? The FBI should look into this...
(Just kidding, Rod. But: "Nicklas Writh"? And surely you do not really believe that multi-user systems have a "bright future" with consumers as opposed to businesspersons?)
The trade journals (ED, EDN, EW etc.) have been expending lots of ink on workstations lately. When you look for hard information you wind up trying to grab smoke. Without producing a chart of manufacturers and model numbers and such, we will discuss the general characteristics of these workstations:
All of them use a 32-bit data bus for the simple reason that all of them have to be (or SHOULD be) good number-crunchers. Some of the numbers needing to be crunched may be integers. This leads us inescapably to the fact that they are all single-user (no sharing CPU cycles here!) but multitasking. Multitasking because they will be used 8 hours a day (at least) and nobody wants to stop one task while another is performed.
An organization which employs 17 engineers who work with these workstations will typically own two or three of the workstations. With typical prices running $137,000 you can see WHY only two or three.
These workstations are sold with software 'bundled'. The price commonly represents more software cost than hardware cost. That means we have a vicious circle with regard to pricing: because the workstations are expensive, companies buy few and share among many engineers. Because few workstations are bought, the software cannot be amortized over a larger number and the prices stay high. (There is absolutely no way Digital Acoustics could purchase a $131,000 workstation short of selling its soul to a trusting bank.)
The workstations are almost all used to design custom LSI. Gate arrays, standard cells, stuff of that ilk. Although this is a very important application it is a very small market. However, it IS a market which can bear the current very high prices. As long as workstation manufacturers continue to target the custom LSI design market, workstations will remain very expensive and the market will remain small.
Most engineers do not design custom LSI. Almost all engineers work with drafting boards or (ugh!) printed circuit board artwork stations ('light tables'). Your FNE has more personal experience with drafting boards and light tables than he cares to remember. The workstation managers have not yet targeted this market, although it is at least a couple of orders of magnitude larger. Why?
Well, the automated circuit board layout market has been around a while and is pretty much minicomputer-based. Although the number-crunching and video graphics requirements are similar, automated circuit board layout systems also need a very good digital plotter - and these have been VERY expensive up to now.
Another factor is that it costs lots of money to break into a market where some companies such as Gerber Scientific have long held a large market share. Automated circuit board layout is electronics' version of a smokestack industry.
The automated circuit board layout market is one which almost anybody can see can be revitalized using newly arrived technology. But upon examining the time, investment and risk, business planners are very willing to let somebody else tackle the job.
This brings us to automated drafting, nowadays called 'schematic capture' in the trade. A good 16-bit machine has been proven to be adequate for this market, so the higher priced - and more capable - 32-bit workstations are not needed. Identical machines are used in architectural, civil engineering and structural engineering applications although the word is no longer 'schematic'.
All of the things we have been discussing are aspects of the CAD/CAM industry, which stands for Computer Aided Design/Computer Aided Manufacturing. The first thing you should know is that there is no such thing as computer aided design, just Computer Aided Analysis (the distinction is quite critical). The second thing you should know is that although CAA/CAM systems do exist, they are VERY expensive. Make that VERY VERY expensive and substantially customized, so we won't pay any more attention to them. The third thing you should know is that most outfits which claim to be in the CAD/CAM business are instead in the CADD business (Computer Aided Drafting and Design - and once again, you can throw out the 'Design' part).
A good CADD station needs lots of disk storage, high resolution graphics and a decent 16-bit CPU. Access to a plotter is also needed, although one plotter can service several CADD workstations. All of these things have been available for a long time in the minicomputer world but minicomputer prices have restricted the size of the market. All of these things are now a LOT less expensive than they used to be - see the front page of our rag, issue #29, for instance. All of you know how inexpensive 10 or 40 megabytes of Winchester disk is these days, but most of you don't know about the recent astonishing advances in plotter technology.
Plotters which until recently cost $75,000 have been replaced with equivalent performing $8,500 models! That's for E size drawings, known in the industry as 'bedsheet' drawings for their size. We own a $3,000 Houston Instruments TP-41 plotter for D-size drawings. The TP-41 is a very good, if not super-fast, plotter. We haven't turned it on yet and it is ALREADY obsolescent! Why? The usual - microprocessors, competition and a market that expands as prices drop.
Folks who are used to graphics being plotted at 14.7 nanoseconds per pixel are sometimes astonished at how slow even a good plotter seems to be. For 12 years now (that's how long these plotters have been commonly available) folks have run the plotter overnight if need be or even over the weekend.
An increasing number of workstation vendors have finally figured out (correctly) that providing a mix of text and graphics on a single monitor is too expensive, and are providing a separate monitor as the primary text output device for menus, etc. If one tries to combine text and graphics onto a single monitor one either pays the price in terms of a slow system, as in Mack's Mack Write program or the old LlSA-Write, or you can visit us and try out our Tandy 2000 (sloooow text with the color graphics board installed). Plotting text on a graphics screen like Mack's or the Tandy 2000's is like hitting a racehorse between the eyes with a 5-lb sledge hammer; things essentially come to a halt! (Our Tandy 2000 does NOT have the graphics board installed right now so that Multi-Mate will run real fast when we get around to running it.)
The next alternative is to spend a great deal of money on special circuitry to circumvent the slow text problem. So to combine text and graphics on the same monitor you either pay in time or you pay hard cash, Both ways, it gets expensive. On the other hand, a separate text monitor costs a few inexpensive chips and a $79.95 monitor you can buy locally.
We have briefly reviewed the 32-bit workstation market which targets custom LSI design, the minicomputer-based smokestack-industry automated circuit board layout market and the 16-bit CADD market. These three markets have many common elements:
By this time the alert reader should be asking, "so what?". So the 68020, which now exists and which will be sampled in about four months and will be in production in a year, provides enough computational capacity for any of those applications. Hardware graphics chips such as the first-generation 7220 and its successors provide enough graphics power for any of these applications. Hard disks are very cheap and getting cheaper and plotters are fairly cheap and getting cheaper.
By this time next year it will be possible to produce a cheap, universal graphics station which will be able to serve ANY of the three markets we have discussed! Add the right software package and AWAY YOU GO!
(Yes, a universal design, and hence cheap through economies of scale, is possible. No, it's not gonna happen. But you knew that, didn't you?)
However, the requirements of a good CADD workstation just happen to be pretty close to our idea of an ideal personal computer, with these little changes...
is one we are going to pay for out of our own pocket, not the company's pocket. That means that we are not a full-time computer professional right there. And since we are not going to be using the computer full-time, we can do long printouts, assemblies etc. when we are doing something else. So we can save money by eliminating the memory management required by multi-tasking. We can also use less memory if necessary (since this is our personal computer, we will load it up with a megabyte soon enough). And the hard disk drive can be a lot smaller since it does not have to maintain separate files for ten different employees who may use that workstation. In fact, with the new high-density floppies we may not need a hard disk.
But most of those workstations have 1024 X 796 (or so) graphics resolution and we can't afford that using our own money at today's prices. But we CAN afford medium-resolution graphics like our POTBOILER and (groan!) we CAN (if barely) afford an $800 color monitor to use 640 X 400 non-interlaced graphics.
Of course, since the text will be handled by a different monitor and by different circuitry it is a simple matter to leave the graphics monitor and controller board off if one is just looking for a software development station, for instance.
With no memory management and no multi-tasking operating system but with a language/programming environment like HALGOL, which directly grabs all of the system resources for the single user and his/her
single task our ideal personal computer is going to be significantly faster than an equivalent (same CPU) workstation! Yes, you read that correctly: our single-user single-tasking personal computer, which is CHEAPER on account of no memory management or multi-tasking overhead or REQUESTS for system resources is FASTER than those expensive multi-tasking systems and their high-overhead operating systems and their meek requests for system resources!
The assembly-language freaks among our readers will realize how much faster a program will run if it can take advantage of the 68000's zero page and direct addressing and direct access to system resources as opposed to the relocatable programs required to run under multi-tasking operating systems.
A jihad is a holy war, and our comments above may have tempted you to start one. Please do not write letters accusing us of opposing multi-tasking systems because we are not so opposed - IN COMPUTERS DESIGNED FOR FULL-TIME USE! However, we favor PERSONAL computers; PERSONAL computers are NOT used full-time (8 hrs/day); and so personal computers do NOT need multi-tasking and all the bullshit, er, toro poo-poo that goes along with multi-tasking. (See also Aug Dr. Dobbs, p24, especially the penultimate paragraph.)
We say again, the needs of workstations which are used full-time are quite similar in many ways to our idea of a good personal computer but quite different in other ways. We are NOT hostile to business computers (or operators) which are used full-time any more than we are hostile to those necessary multi-user airline or hotel reservation systems. We merely want our own personal computer to be as FAST and INEXPENSIVE as possible - and that gives us two reasons to dump memory management, multi-tasking operating systems, relocatable code and meek requests for system resources.
Look: we have a very nice, fast 68000 board with up to a megabyte of RAM which is definitely the cheapest big, fast 68000 board on the market. We have been selling 16081 math boards to work with that 68000 board since Dec '83. We are introducing a good, inexpensive color graphics controller REAL soon. It does not take a genius to see that we are only a keyboard and disk controller (and a couple of parallel and serial ports for printers, plotters and modems etc) away from a complete stand-alone 68000 system. So Digital Acoustics will soon be entering the personal computer market with its own computer?
LIKE HELL IT WILL! There's not a single death-wish anywhere in the company, as far as we know. Nor are we interested in turning control of the company over to some bankers and/or investment brokers. Still... it's sometimes fun to speculate about what maybe CAN be done using existing resources. Speculation follows:
There is nothing that says we can' t bolt a keyboard etc. onto a set (or five) of DTACK boards inside the company, the better and faster to develop HALGOL with. HALGOL is doing quite well, thank you, but it is outgrowing the Apple II as a development system. It is SERIOUSLY outgrowing the Apple II development system.
And since any system or application software we develop for that custom system is not commercial, we will be in a position to GIVE it to any other DTACK customer who might want to graft a keyboard and disk controller onto his/her Apple compatible DTACK boards. [Have we ever complained about how poorly the English language is suited to unsexist prose? That his/her stuff is damnedly inconvenient. Oh, yes: we DID complain about a year ago, didn't we? Please excuse the digression.] So anybody who buys the necessary DTACK I/O board can also have an experimental HALGOL development station. And we really won't be selling a computer...
A funny thing happened when we added multiple statements per line. The line numbers, which had been optional and used exclusively for editing purposes, became a necessity. In the process, one of the three links associated with each HALGOL action address became redundant, sort of like a vermiform appendix.
A very brief review for newchums: HALGOL (like BASIC) is interactive, but is invisibly and incrementally compiled so that when the program runs, it IS compiled and hence very fast. Since there is no such thing as a free lunch, you may ask how the miracle of an interactive compiled language is achieved. We answer: by using enough memory to keep both the source and compiled code in memory at the same time! Finally, we bend familiar BASIC enough so that the source and compiled code for each individual statement are kept together instead of having a huge source module and a separate compiled object module.
HALGOL run-time code consists of threaded action addresses with the necessary parms (parameters). If the phrase "threaded action addresses" made you jump to your feet and shout "FORTH!", sit back down. HALGOL is in no other way similar to FORTH. Specifically, HALGOL uses no reverse Polish notation, it does not use a stack for parameter-passing, and HALGOL is written in assembly language, not HALGOL (everybody knows that FORTH is written in FORTH).
Up until now, each action address had associated with it three address links: an "edit" (really "compile") link, a "list" link, and a "line" link. The edit link handles the compilation at the time a program statement is entered, the list link handles listing the compiled code back out AS IT WAS ORIGINALLY ENTERED, and the line link pointed to code which determined the start of the next HALGOL statement.
It is the line link that just died. By going to multiple statements per line number it became necessary to compile and list HALGOL code in chunks whose size is determined by how many program statements are tied to a particular line number instead of a single program statement at a time. So the HALGOL program now has a table of line numbers and an associated table of line number addresses. That table of line number addresses has obsoleted the line links.
And line numbers have become institutionalized in HALGOL as a permanent part of the language. The only way to get rid of them is to eliminate the multiple statements/line feature, and we don't want to do that.
We spent a morning recently performing a search-and-destroy to eliminate those useless line links. Our resident full-time HALGOL programmer asked (rather cattily, we thought) just exactly how many bytes we were saving by eliminating a few dozen links. We replied that we were not interested in saving bytes at all! We had in mind the confusion which a few dozen conspicuous but unused and useless links would cause anybody who looked at the source code.
Oh, yes: we will make the source code available for the usual $10 per disk with HALGOL release 3, coming up fairly soon.
(continued from p.19)
"Maybe you should be trying the people who have to prototype systems, by trying such places as EDN rather than places for users, like MICRO or BYTE." Eric L., Faulconbridge Australia
We're glad to see that you and Ken are putting our rag's reputation to effective use. You have arrived at OUR opinion regarding our 'market' but that's not much help in planning an advertising campaign. Places like EDN are death - you do indeed get lots of 'responses' but when you audit a couple dozen you discover at least half never checked the reader response code (the computer does that for you to try and kid you that ads in the magazine provide fantastic response). The other half want a complete set of literature for their file cabinets. The prototypers want a schematic and timing diagrams, period. BUY a board? It is to laugh - FNE
In the last newsletter we told you a few tales of sales that aren't being made and of bills that aren't being paid (a poet and we don't noet!). Here are a few excerpts from the very next Electronics News (30 Jul) after we mailed the last newsletter:
(headline) "Say Mackintosh Momentum Slows; Corporate Accounts Cool" (p.1). "Convergent Drops Workslate; Takes $15M Set-Aside" (p.23). "IBM-PC-compatible manufacturer Columbia Data Products said last week its cash position has been adversely affected by what it called the stretchout of payments for its products by distributers. The acknowledgement came as it issued revised preliminary results for the second quarter, which saw it cut $10M from sales after consultation with its independent accountants... (Columbia's preliminary second-quarter results) show a loss of $2.5M on sales of $14.8M.)" (p.24) [Darn those pesky auditors anyhow, right?]
"Televideo Systems Inc. last week said lower-than-expected sales of its IBM PC-compatible computers will cause it to report lower-than-expected earnings for the third quarter ending this week." (p.24) "Founder/Chairman Resigns at Eagle... sources say the company's backers have approached offshore investors to determine if there would be any interest in acquiring Eagle." (p.24)
"Keyboard Mfrs. Asking Cash in Advance... Keyboard manufacturers are demanding cash in advance from certain personal computer firms and are taking other precautions should more of their customers follow Franklin, Victor Technologies, Computer Devices and Osborne into Chapter 11 courtrooms... Key Tronic Cuts Paychecks 10%" (p.42)
"Industrials Tighten Credit As PC Shakeout Looms... Anxious to keep their receivables in check and offset the prospect of sizable writeoffs as a shakeout in the personal computer industry looms, industrial distributers are further tightening their credit policies." (p.70)
We had lunch again (4 Aug) with our friend who is well-connected in a micro peripherals industry. A month ago his company had just laid off half their employees while some other companies in his industry had laid off half and half again, or 3/4, of their employees. Well, his company has just laid off its second half! And some of the well-known names in the industry, names you would recognize, are among the latest to catch the accounts-payable disease and aren't paying their bills anymore. It appears that his industry is approaching the infamous grid-lock with respect to payables... Are you beginning to get the picture?
(We kidded him that since he works daily with much larger companies than ours, it was surprising that he would associate himself in any way with a dinky little outfit like ours. "It's nice to visit somebody who is making a profit!" he rejoined.)
A certain company is developing a product using the 286, and has paid $300 each for some supposedly good parts. But they discovered there are bugs in the interrupt handling hardware. Apparently the bug is related to the operation of the ready line during an interrupt acknowledge cycle. When they informed Intel, the response was "Oh, yes. We already knew about that!" We understand this bug will be fixed sometime during 1985 (!).
Intel really stands behind its parts! Waaaayyy behind...
[LATER: "The interrupt disable mask is ignored when the POPF instruction is executed with an odd stack pointer, unless two wait states are inserted in the fetch operation. The problem can be fixed by inserting two wait states into all data read operations [YE GODS! - FNE] or all even data byte read operations. Generally odd stack operations are not used by programmers because of the great speed penalty on an 8086/186/286. However, there is no additional speed penalty on the 8088 because it is always slow. Running a version of the less popular operating system revealed this bug." - Understandably Anonymous)
U.A., are you asserting that there might be some 8086/186/286 software out there written for the IBM PC, which uses the 8088? We didn't know that! - FNE
We are sure you have seen by now the ongoing rumor that IBM has purchased 1.5 million of the CD-4 digital disks that are being used for home phonograph replacement. That may well be true, but please understand that 1) the rumor covers disks, NOT DISK DRIVES, and 2) the disks themselves are absolutely useless until a bit pattern has been stamped into them. In other words, IBM may well have a stock of 1.5 million blank digital disks.
Since each blank disk is capable of holding nearly 400 megabytes, that's, let's see now - 1.5 million times 400 megabytes... damn! Our calculator just blew up! Let's just say, thatsa lotsa megabytes!
NOT INCIDENTALLY: We now have a CD-4 player at home, and we love it! Ours is a Magnavox/Philips and set us back $500. Fantastic sound!
All of you know that Victor is in bankruptcy. What you may not know is that Victor continues to operate successfully under Chapter 11. They are actually generating a positive cash flow by turning parts inventory into computers and they are selling all the computers they can make, mostly overseas. Since Victor management can operate very nicely with Chapter 11 protection from its creditors, Victor management wants to just go along as things are!
The creditors (read: Walter Kidde) want their money back - at least, as much as they can get. That means they want to sell the assets of Victor, which are larger than you might think. Victor's British agents, ACT Ltd, want the company and have offered about $13M. A Swedish outfit (Datatronic) topped that bid and then Beta Systems of Mannheim, N. Germany bid $30M. The Victor creditors said fine, hand over the $30M. Well, uh, that is... the $30M did not arrive on schedule. No dinero, no sale. (Beta says the problem is, Kidde changed the credit terms so that the final cash price became indeterminate.)
The creditors still want top dollar for Victor's assets and since Victor management is operating so successfully there is no hurry to sell, which pleases Victor management. Some folk think the Norwegian bid is suspect, too, which could leave ACT as the lone bidder with real money (we think). And that is where the situation rests.
OH, YES: THE ULCER! With a loused-up situation like the above, you just KNOW somebody is getting an ulcer. In this case it seems to be ACT, which seems to really want and maybe need Victor, and in a timely fashion.
We have been a subscriber to Electronics, the McGraw-Hill magazine which was published every two weeks, for the past 23 years. That magazine is no more, having been replaced by Electronics Week, which publishes once a week (we'll bet you knew that already, you clever devil!). We'll try to refer to it by its full name for a couple of issues before switching to EW. (The new Electronics Weekly format looks exactly like InfoWorld - #%&*@$!)
We are going to stop printing the year of periodical publications unless the date is not the obvious one; i.e. proximate to the date of the newsletter the reference appears in. We apologize to the old guard amongst our readers who will consider that a shocking informality.
The British government has finally found a patsy to purchase INMOS on its own terms - Thorn-EMI, a British electronics firm. Within two days after the announcement, Thorn-EMI's stock dropped 10%. (Gee! Doesn't the stock market understand what a world-whalloper that Transputer is?) According to Electronics Week (23 Jul), the British are already planning to standardize all their fifth-generation projects around the Transputer! Isn't that a world-class "count-before-hatching" situation?
A T & T and other outfits who were casting lustful eyes at Inmos' wafer fabs will have to go to plan B.
The Aug issue is lovely, with its color renditions of Mack displays on the cover. All Mack displays, of course, are really black and white. If you read Viewpoint on page 4, be sure to read the last paragraph carefully. Read the letter in the center of page 9, 'Cry of the Hacker'. On page 18, you can learn that Apple has a disk-based monitor for Mack, only they won't let you have a copy. It might strain your brain. Remember, only computer illiterates are supposed to buy Mack! Else that monitor would be released, hmmm?
Love Apple's TV ads, seen during the Olympic coverage, which assert that Mack is a business machine. If Mack in its present form is a business machine, then your FNE is Queen of the May. (You will keep in mind that Digital Acoustics is located in conservative Orange County, CA - NOT the San Francisco area.)
Meanwhile, PCjr has a new keyboard featured in its TV ads - with a selectric layout, no less!
Market watcher Brian Doyle, as reported in Electronics Week Jul 23, p.91, asserts that the microcomputer software is enjoying rapid growth. In terms of dollar volume, he is absolutely correct. It is unclear whether he believes the number of software companies and products are also growing. We hope he doesn't believe that because it emphatically ain't so. When Hayden shut down their Software magazine, they pointed out that a year ago, 4000 software companies were running magazine ads. Now only 1000 run ads (and you wondered why those magazines were getting skinnier!).
Like we told you a couple of issues back, the microcomputer software industry is undergoing a consolidation and shakeout that is much more advanced than the corresponding shakeout in the hardware end of the business. Where's all the money going? We have
seen TV ads for 1-2-3 and Symphony four times during Olympic coverage in the past two days. A 1/8 page ad in SofTalk doesn't cut it these days.
(Brian Doyle is also the author of "The Legend of Supermicro" in the premier issue of UNIX World.)
As lots of folks have written lately, 256K DRAMS are MUCH more expensive than 64Ks, right? Sigh. In fact, 256K DRAMs currently cost only 2 to 3 times the price of the same amount of memory made up of 64Ks when purchased in the same distribution channels and in the same quantities. Since the SYSTEM price is then lower - fewer sockets and a much smaller PC board, you know - the system cost is significantly lower than a ratio of 2 or 3 to 1.
So why don't we design in 256K DRAMs? Well, we can't see DTACK customers paying $3500-$5000 for a megabyte made up of 256Ks when they can get a megabyte Grande with 64Ks for $2100. Secondly, some big outfit (IBM) can come along and suck the market dry of 256Ks at any time. If we depend on 256Ks that puts us out of business. No thanks!
We have at hand the premier issue of Computer LANGUAGE. Why the title is singular we dunno, since it is apparently the intent of the publisher(s?) to cover a wide variety of languages. They have enlisted Ron Jeffries to write a newscolumn for them, which partly explains why, on Aug 6, our most recent issue of "The Jeffries Report" is dated April.
In this first issue, the editors take an even-handed approach to various languages. There is even an article extolling the virtues of COBOL, we kid you not. (The author is a damned good writer and makes a persuasive case, incidentally.) If this first issue is typical of what is to come, we would not hesitate to suggest that you send $19.95 to Computer LANGUAGE, 2443 Fillmore Street Ste #346, San Francisco CA. Assuming you are interested in computer languages.
Why do we have this nagging suspicion that three or five issues down the line a majority of the editors will line up behind The One True Language (whichever one that might prove to be) and force dissidents out? Then we would be left with the PASCAL Paper or the Forth Force or the HALGOL Harangue or... you get the idea. Too many folk support their pet languages with a religious zeal to permit long-term even-handed evaluation. We think. We hope we're wrong, and that Computer LANGUAGE has a long run.
Most of you know that Esther Dyson edits a professional ($395/yr) newsletter on personal computers. Now, most newsletter folk are downright secretive about their publication's circulation. Esther frankly claims 2,000 - just over 3 times our 600+. Of course, that gives her $800,000 a year income vs our $12,000. Maybe we can increase our subscription rate by $380/yr? No? Aw, rats!
Unless you read Newsweek's new Access magazine, which is sort of People magazine for personal computers, you would not know that Esther is the daughter of physicist Freeman Dyson. If you are a science fiction fan, you will have heard of the Dyson Sphere and other way-out speculations...
The local Motorola sales office just sicced a genuine 68020 salesperson (hair: blond, name: Alice) on us. It was refreshing to hear Alice start out by asserting that Motorola was going to be sampling (for little outfits like us) the first of next year and would be in production the middle of next year. We seem to recall suggesting a schedule something like that ourselves a long while back...
Meanwhile, you may read that Motorola is selling samples for $487. That is correct. The total number of samples in Aug is something like 100. Which means even Tektronix, Cal Comp and TRW can buy one (that's ONE) now. Littler outfits like us and Houston Instruments will have to stand in line for a bit longer. But the assertion that Motorola is selling samples NOW is true. And misleading, for some.
We have here a promotional insert with the heading "Share The Computer Journal With a Friend" Hmmm. Friend = giggles a lot, wears Jungle Lust often and a negligee occasionally? Perhaps they really meant, Share with a Colleague? We dunno. We also don't know why they capitalized 'The' and 'With' but not 'a'.
Apple Softalk (Jun p.4) is running a design contest for a computer you can take with you when you go backpacking. Gosh, now we can take dBase II and 1-2-3 with us on a backpacking trip? NIRVANA! Now, if someone will only design a matching portable time clock so we can punch in every morning...
Gantz' column has been a fine addition to InfoWorld. John has the rare combination of the ability to write
plus something interesting to write about. Our usual reaction to one of his columns is, we agree completely BUT... In the Sep 3 issue, for instance, John writes of 'BOOM TIMES AHEAD'. We agree but... the 130 vendors of personal computers he cites (a reasonable estimate) will have to drop to 19 (or so) before his scenario will come about. That's 111 companies which won't be sharing that 'BOOM AHEAD'! Yet in his penultimate paragraph John talks of losing only five of those 130 suppliers! C'mon, John! And his final sentence may be correct but we HATE it!
You have probably read by now that Commodore has purchased manufacturing rights to the Amiga Lorraine 68000-based microcomputer (by buying Amiga). You probably read about the Commodore folk chortling over purchasing the Lorraine rights out from under Kindly Uncle Jack, which is true. You probably also read about the Lorraine being demonstrated a while back at the Consumer Electronic Show, which is not in accordance with the facts. The Lorraine depends heavily on two custom graphics chips (shades of the MindSet?) and neither of those chips were completed by the time of the Consumer Electronics Show so two heavy-duty minicomputers were pressed into duty to simulate those chips. The Lorraine that was demonstrated was just a mockup. Question: do those chips exist now?
And after the example of the MindSet, which claims superb color graphics but which in fact is not much better than the C-64 or Atari 800 graphics, we all know to regard claims of superb color graphics with a grain of salt, don't we? Still - it IS a 68000 machine.
The Commodore folk did not even attempt to disguise their glee in yanking the Lorraine out from under KUJ. KUJ responded in, for him, a temperate manner - the lawsuit he filed against Amiga is for only $50 million. You do understand that when Commodore completes the purchase of Amiga that it has also purchased the lawsuit? You may be certain that this is only the first of MANY lawsuits to be filed between KUJ (Atari) and Gould (Commodore). When friends fall out of bed, they fall all the way, don't they?
You will recall that the original IBM PC, which came with an audio cassette interface for program and data storage, BASIC in ROM, and 16K RAM was really aimed at Apple Computer's home market and the Apple II, right? But that folks, seeing the IBM logo, immediately added enough memory and disk drives to turn the PC into a business machine which became vastly successful despite a deliberately busted keyboard to keep folks from using the PC as a word processor?
We did not have to tell you that Peanut (aka PCjr) was really aimed at the Apple Computer's home market and the Apple II because lots of other people (remember the WSJ's 'a rifle aimed at Apple's cash cow'?) did that for us. A few of those other people also asserted, as we did, that any machine with that IBM logo would be judged exclusively as a business machine and that the Peanut was doomed as a home (non-1-2-3) machine.
In the last issue we suggested that IBM was faced with the Apple III problem - lots of completed inventory (of PCjrs) gathering dust in warehouses. TOO MUCH inventory to be overlooked by the independent auditors which are legally mandated by SEC rules for publicly traded companies. (By the way, IBM's lawyers are much more experienced and professional than Apple's were three years back - our issue #4 caused a ruckus in Apple's legal henhouse back then.)
IBM has just applied Apple's classic fix - bolting some new geegaws on the old turkey and in effect reintroducing it. This slides IBM by one audit and buys them a year before the next audit. The Apple III geegaw three years ago was the ProFile 5Mb hard disk and the Peanut geegaw is a very slightly improved keyboard and a lot of VERY EXPENSIVE add-on memory modules. Enough add-on memory to take the Peanut out of the home computer price range (surprise?) and allow it to run 1-2-3 (double surprise?).
Let us emphasize this: IBM's first personal computer was intended to be a home computer, but because of its open architecture it was possible for third parties to expand it into a business machine whether IBM liked it or not. IBM's second personal computer was not only intended to be a home computer, but IBM (having learned from the PC) gave it a closed architecture and an abominable keyboard so that third parties COULDN'T turn it into a business machine. IBM has just backed down. Actually, they were FORCED to back down because of 350,000 Peanuts (our estimate) gathering dust in warehouses!
(There are now some IBM apologists running around asserting that Peanut was intended to be expandable all along. SUUURE IT WAS! That's why peanut has to have an auxiliary power supply bolted on if you add more than just one memory module. YOU CANNOT EVEN ADD A PRINTER PORT WITHOUT ADDING THAT AUXILIARY POWER SUPPLY (after adding ONE memory module). Now you know whether to believe those IBM apologists!
As we told you once before, the revised Peanut is going to sell in very moderate quantities (while being hailed as a fantastic success by IBM's flacks) until the warehouses are emptied enough to quietly close out the line. You will recall that took three years in the Apple III's case.
Since EVERYBODY, including IBM, has written off the dinky diskless $595 'home' version of Peanut, we once again point out that IBM has now taken two swings at the home computer marketplace (and the Apple II) and has now struck out twice. (The first strikeout proved to be highly profitable for reasons which IBM had not anticipated.) Question: is IBM dumb enough to try for the home market for a third time with a personal computer bearing the IBM logo?
Do not believe anything you read about the new Peanut being a $995 business machine. Add up all the accessories needed to run the most rudimentary business programs and see if you are not over $2000 without including the cost of a printer and the software itself. Remember, all of the cables cost extra (honest!). You can't steal the kids' TV for a monitor. Once you have added even ONE extra memory module you have to purchase that auxiliary power supply before adding ANYTHING else, including the printer interface (which costs extra). Check the price of those extra memory modules, but sit down first so you don't hurt yourself when you keel over. Check the tiny capacity of each of those memory modules. And SUUURE you are going to get by on just one disk drive...
Once you have correctly identified the true price of the revised Peanut you will understand why IBM does not care whether sales of the PC are impacted.
IBM has rolled out its big gun, which turned out to be a popgun. Popcorn, now renamed the PC/AT, has just shot down exactly one personal computer now in the marketplace - IBM's own XT! The lowest price AT is more than double that of the PC clones (and the PC itself, for that matter). For $4000 you get ONE floppy and NO video monitor or video board. And that is the cheapest AT, which we predict will sell (at that price) like coldcakes.
The $5800 version with a 20Mb hard disk and provision for three users is something else. Nobody believes you can do anything with that stripped $5800 version, but it will, when fleshed out into a $10,000 to $12,000 four-user system, give Fortune, Altos and some others fits. Mostly because of the IBM logo, but what else is new? A few rugged individualists and status-seekers will buy the fleshed-out $7000 version as a single-user package. (As usual, the prices we are talking about do not include a printer or software.)
AS THE MIGHTY PC/AT DREADNOUGHT SWEEPS FORTH, DEVASTATING ALL WHO WOULD CHALLENGE ITS POWERFUL 80286 ENGINE, IT... what's that? You ask whether the 6MHz
80286 used in the AT is faster than the 8MHz 80186 used in the Tandy 2000? Gee, we dunno. IBM claims that the AT is 2 to 3 times faster than its PC. Tandy has claimed for some time that its 2000 is 2 to 3 times faster than IBM's PC. Both of those claims look accurate to us. If you have been swallowing that blather about how powerful the 80286 is you should be ashamed of yourself. At 6MHz, it is about 2/3 of a 12.5MHz 68000. At 10MHz an 80286 would have an edge on a 12.5MHz 68000, but you have to remember we have been shipping 12.5MHz 68000s for 27 months while it will be a year before you will see any 10MHz 80286s being shipped. (These figures are based on an average of a variety of different applications. As we have told you repeatedly, the 68000 and 80286 are each, for comparable clocks, superior to the other in specific areas. However - 6MHz is not a comparable clock!)
As far as we can see, the Tandy 2000 and the PC/AT have equal CPU capacity for single-tasking environments, which is the only sort of environment we are interested in. It is true that the AT's 80286 is well suited to multiuser/multitasking environments, but it is ALSO true that the 286 runs a lot slower with that memory management unit switched on.
For $2995 you get a Tandy 2000 with 128K, two disk drives (1.44 Mb total) and a monochrome monitor and (the equivalent of) the 'video monitor card'. For $3995 IBM gives you a 256K PC with ONE 1.2Mb floppy and no monitor or monitor card. With equal CPU power for single users, which do you think is the better deal? We both know which will sell better - the one with the IBM logo!
But the real message of the PC/AT is the chains and shackles for those three extra users. The AT will not be purchased because of what it can do for the user; it will be purchased to be DONE TO the peons shackled to it. Almost all of the AT-specific software development for the 80286 is going to be aimed at the four-user version, and that software will wear four-user minicomputer price tags. Count on it.
You do remember that we have been telling you for 20 months now that the 80286 is best suited to locking several users into separate memory spaces, don't you? You will notice that IBM figured that out, too...
How many of you remember that less than a year ago there was just one IBM personal computer (the PC) for the clone-makers to shoot at? And only one operating system for the software vendors to target? IBM has (deliberately) done a good job of stirring up the mud. Already we read speculation about 80286 clones. And IBM has a briefcase portable and a file-server coming
up - a file-server being a big disk networked to a bunch of PCs. Yes, we know Corona just introduced such a product, but it does not bear the IBM logo. That will stir up even more mud, and when the mud settles most of the clones will have disappeared.
The winds of change in this industry, which have heavily buffeted the magazines, are apparently striking out at some newsletters as well. For instance, the last issue of The Jeffries Report was dated May. The masthead at the bottom of the front page asserts that The Jeffries Report is published monthly. Really?
Unique, the UNIX newsletter published by David Fiedler under the corporate imprint InfoPro Systems, also carried a bottom-front-page masthead asserting that it is published monthly. The last monthly issue was dated April; on Aug 25 we received the 'Summer' issue. The masthead has been moved to the back page but still asserts that Unique is published monthly. Again: really?
Unique's subtitle has been changed from 'Your Independent UNIX and C Advisor', last seen on the April issue, to 'The UNIX System Information Source'. On page 1 of the 'Summer' issue Fiedler writes: "we'll talk about current events in the world of UNIX'! CURRENT? In a publication which publishes an April issue followed by a 'Summer' issue in August?
Our impression of Unique's new format is that it is now an advertising-less UNIX magazine and hence should be compared more to UNIX Review or UNIX World. Accordingly, the masthead, which formerly listed David, his wife Susan and a circulation manager has now been expanded to list four more persons. An awful lot of the stuff in the 'Summer' issue was not written by David (as is typical of magazines as opposed to newsletters). But Unique needs to find some magazine-quality proofreaders: on page 23 we find "...nor does they treat the interaction...", for instance.
We regret Unique's change in format. For one thing, UNIX World is clearly the best of the available UNIX-related magazines with whomever is in second place trailing far behind, and that includes Unique's new format.
With a subscription price of $54/yr, Unique is clearly intended to be a profit center for InfoPro Systems. The recent change in format may be a deliberate move on InfoPro's part for business reasons. The Jeffries Report goes for $15 per six issues and considering the expense of typesetting cannot possibly be a profit center for Ron Jeffries. Translation: we suspect that Ron has published the Jeffries Report for the fun of
it! (Ron does not even use his newsletter to advertise his own products! How unbusiness-like!) The recent non-publication of his Report is almost certainly related to considerations of making a living.
We hope our readers understand that we do not mean to pick on Ron or David personally. We are reporting the facts, and the facts show that two publications which claim to be published monthly just plain aren't. Which is a shame, because those two publications (Unique in its old format) were among our personal favorites.
There are too damned many computer magazines. It would be a very good thing for us readers if about two-thirds of them went out of business. While we stand on the truth of that last sentence without reservation, we also recognize that a lot of individuals are going to be hurt as (if) two-thirds of the computer magazines go out of business.
There are some wild goats on San Clemente Island, just off the coast of Southern California. Since San Clemente Island has no natural goat-predators, the goat population has expanded beyond the capacity of the environment to support it. So the Navy, owners of the island, are shooting a lot of goats for the general benefit of the goat community. This is a very good thing for the goats which survive and a very bad thing for the goats that got shot. [Does anybody want to hear our sure-fire engineer's solution for eliminating 90% of the smog in South Coast (Los Angeles) air basin?]
A short word about our own publishing frequency: it was obvious to us from the start that a newsletter written mostly by one person could not promise monthly publication. In our first three calendar years, we published 33 issues. That makes DTACK GROUNDED a mostly-monthly publication, which is what we suspected it would be when we started out.
On page 107 of the Aug issue, Terry Peterson's improvement to the original Dr. Dobb's benchmark is mentioned. On page 108 the interpretation of the errors reported by Terry's benchmark is mentioned but somehow the name of the originator of that interpretation got misplaced although DTACK #26 was correctly credited. And although it is mentioned in the final paragraph that the accuracy (actually, the resolution) of the floating point package can be estimated it is not explained HOW, nor are any examples cited.
Dr. Dobbs did report two of the benchmarks from issue #26 in the following pages: Terry Peterson's 515.4 sec Commodore/Petspeed and our 2640 sec Eagle II/CBASIC.
is what we have sometimes heard the Apple Computer Co. called on account of that's where one writes with a technical question or seventeen and never gets an answer. Our sympathy is with Apple because we have had the opportunity to see the other side of the issue.
A long while back one Peter R., who is an attorney from back east, wrote us about the Dtack boards. He wanted answers to lots of questions. We sent a reply inviting him to subscribe to our newsletter. He sent a second letter, telling us in no uncertain terms that he certainly was NOT going to spend his valuable $15 until he had these here minimum questions answered, so there! We glanced at his letter and estimated that a six-page letter would answer about 85% of what he wanted to know. The letter would naturally contain much (most) information overlapping the content of the newsletter for which he did not wish to pay. For some reason, rather than simply crumpling the letter and round-filing it as we usually would, we wrote "If you no wanna pay, go away!" across the bottom of an invitation to subscribe and mailed it.
More than a year later we received a newsletter from back east containing a writeup by this same Peter R. It seems he had been infuriated by that reply (no surprise there) but, after a year of searching for information about the 68000 he had (in desperation) subscribed to this rag anyway. As we recall, he expressed surprise (the pleasant kind) about this rag. Apparently it really DID answer his questions (as we had suggested year earlier), because Peter had purchased a DTACK board!
After reading that eastern newsletter with Peter's writeup, we wrote him a nice letter pointing out how impossible it was to write long letters answering initial inquiries and urging him to reconsider his opinion (repeated in his newsletter writeup) that we were duty bound to reply to all requests for information. You know what?
The #*&$%@! didn't reply to our letter!
Apparently the admonition to answer all letters applies in one direction only. Sigh. Look, we don't answer those letters because it is flat impossible to do so. For instance:
"Dear DTACK: I want timing diagrams of your 68000 board and information on hardware and software requirements to hook your board to an Atari 800. What software do is available for your board? How does the 16-bit 68000 run on the 8-bit Apple bus? I want to use your board to digitally equalize my stereo system." signed Mugwump
That, gentlepersons, is a very typical letter which is IMPOSSIBLE to answer given finite time and finite resources. (Let us correct that last sentence: we CAN answer that ONE letter, if we have nothing else to do on a given day. but we can't answer dozens of such letters, and dozens is what we get!) We won't even TRY to answer it, so we guess Digital Acoustics is ALSO a "Black Hole of Calcutta!" It is instructive that some folk who demand answers to their letters are themselves unwilling to reply in kind...
Some of you may remember our MAX-80 CP/M machine which is gathering dust in a storage room at home (Digital Acoustics' facility is getting crowded). When Lobo first shipped the MAX-80, they provided a 400-page technical manual and a toll-free 800 phone number for answering technical questions. So every MAX-80 owner who needed an answer would heft that 400-page manual, think for a moment, and then reach for the telephone. Guess how long the toll-free number lasted?
Newsweek magazine has spun off a new publication about the personal computer industry called Access. Our copy features lovable Steve Jobs on the front cover. One of the articles in that issue describes the hilarious story of how and WHY Bill Gate's Altair BASIC got ripped off early in the personal computer game, making Bill's BASIC an industry standard and Bill a wealthy man. It seems that the rippers thought information should be free. Well, a computer program IS information, isn't it?
Peter R. and his numerous philosophical brethren take this one step further: if they want information, someone at the Apple Computer Co. or at Digital Acoustics Inc should provide that information free, even if they have to compose a six-page letter especially for the purpose! (Even if they have to compose DOZENS of six-page letters.) Anybody who is on the other end learns REAL quick that information just ain't that free. What REALLY frosts our cake is that you can patiently explain all this to some folk and they will STILL demand that you answer all information requests! Mama mia!
This brings us to our first letter:
"On my answering machine the night of July 22 was a voice claiming to be you. The voice said that FNE was wondering how I like my RANA drives. Obviously this person was an imposter since you have told your faithful readers time after time that you are far too busy to spend time chatting with them." Robert L., York PA
Robert, it's BEGINNERS that we have no time to talk to.
Perhaps you are mistaking what we have said about correspondence? We send a 28-page note, using compressed type no less, to each of our readers once a month (appx). It is too much to expect of us to ADDITIONALLY compose a six-page letter for a PARTICULAR reader, and we won't. But we can talk faster than we can type, and our regular subscribers and customers are a valuable source of information (once they are out of the beginner category) - FNE
"...Among my chores is the teaching of a beginning course in Computer Art at Washington State University. First time around I was a good boy and tried to teach structured programming from the start. I ended up rather puzzled at the slowness with which the students grasped what seemed to be very clear concepts. Next I taught a 1-week workshop to high school students. No time for theory - let's just get 'em going! I feel that this second, messy approach was far more productive. In short, I agree with the thrust of David William's article excerpted in DG #33." Jim H., Pullman WA
Our immediate reaction is that teaching PASCAL to (aspiring) artists is like teaching bricklaying to ballerinas. Jim, have you and Robert (see above) ever considered that writing a letter is very much like writing a PASCAL program, while a phone conversation is very like writing a program in BASIC? Phone calls are unstructured, interactive, and filled with sudden pauses, tests, and changes of direction. A phone call often winds up in a place unrelated to where it starts out. An up-tight PASCAL type would doubtless ascribe that to a lack of discipline - FNE
"Since you are no longer in the digital acoustics business, why not change your dba?" Michael S., Pacifica CA
What's a dba, Michael? If you mean business name, it is Digital Acoustics that is on our corporate seal and letterhead and is registered with the state and the city and which has VERY good credit all over Orange County and which has been in business for over 12 years and...
Oh, yes: and which has many thousands of newsletters (some of them photocopies) out there bearing the name Digital Acoustics. All of the above goes out the window with a new name - FNE
"I saw a message on Compuserve that indicated you are (an) Anti-IBM mentality. I'm glad somebody is trying to organize a movement against IBM's type mentality and I'm with you all the way." Vic S., Cleveland OH
Sigh. Where do we start? There is an outfit called AIBMUGO which 1) does not like IBM, and 2) is apparently a fan of Digital Acoustics/DTACK GROUNDED. Unfortunately, their flyers are worded so that it is easy to mistake us as the headquarters for AIBMUGO, which we are not.
Explanation: we really are not IBM fans and don't try to conceal that fact. AIBMUGO, on the other hand, is a secret organization - their address is a mail drop. If DTACK was behind super-secret AIBMUGO, we would most certainly NOT place our name and address so conspicuously on the AIBMUGO flyer. Is that not reasonable, hmmm?
In the meantime, lots of folk write to us assuming we are AIBMUGO. We don't answer those letters. In fact, we wish the AIBMUGO folks, whoever they are, would be of less 'help' - FNE
"...I enjoy reading your newsletter. No ONE else discusses in readable English things like the implications of micron design rules on speed and product availability, floating point hardware and software, parts allocations, etc.
"I'm an assembly language, operating system oriented programmer/manager. I've been programming 68000's since early 1980 (password = bit operations to memory didn't work on the first masks). I have developed a 68000 macro assembler on my Radio Shack Model 16. I would like to contact a DTACK owner in the Baltimore or DC area willing to trade some weekend time on a DTACK for a fast native assembler. I need a site with a cross-assembler and XMODEM file transfer capability." James H., Columbia MD
James, don't you know that John Dvorak asserts that this is a pure gossip rag? Your insinuation that it might actually be useful is clearly out of line. We will forward any nibbles on your macro assembler from folks in your area to you - FNE
"Speaking of Weitek chips, have you looked at the Sky Floating Point Processors used in many 68000 machines? Based on the very same chips." Roy M., Santa Cruz CA
Sorry, Roy, that ain't so. The Sky board has been around for a year and performs double precision arithmetic while Weitek's double precision chips don't exist yet. (The Sky board set - that's two boards - also performs transcendentals.) The Charles River/Sky Dr. Dobb's benchmarks of 20.1 sec (SVS FORTRAN) and 14.6 sec (ABSOFT FORTRAN), both double precision, don't impress us much since HALGOL now runs at 18.6 sec, double precision, WITHOUT a hardware boost. The new
(announced?) Sky array processor that you mentioned we don't know about.
By the way, we still do not make up our letters ala Penthouse (bet you thought we had forgot!) - FNE
"I hesitate to write for much fear that I may receive loud and undeserved abuse via the newsletter, but since I am enclosing the purchase price of a full gallon Grande... I do, however, have a few questions for the exhalted FNE. With the introduction of the APPLE IIc, and with APPLE's apparent abandonment of the IIe's, what are your expectations for maintaining support of the APPLE? Are you still frowning on Franklin? ...I expect no thrashing over some of my spelling, grammatical, or structural errors in this treatise." Nathan S., Angleton TX
Nathan, APPLE's abandonment of the IIe will prove temporary when it sees the orders pouring in from its distributors. There is a genuine shortage of IIes in the L.A. area; the few IIes that are available are selling for pretty much list price on account of their scarcity. The hackers of the world adopted the II/II+/IIe a long while back and we can see no sign that they have abandoned it. That means that we will be continuing to support the IIe for a long while yet. We will not ever support any Apple II rip-off. In view of the nice check you enclosed, we certainly will not thrash you for including an 'h' in 'exalted' - FNE
"In answer to Rowland B.'s question, YES, it is safe for him to return his disk for updating. (Since our programs are not copy protected, it would seem reasonable to make a copy before sending the master anyway.) As always, all updates are free upon return of the original disk.
"The most recent update for ASSEM68K was in March, 1983. MINOS has been updated to work on the Grande. Unless presented with evidence of major problems, we plan no further revisions. If any users would like to modify any of our programs we would be happy to discuss commercial arrangements.
"Rather than sending letters, it would be best for people to call me evenings between 8 and 10 Mountain time at (602) 325-5760 (or leave a message in CompuServe mail for 70014,2104.
"Thanks for giving me the opportunity to set the record straight." Dan Davidson, Manager (and shipping clerk)
"You write TramEIl while newspapers and magazines write TramIEl. Huh?" Anon, Oakland CA
Obviously, one of us is wrong, Anon. Did you really write all those naughty books located at the start of the B. Dalton fiction section? - FNE
"I wrote to you on 24th May regarding an upgrading of my 256K DTACK Grande (arrived on 18th April) with a further 256K RAM. You have given no reply as yet, which is unlike you. I request again for advice regarding both the procedure and the price." Graeme J., Auckland NZ
Graeme, let us first hasten to explain that we DID NOT ship a Grande to New Zealand on account of Cap Weinberger and the Pentagon folk would get very unhappy with us for shipping same to communist subversives like you. Presumably somebody in the U.S. bought a board and transhipped it - fine with us.
As we have said before in this newsletter, the price of a memory upgrade is the difference in the current list price of the two memory sizes plus $50 for handling & return shipping. Right now, that's $350 total. We have to do the work here in Santa Ana, CA and we have to return ship to a UPS address in the U.S. - in other words, we have no customers in New Zealand. Now, you and we both know that doesn't fit your circumstances.
So find a local with some knowledge of micros and enough sense to avoid problems with static discharge. Have him buy some 150nsec 64K DRAMs locally (we suggest Toshiba or NEC) and install them. They should work fine. No, we cannot sell you the individual chips on account of some rinky-dink regulation having to do with the fact that we buy our DRAMs from trading companies. (We can't even sell chips in the U.S.! Honest!)
We replied to your earlier request for info right away, Graeme. Perhaps it went awry somewhere within the halls of Auckland Hospital? - FNE
"...I for one am tiring of the ongoing Jim Strasma issue. I suspect many of your readers are far more interested in your technical insights regarding high-performance microprocessors. Jim is not a bad guy, although he isn't always right. But then, I'm not always right either." Ron J., Santa Barbara CA
We're concluding the Strasma thing this issue, Ron. Like you, we think he is not a bad guy who is not always right. The problem is, we were unable to persuade him (in a letter and two long phone calls) that he wasn't right! And he DID dump hard on an individual and a company, neither of which deserved to get dumped on. We bet you would be tiring less quickly if it were you or the Code Works which Jim had dumped on - FNE
"It seems that FNE has not gone off the deep end yet, since in a back issue, he returned from the mountain with stone tablets rather than representing himself as the Author.
"I'm now legally blind from your dot-matrix print (you'll be hearing from my attorney soon)! Can't you please do something about your print quality? Though it no longer concerns me, I'm sure there are still many readers who might arrest further degradation of sight. Do you publish Braille editions for your long-term readers?
"In issue #34, page 2, $600 X 500 sets DOES NOT equal $30,000, guys! WHOOPIE-DOO! I get the point, but... did you calculate that figure with HALGOL?" Pen U., Warminster PA
Surely you are referring to an issue when we suggested that lots of folk think that when Moses came down from Ararat, he had the log tables on the back of one tablet and the trig tables on the back of the other? No personal involvement here, Pen. We don't like chiseling in any of its senses and we are much too smart (and lazy) to go around carrying stone tablets.
About $600 X 500 sets: that is commonly referred to as a $270,000 mathematical error. We did that all by ourself without the assistance of HALGOL. Let us explain to our other readers that "Pen Umbra" is the nom de plume of a 'shadow reader' of Rafael S's subscription. Can we just THINK about clobbering Pen over the head with the Dobsonian mount of our 13.1 inch reflector for such a lousy pun? - FNE
"I just got the back issues I ordered, and read in short order #1 - #26, as well as skimming the rest up to the current issue (#34). Wow, you may be right about overdoing it, but I wanted to finish before going on vacation.
"...Your attitude (toward virtual memory) reminds me of another all-out organs-to-the-wall speed freak, Seymour Cray. Seymour is reported to have said "If you want a million words of memory, BUY a million words of memory!" One could also summarize as TANSTAAFL. However, it could be mentioned that with large multi-user VM systems, the processor can be happily busy servicing someone else while the I/O system reads YOUR memory from disk. For a single user system this does not apply, and substantially cuts down the usefulness of virtual memory.
"Re #27, comparing with the Cyber 205 - you are starting to sound a lot like AMD's tricycle! Tell me who rents a Cyber at $1000/hr to do scalar math? Yes, it does do scalar math at the speed you say, and yes it
does rent at $1000 per hour, so therefore people pay $1000/hr for 8 MFlops. Sound familiar?
"'Implementing Software for Non-numeric Applications' (mentioned somewhere) is not by William White, but the same William Waite who is co-author of 'Software Manual of Elementary Functions'.
"It seems to me that you usually (with some slips) follow the dictum: 'Never offend with style when you can offend with substance.' Perhaps it should be on your masthead?
"Before I received and read the back issues, I wondered how you had gotten into this newsletter stuff. Now I know: I can picture you telling us about the FNE thing that happened to you on the way to the Market..." Zhahai S., Boulder CO
Sigh. Re: virtual memory. What you wrote in the last two sentences is EXACTLY what we have written on the subject, and more than once! It seems that people read what they want (or expect?) in this newsletter rather than what is clearly printed (in small dot matrix, true...). We CONSTANTLY receive letters (and are written about in Dvorak's column) as an assembly freak who hates high level languages. Well, this assembly freak has been working on a high level language for over two years and has caused over $100,000 to be expended (on HALGOL)! Suuure we hate HLLs!
What alternative machine has scalar performance close enough to 8 MFlops to get a complex calculation done by next Tuesday? We understand that Cyber 205s (and Cray 1s) are indeed used for this purpose but will have to yield to our resident Cyber expert, Bill J. What say, Bill?
We got into newsletter writing because we discovered a nice passive audience which never argued back - FNE
"Poor Rafael [sniff! - FNE] must spend hours writing REAL software to DoD specs. None of this hacker-type stuff the little companies are making millions off of, but professional stuff the government is paying for with the taxes it gets from those millions. There are a few layers of 'proper' specification that CS departments never tell their undergrads about so they won't switch to poetry.
"Specs are written in an HLL called English and each layer stepwise refines the previous gibberish into more ordered gibberish. Several thousand pages later, one is ready to code the 50 lines or so of APPROVED DoD HLL. 'C' and PASCAL are not APPROVED DoD languages - there are no OIT/FSTC (another agency) certification procedures. In the end, only ADA will be permitted.
"Once there is certification, the EXACT compiler to be used must be verified. But first, comments. You must write the comments in English and also in 'PDL' - Program Design Language. PDL is another HLL used in comments and documentation. One big attraction of ADA is that it may be allowed to be its own PDL.
"The very last action in writing an approved program is to whip out the 50 lines or less of APPROVED HLL in the easily-understood module. Its function in the system is now totally incomprehensible. Before you follow the advice to learn ADA to insure your employability, investigate the ramifications of the decision. Wanna talk about SHACKLED SHEEP? Once long ago on a project far, far away, the hardware folks informed a young programmer that they had goofed and the D-A for the gyro torquer worked on the opposite polarity: was that serious? 'Oh, no, it's just a little code change', he replied. An hour later, the software project leader informed this fellow, in front of all of us to make the consequences crystal clear, that he was responsible for changing each reference to the DACs in every piece of software documentation for the project!" Rafael S., Warminster PA
Raphael, one might think you did not appreciate correctly structured programming languages. By the way, that Atari-68000 interface you are designing is STILL miserably complex! You won't even get credit for brilliantly slowing the 68000 down to the speed of the Atari - the Mackintosh folks beat you to it! - FNE
[Boy, is your FNE gonna catch heck from John B. and Jeff H. for that one!
The latest issue of Jeff H.'s dinky four-page excuse for a newsletter has just arrived (misprinted, of course). Since his newsletter is shorter than 10% of the letters we get and it arrives in the mail, we guess this is the place to discuss its contents. So if one of you will just pull this here knife out of our back - ugh! - thanks.
Page 2, col 1: "All bets are off if and only if Hal builds his own 68020 minicomputer emulator. He has said oh, so many times he won't." Huh? Jeff, what in the hell do those two sentences mean? We most certainly ARE going to build at least one 68020-based product at our earliest opportunity, and no one who has been following this newsletter could ever doubt that for a moment. "His own minicomputer EMULATOR"? We dislike multiuser multitasking minicomputers enough; what's this about an EMULATOR?
Jeff asserts that Blue Sky One is unattainable. To the contrary, Blue Sky One requires about ten or twelve
pages (source code) of software and NO additional hardware. Anybody who thinks your FNE cannot turn out 10 or 12 pages of working source code has not been paying attention for the past three years. See, for instance, the five pages at the end of issue #28 which is a VERY primitive 'DOS'! It is true that Digital Acoustics has more desirable things to do than it has hands to do them, but that condition prevails generally in this industry. Remember, even the no-longer-mythical case arrived eventually!
If Apple Computer can misspell McIntosh why can't we? Or does Jeff really think Apple named Mack after a raincoat?
And bless Jeff's naive, trusting heart: he just can't figure out why Paul Lamar, in his MICRO article (Aug) seems to favor the Sage II and its p-system. Here, Jeff, let us help you dry yourself behind the ears. You don't suppose that Paul's prejudices are in any way related to the fact that he owns a personal computer retail outlet which features the Sage?
Let's see, now: Jeff asserts that the IBM PC with a compiler and a math chip outruns HALGOL, which currently uses 68000 software to perform math functions. In issue #2, over (yawn!) THREE YEARS ago, we went to some effort to point out that ANY general purpose microprocessor needs a specialized math chip for optimum floating point performance. Uh, Jeff? How come your precious Mack requires nearly 700 seconds (double precision) to run that Dr. Dobbs benchmark using MackBASIC, while HALGOL splashes through at 16.8 seconds? Huh? How come, Jeff? - FNE
"This is just a short note to try to convince you of the merits of the Mackintosh from the perspective of a thoroughly satisfied customer who is not totally ignorant of the complex issues surrounding the micro world. As you are the leading spokesman for the 68000 hacker community I think your hostility to the Mack is extremely unfortunate and unjustified. There was a time when I would try to seriously discuss some vital aspect of software engineering with you but I have concluded that talking to you about the subtleties of creating good software is like committing suicide with armor piercing bullets... it goes in one ear and out the other. Suffice it to say that the Mack software is extraordinary and, more importantly, it is filled to the brim with unlimited promise.
"When you consider that the imminent next step for the desktop Mack is 512K which will provide a RAM disk to alleviate the performance of the slow floppy (which is Mack's only real problem) and when you consider that the 68020 lies somewhere in the future for the Mack I just can't understand why you can't rally behind the
best 68000 system in existence and lend your wholehearted support in making it better. Instead you bitch and moan and make vague whining noises about 'only going North' whatever that means. In short, I am very disappointed in your negative waves, Moriarity! I always thought you had the best vision of any commentator on the micro scene, now I find out you are as blind as Pournelle, maybe you should get yourself a nice S-100 bus machine that will crawl East, West and South to your heart's content." John B., Burlingame CA
GULP! John, this is the first time a reader has ever accused us of being vague! It would be hard to decide whether you or Jeff H. (another Mack owner/fan) have been hardest on us over what we have said about Mack. And to think that we thought our comments about Mack were moderate!
If we are the leading spokesperson for the 68000 hacker community then that community is in big trouble, because we have fewer than 650 subscribers as we write this. We think it would be more accurate to say there IS no 68000 hacker community - and that's a damn shame. There should be one and there WILL be one once TRASH-68s become common (there ain't none on the U.S. market right now). But the leader (Bob Tripp?) must command a much larger circulation.
Why is it that every time we prove we are willing to publish negative letters the signature is John B.? (See issue #31, p.10) - FNE
[Moriarity is to Sherlock Holmes as Lex Luthor is to Superman, in case anybody was wondering. Definitely a black hat.]
No, this is not an 'Annie' road show. But we are tired of reading how great Mack will be... TOMORROW! Tomorrow Mack will have software; tomorrow it will have 512K; tomorrow it will have a development system; tomorrow the 'finder' program will be fixed so 40 disk swaps will not be needed to copy one lousy floppy disk; tomorrow the existing hard disk problems will, no lie, be fixed (see the Tecmar review in Aug St. Mack)... Mack will be great tomorrow!
Let us know when Mack was great yesterday, O.K.?
"Correct me if I'm wrong, but I get the impression from reading DTACK that you don't like UNIX. Astounding if true, but I can only assume your dislike is from lack of technical information on it... You also seem to have some misconceptions about the world in general, and UNIX in particular. [Umpty-eleventh exposition of
10%-90% rule deleted.] ...let me direct you to an excellent book, "The UNIX Programming Environment" by Rob Pike and Brian Kernighan. Unlike other books, this one shows you how you can do nearly everything in one command. (Also, Rob just bought his first house and can use the money.)
"My dad (same name, ex-JBL VP, now with Pioneer) says he has delt with you before, and some of his acoustic friends (Ludwig Sepmeyer and Bruce Walker) say likewise. I called up Bruce to find out if he was the Bruce Walker I see contributing to your mag, but he disclaims that and any schizoid condition that might have him doing so without his knowledge." Bart L., Murray Hill NJ
Bart, we do recall running into your dad at JPL but you have to realize that it was in reference to our somewhat largish speaker system, which features 12 each JBL LE-15A woofers and a pair of Altec 288C drivers and 311-90 horns. Like the current Motorola (68000) & Nat Semi (16081) brouhau, neither the JBL folk nor the Altec folk were especially enthused with our system! In fact, we remember your dad's 1967 transistor amplifier design and have more recently seen his name on some digital audio technical committee reports (along with John Eargle, his former Altec counterpart). We have forgotten Ludwig and the original Bruce Walker - it has been 15 years since we built that largish speaker system, you know - and the current Bruce Walker ain't the same guy.
We have heavily condensed your letter re UNIX - and we warn future letter writers that the next pinhead who explains that damn 10%-90% rule to us again is going to be openly called a pinhead! You are not the first UNIX supporter to suggest that if we do not like UNIX, we must not know much about it. (We do appreciate you sending us yet another tech journal reprint.) In fact, we have studied the subject more than most of you think. The following reply is just an outline; the reader is expected to flesh out the argument on his or her own - FNE
Therefore it appears to us that UNIX is well suited to a well-defined but small portion of the multi-user market.
"I don't know whether or not what I asked is too obvious to merit an answer, but I am sending this $10 check on the grounds that that seems to be your price for interim software, which maybe is what I want. Maybe you didn't get my previous letter?" Maurice S., Houston TX
The above letter, reprinted in its entirety, is a perfect example of how not to get a reply from us. First, we do not offer ANY interim software because that implies interim documentation. Second, the letter writer did not specify IN ANY WAY exactly what he wanted. Third, it costs us over $1,000 to turn the key in the office door each morning - $10 doesn't impress us much. ($10 WILL buy a copy of a disk that is all set up and ready to be copied - obviously not what Maurice is after.) We think - we aren't about to
search through our massive correspondence file to prove this - that Maurice is unhappy because he wants more than the three demo disks we provide with the Grande, two of them filled with source code (that's two more than you get with, say, a Mack!).
To provide a variety of samples of different code while staying within the bounds of two disks, it is quite natural that some of the source code files do not describe complete systems. NO AMOUNT OF MONEY CAN PERSUADE US TO RESEARCH A SHOEBOX-FULL OF DISKS TO TRY TO RESURRECT A 'COMPLETE' VERSION OF ALL THOSE DEMO FILES! (We're returning the check.) - FNE
[The following note was pasted over FIGHT! ROUND 2: on page 21 of the last issue, but we think it might refer to PLAYBOY WATCH: just underneath.]
"Guess I can do without your publication, I don't care to happen on stuff like this in ANYTHING I read.
"Cancel my subscription, I would like a full refund. Thank you. Most sincerely," Carolyn R., Milpitas CA
Gee, Carolyn, we're sad to have upset you. Uh - your brand-new subscription has a mailing address c/o RACAL-VADIC. In such cases, the payment was generally made with company funds. To make certain we refund your subscription properly, please send us a copy of your check (just the front; we trust you!) if you paid for the subscription personally. If the subscription was paid for by RACAL-VADIC, we need a letter on company letterhead, signed by an authorized person, requesting a refund and telling us what department to send the refund to. In either case we will issue a refund promptly - FNE
"I like your justifications for doing DG, and hasten to assure you that it does keep DG's name on many lips. Indeed, Ken Ozanne and I often use it as authority for all manner of outrageous remarks, attributing them to you, of course. At which our listeners nod and agree that you would say such things.
"...I'm stuffed if I know who your market is, nor who to aim your ads at. I thought originally it was hackers, who just wanted to play with the best chips, or people like Ken with a task that needed close to mainframe performance. However, I gather most of your sales are to OEM types for CAD stations and the like. I get a lot of enthusiasm showing a DG board to the more expensive hackers, but relatively few have the spare cash to arrange to get one. [That's it! We need more affluent hackers! Will the rest of you peons please leave?]
(Continued on p.6)
This is the second and concluding installment covering a disagreement we have with Jim Strasma of The Midnight Software Gazette, which is a PET/CBM - related publication. In last issue's installment we believe that we proved that some copyrights which Jim had questioned the validity of were in fact 100% valid. We believe we also proved that it is completely and unquestionably legal for anyone whomever to incorporate public domain software into a commercial software product [the DEFINITION of 'public domain' makes this legality implicit]. In this concluding installment we will discuss the MORALITY of commercial use of public domain software and the morality of not revealing the identities of the original authors of public domain software. We begin with Jim Strasma's reply to our last issue:
"FNE will be pleased to note that Jim Butterfield's 'Supermon' for the PET explicitly credits the Woz for its disassembler routine. If one could say the same of HESMON, none of this would have come up for discussion, but then what would there have been to read about in the last two issues of DTACK Grounded?" - Jim Strasma
Well, Jim, YOU attack persons in your newsletter by name, hmmm? And you publish under your own name, hmmm? Are you supposed to have the privilege of attacking by name while retaining the comfort of being criticized anonymously in return? Doesn't seem fair to us! (By the way, we edited your reference to our real first name to 'FNE' because otherwise a lot of our readers would not know who you were referring to. Anybody who wants to know Eloi's real name (misspelled, of course) can look at Midnite - it's not really a secret.)
This time we're going to discuss moral judgements, Jim. Unlike the question of the validity of a copyright, which is a purely legal issue, there is no absolute standard for moral judgements. I am sure that there is some whingdilly in the Bible belt who thinks it is immoral for unmarried young men and women above the age of 11 to come within 50 feet of each other! (The same whingdilly probably thinks the world is coming to an end next Thursday afternoon.)
Here is our best shot at establishing some ground rules:
While we cannot be sure that Jim agrees with us on 1) and 2), we know for certain that Jim agrees with 3) because he has addressed that issue himself. In Midnite #17, page 12, Jim discusses copy programs which will not copy themselves: "Frankly, the moral inconsistency in preventing backups of their own programs bothers me far more than that such programs exist."
I think we can agree, Jim, that the 'community' in question is national (at least), since Midnite is has a national (at least) audience. But the real sticker, the one which we OBVIOUSLY disagree on, is the question of whether moral standards are related to legal standards. Since it is unquestionably legal to incorporate public domain software into a commercial product, and since you evidently feel that this is immoral, you must therefore feel that moral and legal standards are not closely related. WE STRONGLY DISAGREE WITH YOU!
But we cannot prove that you are wrong. And you cannot prove that we are wrong! This state of affairs is a natural consequence of an absence of absolute moral standards. We will say this:
BEFORE APPLYING YOUR PERSONAL STANDARD OF MORALITY TO CRITlClZE (BY NAME) OTHER'S ACTIONS AS IMMORAL (AND YOU HAVE DONE THIS) YOU SHOULD BE CONVINCED THAT THE ACTIONS ARE IMMORAL NOT ONLY BY YOUR OWN PERSONAL STANDARDS BUT ALSO BY THE MORAL STANDARDS OF THE COMMUNITY BEING ADDRESSED. Jim, you are writing not as the pastor of a small Protestant congregation in tiny Mt. Zion in downstate Illinois but as editor of the 100th largest (by circulation) personal computer magazine in the United States!
Jim, if you say you ARE applying those national standards may we respectfully ask just whom you consulted with outside your local, rural location? Remember, Jim, as editor of Midnite your inkwell (well, ink ribbon) gives you ammunition and power that is not available to the average person. We believe you are duty bound to bend over backward to be sure that you are applying the appropriate standards before attacking someone on moral grounds. AN ACCUSATION OF A MORAL OFFENSE IN A NATIONAL PUBLICATION IS EXTREMELY DIFFICULT TO REFUTE! Especially if some folk (like us) don't think a moral offense has occurred at all...
It appears that Jim also feels that it is immoral to use public domain software without acknowledging the authors. Specifically, Jim climbed all over Terry Peterson and HES in issue #17 for not revealing the original authors of Micromon, a program which is widely believed to be in the public domain. Jim further asserted that the original authors are Bill Seiler and Arthur Cochrane.
THAT IS NOT IN ACCORDANCE WITH THE FACTS!
We have done some research on this matter and it turns out that the situation is CONSIDERABLY more complex than Jim has indicated (you may want to take notes!).
Harry Saal of NESTAR, the networking company, took delivery of his PET 2001-8 during Christmas of 1976. That PET had a monitor on tape rather than in ROM. He laboriously used this monitor to put in, a byte at a time, Baum & Wozniak's disassembler as published in Dec '76 Dr. Dobbs. (Harry also picked on your FNE for misstating that that disassembler was published in Sept '76 Interface Age. Wrong, Harry! It was published in Interface Age AND Dr. Dobbs! But the Interface Age article had priority. NEITHER listing included a copyright notice, it should be noted.) Harry had to change a part of the code because Apple and PET alpha code are different (neither uses ASCII).
After a lot of painstaking work, all done without an assembler, Harry had a source code-less monitor called HIMONDIS which included a working PET version of the Wozniak/Baum disassembler. (Some of Harry's friends also assisted in the latter stages of preparing HIMONDIS.) Harry in effect placed the additional work he had done to convert that disassembler to the PET into the public domain. For instance, he turned a copy over to Dick Tobey at a Silicon Valley Pet User's Club meeting. Dick owned Convenience Living Systems, which later became Computhink.
A copy of Harry's tape wound up with Bill Seiler, who was with Commodore Santa Clara at the time. Bill very cleverly used that disassembler to produce the world's most compact assembler. Suppose you wrote 'BCC': Bill's assembler would try HEX(00) and pass that to the disassembler. Is the output 'BCC'? No, the output is 'BRK'. Well, how about $01? No? Well, how about $02...
Bill's assembler would simply try successive values until the disassembler output matched the operator's input! Not very elegant, but DAMNED compact!
A SHORT REVIEW: the tape-based machine monitor provided by Commodore with the very first PETs was
itself used by Harry Saal to add Wozniak and Baum's 6502 public domain disassembler to that monitor. Commodore's Bill Seiler later added a crude assembler to the package which was heavily dependent upon the disassembler. (Question: was the original tape-based monitor in the public domain?) Once Bill had a working assembler, he added the original comments in the Wozniak/Baum listing(s) to the PET code, which became known as Micromon somewhere around this time.
The plot thickens. In Nov or Dec '80, at a Silicon Valley PUG (Pet User's Group) meeting HES's David Malmberg got a copy of Micromon from Bill Seiler and passed the file on to Terry Peterson. When Terry listed the files here is what he found:
MICII4 11/04/80 MICROMON FOR 4.0 ROMS (C) COPYRIGHT 1979 BY BILL SEILER ALL RIGHTS RESERVED
In other words, Bill Seiler had placed his own copyright notice on a program which included substantial public domain work by Wozniak, Baum, Saal, and others! Further, Bill did not acknowledge those authors anywhere in the source listing or elsewhere!
Terry decided that Micromon, which was not available for the VIC-20 at that time, SHOULD be made available. So Terry, cognizant of the copyright notice on Seiler's big-PET version, phoned Seiler at his then place of work (Victor Technologies) and asked for, and received, verbal permission to use Bill's copyrighted code in a commercial product to be written (in part) by Terry.
(The reader will now appreciate why we have used the phrase 'widely believed to be in the public domain' to describe MICROMON. The program was, for a while at least, copyrighted by Bill Seiler. As we explained last issue, that copyright is unquestionably valid. And we are not at all certain that Seiler has since placed that program in the public domain...)
Terry did in fact adapt Micromon to the VIC-20, and that product was (and is) marketed by HES under the name HESMON. HESMON is a $35 cartridge, so if Terry's royalties are the usual 5% (we haven't asked) then he gets a munificent buck or so (5% of retail or wholesale?) for each cartridge sold.
HESMON was written by the anonymous authors of the original PET monitor, by Allen Baum & Steve Wozniak, by Harry Saal and friends, by Bill Seiler, and by Terry Peterson. So you know what Jim Strasma did?
Jim Strasma in effect accused Terry Peterson and HES of immorally failing to acknowledge that Bill Seiler and Arthur Cochrane are the original authors of HESMON!
Further, Jim has repeatedly asserted that Seiler and Cochrane are the original authors of Micromon - THEREFORE JIM IS HIMSELF GUILTY OF THE MORAL OFFENSE WHICH HE ASCRIBES TO TERRY PETERSON AND HES! Jim himself has not acknowledged the true authors of Micromon! (To the best of our knowledge, Arthur Cochrane is NOT 'one of the authors'. Arthur has done a small amount of debugging, no more. We base this assertion on correspondence we had with Arthur about 2-1/2 years back.)
Further, Bill Seiler, who is sort of a hero to Jim Strasma, himself copyrighted public domain software written by others! But Jim, who insists on consistency in moral judgements, has never criticized Bill Seiler for doing exactly what Terry Peterson and HES did!
Like Jim, we like consistency in moral judgements. We are therefore pleased to point out that Jim, if he is to be consistent, must in a (near) future issue of Midnite either apologize to Terry and HES or else denounce Bill Seiler and himself! After two issues of Midnite have passed, we will print an update in these pages. Has anybody noticed that the question of whether Micromon is or is not in the public domain is being left unsettled?
Oh, yes: Jim wrote, "...what would there have been to read about in the last two issues of DTACK Grounded?" Well, Jim, you might check out the very humble personal apology we printed (GOOFUS GIGANTICUS) on page 7 of issue #33. That's in case you have forgotten how to write an apology.
Jim, will you PUHLEEZ quit misspelling Eloi? See, for instance, inside the back cover of Midnite #19. As always, we have lots of space for a rebuttal.
If you have a business background please skip this; this writeup is really aimed at the 50% of our audience who do not completely understand why a company cannot buy something for a dollar and then sell it for a dollar and prosper.
Let's describe a hypothetical company: sales of $10M, after-tax profits of $1M (10% of sales), corporate overhead including promotional expenses equal to $2.5M (25% of sales). The product is electronic assemblies,
so 10% of the parts (bill of materials) are wasted/defective/fail in warranty. Sales go through distribution, so the company gets 60 cents on the retail dollar. (A company with only $10M/yr sales cannot afford to deal directly with retail outlets.) That means the sales of $10M represent $16.67M retail.
The numbers listed above are fairly typical except that the corporate overhead plus promotional expenses (advertising, shows etc.) is extraordinarily low - the Apple Computer Company spent 28% of sales on promotion ALONE in a recent quarter! So let us call this company "A" and include in our calculations companies "B" and "C" with corporate overhead plus promotional expenses of 33% and 40%. Company "B" is a sort of middle ground while company "C" is somewhat profligate (in real life there are more company "C"s than "A"s). We will group the corporate overhead plus promotional expenses under the catchall title "EXPENSES".
Let's estimate direct labor, which includes testing and packaging, at 8% of sales. (A company with sales of only $10M/yr cannot afford a fully automated plant and hence has higher expenses, which necessitate higher prices - that's why few folk start a company to compete with Ford and General Motors.)
Now, about that after-tax profit of $1M, or 10% of sales: although the exact numbers vary from state to state, federal plus state corporate income taxes eat almost exactly half a company's profits at this level. (The feds want 48% of all profits over $100,000. State taxes are typically well over 2% but they are deductible, just like YOUR state income taxes.)
Here is a financial profile of our three companies:
company A B C TAXES $1.0M $1.0M $1.0M PROFITS $1.0M $1.0M $1.0M EXPENSES $2.5M $3.3M $4.0M LABOR $0.8M $0.8M $0.8M PART COSTS $4.7M $3.9M $3.2M ---------- ------ ------ ------ TOTAL $10.0M $10.0M $10.0M
Remember that we must allow for 10% wastage/defective/warranty failure. After all, if Digital Acoustics determines that a 68000L12 is defective when testing a Grande, the buyer would be unhappy if we didn't replace it, right? Even when the Grande has been used 8 months but is still in warranty? But a lot of that 10% represents mismatches in quantity buying over the life of a product, parts damaged in production, the occasional circuit boards that we can't make work after wave soldering and such. Add it all up and and it comes to 10%.
company A B C PART PRICE $1.00 $1.00 $1.00 PART COST $1.10 $1.10 $1.10 WHOLESALE $2.34 $2.82 $3.44 RETAIL $4.07 $4.70 $5.73 3.90
Back in 1961 we spent the summer at the Tektronix engineering facility in Beaverton OR. That was when we first learned that the parts cost had to be marked up a minimum of 4 times and an average of 5 times in the final selling price. In 1961 you could get super-neat transistors if you wanted to pay, say, $35 each. We still remember the screams of the engineers when management would tell them a pair of such transistors would raise the price of, say, an oscilloscope plug-in by $280 all by themselves!
It is understood that the costs and margins outlined above are appropriate for light electronic manufacturing outfits like us (although our sales are a LOT less than $10M/yr) or for instrumentation companies whose manufacturing, test, and sales procedures are similar. (A company which wraps aluminum around a couple of Tandon floppy disk drives and installs a power switch and pilot light on a front panel and a third-party switching power supply, fuse, and power cord on a rear panel can get by with somewhat smaller margins. Real estate agents, who have no manufacturing, test, or inventory problems, can get by on a 5% or 6% margin.)
But most companies who make products like ours who mark up the basic parts cost less than 4 times are on their way out of business. There are only two ways to get by on smaller margins.
One is to drastically cut overhead. The best way to do this is to copy somebody else's product. Have you noticed how similar a lot of Apple lI-compatible disk drives are to each other and to Apple's Disk II? VERY low development cost there. Franklin became, for a time, the 12th largest personal computer manufacturer in this country while keeping development costs very low. Most, but not all, IBM PC clone-makers have very low development costs. Another way to hold down development costs is to hire an engineering team to design a product and then lay them off as soon as the design project is completed. We understand that last practice prevails in the low-end Winchester disk drive market - nobody wants engineering departments, just production departments. This gives very good short-term financial results! The long-term effects are questionable but in this country a manager is hired or fired based on short-term results.
The second way is to market via direct mail. We do this, and so does Lobo with their MAX-80 machine.
Sinclair in Great Britain has adopted this marketing stratagem for their QL machine, and has announced that the machine will eventually be marketed the same way in this country. The figures we just printed suggest that an efficient company can sell a dollar part for as low as $2.34 by selling via direct mail - and that is in fact very, very close to what we (Digital Acoustics) have been doing for the past six months!
Still, there has to be some special circumstance for a mail-order company to be successful. Most folks prefer to buy in local stores where they can see and touch and get support without having to send a box via UPS.
So there you have it - you can buy dollar parts from mail-order suppliers like us for as little as $2.34 or you can buy retail for as little as $4 (less with a 10 or 20% discount, naturally). Can any company sell dollar parts for $5.73 retail and stay in business? SURE! Hewlett-Packard does it all the time. In fact, lots of folks think HP stands for High Prices. If you have super-good engineering and big promotion budgets, as HP does, your overhead is going to be a lot higher than average.
We have seen folks selling dollar parts for less than $2.34 - usually memory expansion boards for, say, S-100 systems. And we have regularly seen those folks go broke while leaving some paid-up customers stranded. What's that? You say YOU know how to operate on smaller margins and still make a profit? Then for heaven's sake start your own business and get rich, because you know something that nobody else does!
There is no such thing as a free lunch.
is a science-fiction writer whose career happens to coincide with both our interest in science fiction and the modern development of the computer. Naturally, computers have heavily influenced many of his novels. We are going to take a brief look, specifically, at five of those novels. They are:
THE ROLLING STONES (1947, see note) STARMAN JONES (1953) CITIZEN OF THE GALAXY (1957) THE MOON IS A HARSH MISTRESS (1966) THE NUMBER OF THE BEAST (1980)
(Note: Heinlein's The Rolling Stones is a revision of his Boy's Life serial, Tramp Space Ship, which was written in 1947. The actual copyright date is 1952. The dates given for the other books are the original copyright dates; it can be assumed that the books were written a year or two earlier than those dates.)
The first significant action in The Rolling Stones is when the Stone family, then in residence on the Moon, purchase their own private spacecraft and blast off for Mars. Just like that? Well, no. First they have to stock the craft with the necessities of life: "By the orbit most economical of fuel the trip to Mars from the Earth-Moon system takes thirty-seven weeks. Thus it would appear that the seven rolling Stones would require some seventy-five thousand pounds of consumable supplies for the trip, or about a ton a week. Fortunately... air and water in a space ship can be used over and over again with suitable refreshing." (You will remember that this was written in 1947!)
So they stock up, fuel up, and blast off for Mars? Uh uh. It seems the most fuel-efficient method of getting from the Moon to Mars involves boosting from the Moon just hard enough to nudge over the Moon's gravitational well into the Earth's well. Then, with a correctly calculated orbit, the spaceship falls very close to the Earth, just missing the atmosphere. At that moment, with the fuel having the added kinetic energy of motion (1/2 em vee squared) the spaceship blasts for Mars. With that correctly calculated orbit.
So how did Mr. Stone calculate that orbit? With pencil and paper! (Slide rules and three digit accuracy don't cut it for orbital calculations.) A computer is NOT used for the calculation and in fact no computer appears at any time in the novel. So how did Heinlein, the author, correctly calculate the orbit? With the help of his wife, with pencil and a LARGE piece of brown butcher paper, on the floor of their kitchen!
In Starman Jones (1953) the care and feeding of a computer on a spaceliner (passenger spaceship) is central to the plot. It seems that humans speak decimal and computers speak binary, as in ones and zeros. Transferring decimal information into the computer is a vitally important task, performed using a book with tables. One ship's officer reads from the book, a tech flips toggle switches to the indicated one/zero pattern and a another officer double checks using another book.
And then the books are lost.., but not to worry, our hero has a photographic memory and has all the decimal to binary conversions (and other stuff) memorized!
In Citizen of the Galaxy (1957) the protagonist (our hero) destroys an enemy spaceship with the assistance of a computer which appears to be more analog than digital although the description is not complete. He uses a nuclear-tipped missile which has little self-guidance and which travels at velocities comparable to the enemy craft. It is therefore necessary for the launch officer (our hero) to guess the near-future position of the enemy craft, which is closing. And he can't just wait until they get real close on account of they have a paralysis beam... This takes place while he is on a commercial trading (space) vessel.
Later, as a recruit on a Space Patrol cruiser, he lets it slip that he once destroyed an enemy raider. When a black-hat sneeringly asks how, our hero replies: "I used a Mark XIX one-stage target-seeker, made by Bethlehem-Antares and armed with a 20 megaton plutonium warhead. I launched a timed shot on closing to beaming range on a collision-curve prediction!"
In The Moon is a Harsh Mistress (1966), the entire action revolves around a very large, very fast computer which 'comes awake', reads every book ever written pertaining to politics, economics, warfare, history etc - all in a few hours - and then leads a successful revolution freeing Moon-dwellers from the tyranny of Earth. (In this novel, the Moon has for 100 years been a sort of Botany Bay for dumping political prisoners, criminals and other Earth-undesirables.)
This computer has almost 100% free will, and is almost never 'programmed' (once it 'comes awake'). Exceptions are handled verbally. Interestingly, after the decidedly portable computers of Starman Jones and Citizen of the Galaxy, we find ourselves back with the huge cavern-full-of-electronics computer beloved of B-rated movies (remember the IBM 360/195?).
Question: how do the Moon folk fight a war with Earth? Answer: they throw rocks (honest!).
In The Number of the Beast (1980) we find a computer even smarter than the one in The Moon is a Harsh Mistress and highly portable - she seems to be built into a latter-day hot rod. This latest book is a tour-de-force initially involving random numbers and statistical analysis. When Heinlein tires with those concepts, it turns out that the universe can be described mathematically and therefore can be CONTROLLED mathematically!
This has some rather interesting consequences. How would you like to personally, for real, have lunch with Alice and the Mad Hatter?
We are sure that some folks who have read The Number of the Beast felt that Heinlein, in that book, fell over the borderline separating hard science fiction from fantasy. Not so. We have no contact with Heinlein directly or indirectly but we are almost certain that The Number of the Beast was strongly influenced by Information Mechanics by Frederick W. Kantor (John Wiley & Sons, 1977). Information Mechanics takes a new and startlingly different view of physics. We cannot possibly cover this book (which we have in our personal library) in the space available here, so let us be content to reprint a paragraph from the book cover:
"This new conceptual basis looks at physics by keeping track of amounts and representations of information. It is guided, in part, by the fundamental insight that physics must deal with, and only with, all of the information accessible to an observer. It explores aspects of a physical picture in which information represented by physical systems appears fundamental, and explains how it may be formalized."
Wiley & Sons is a big - make that VERY big - publisher which is conservative and does not publish books by ding-bats or wild-eyed rug-chewers. We don't guarantee that Information Mechanics is 'right' but we DO guarantee that it is thought-provoking.
Why are we sure that Heinlein would have known about this book? For starters, there is a five-page review in the Apr '79 issue of Analog, the hard-core science fiction magazine (no magical sword, dragons or fairy princesses here!). See pp170-174.
Take one highly skilled science-fiction writer, toss in lnformation Mechanics and one hell of a smart computer, then blend well. Result: The Number of the Beast and its mathematical manipulations of reality.
Now go back to 1947 and picture this same highly skilled science-fiction writer beginning a novel by calculating a Moon-Mars orbit using pencil and paper and brown butcher paper on the kitchen floor...
If you follow the WSJ or the Business Section of the Times (NY or LA) you will have noticed that the Warner/Atari/Tramiel deal seems to be unraveling. It seems that KUJ has only been able collect about 10% of that $300M in receivables he inherited from Warner. Today is the last day of August, and KUJ is reported to be seeking a cash payment from Warner on the order of $50M. Remember, Warner did not receive so much as one
copper cent from Tramiel Technologies Ltd. for Atari. Why does KUJ think Warner would pay up? Is Warner afraid that KUJ might walk away from the deal? Would that create a REAL can of worms for Warner? Was the Commodore/Amiga deal a much bigger setback for KUJ than initially appeared?
Remember, KUJ had cash from lots of full advance payments (right, Le Roy?) when he was bringing the PET on line in late '76. He had lots of cash flow from (international) sales of the PET when bringing up the VIC-20 and lots of cash flow from VIC-20 sales when he brought up the C-64. Question: where is KUJ's cash flow today? Answer: we dunno. Worse, KUJ doesn't either. [Newchums: KUJ = Kindly Uncle Jack (Tramiel).]
PERMISSION IS HEREBY granted to anyone whomever to make unlimited copies of any part or whole of this newsletter provided a copy of THIS page, with its accompanying subscription information, is included.
THE FOLLOWING TRADEMARKS ARE ACKNOWLEDGED: Apple; II, II+, IIe, soft: ProDOS, LISA, Mackintosh?: Apple Computer Co. Anybody else need a 186th million ack, have your legal beagles send us the usual threat.
SUBSCRIPTIONS: Beginning with issue #19, subscriptions are $15 for 10 issues in the U.S. and Canada (U.S. funds), or $25 for 10 issues elsewhere. Make the check payable to DTACK GROUNDED. The address is:
1415 E. McFadden, Ste. F
SANTA ANA CA 92705
THIS ISSUE'S REDLANDS is the latest schematic of our experimental math board. Note that we've made provision for 12.5/6.25MHz clocks or 10/10MHz clocks. We've added a monostable multi, U6, to limit the width of the *SPC pulse. With wait states, as in our Grande, the *SPC pulse proved too long, absent that mono.
As with the schematic on the back of issue #25, there are no secrets hidden inside the PAL U1. That's just a GENERAL address decode device. Our experimental math board uses two or three standard glue LSTTL chips for memory decode because we need a SPECIFIC decode.
The Nat Semi ap note is hot off the press - so new that we do not have time to analyze it for this issue. The schematic they suggest is a heck of a lot more complicated than the one on the next page, though. And it does not allow for a 12.5/6.25MHz clock ratio. THE SCHEMATIC ON THE NEXT PAGE HAS BEEN THOROUGHLY TESTED! In fact, we have sold about 20 boards since Dec '83 based on that schematic. By 1985 you might start seeing some OTHER companies sell 68000 systems with 16081 superchargers...
MOTOROLA 68000 TO NAT SEMI 16081 INTERFACE CIRCUIT WITH DUAL CLOCK
Now this is color. Brilliant color to clarify CAD information and increase comprehension. RGB color on a high-resolution 19" raster screen to speed up work and make you more productive. Vivid color on the fastest-selling low-cost CAD system in the world today.
Color is a new option on the Cascade X.
But not the only option.
Now with the Cascade X, 160 separate commands can be activated by voice alone. So instead of using keystrokes, you can merely say "redraw"; the Cascade X matches your voice-print with one in its memory and executes the command.
The Voice Option can substantially decrease learning time, and increase speed and productivity.
But that's not all there is to the Cascade X.
Cascade's "the Associate" for architects and MDAS for engineers are but two examples of our high-powered applications software designed to give you mainframe performance. And the Cascade family is both compatible and networkable.
So you can start with a single workstation and add more later.
If you're serious about CAD, call or write Cascade today. We'll give you the full story on the Cascade X and the name of our representative in your area.
The Cascade X. Simply brilliant.
1000 South Grand Avenue, Santa Ana, California 92705 * Telephone: (714) 558-3316 * For more information, ask for Lisa Anderson. Distributor/Dealer inquiries invited. A member of the KTI Group of Companies worldwide. Cascade products are available to educational institutions through an exclusive marketing agreement with McGraw-Hill.