Issue # 34 DTACK GROUNDED Newsletter - August 1984

eZine lover (@eZine)

Published in 

DTACK GROUNDED Newsletter

 · 1 year ago

UNDER A CURSE?

Over one and a half centuries ago, a small trading vessel (a sailboat, naturally) was trying to round the Cape of Good Hope from east to west. Well, the prevailing winds are always against such passage but on this occasion they were fiercer than usual. It seemed that the small trading vessel was, at times, making more progress backward than forward. The First Mate suggested that perhaps it would be best to seek a safe harbor and wait out the blow. The Captain - van der Decken by name - instead swore to press onward 'even though God Himself may oppose us!'.

This single impious outburst apparently caught Someone's attention. It is said that Captain van der Decken and his ship 'der Fliengende Hollander', sometimes called 'the Flying Dutchman', are sailing under a curse and are STILL trying to round the Cape from east to west! We do not recall being similarly impious...

A FOURTH YEAR? REALLY?

We mailed issue #1 of this rag on 25 Jul '81. Even though that issue bore the date July and this one bears the date August, it is THIS issue which marks the beginning of our fourth year as a newsletter editor instead of an honest hardware designer. Being a newsletter editor is kind of a pain at times and a lot of fun at other times but one thing it is not, at least directly, is profitable. We justify spending time on this rag as a relatively cheap way to keep Digital Acoustics name before a significant portion of the public. (You will please excuse us while we make a small but significant bow to Mammon.)

But you would think, after completing a three-year apprenticeship, that we would know something about writing a newsletter. Nope. We can't even find a textbook on the subject. So you will please excuse us if we seem ignorant of proper newsletter-craft at times; we have a legitimate excuse!

Page 1, Column 2

SIMPLE SYSTEMS REPORT:

It is hard to believe, but three years ago we were the ONLY outfit selling 68000-based products for personal computers. The 68000 was, as everybody but us then knew, exclusively for use with $20,000 (and up) minicomputers. The wonder of our 4K minimum 68000 static RAM boards was not that they were so great but that they existed at all.

Times have changed. Back then, the 68000 boards were a fun sideline to our main business, which was making environmental noise analyzers with imbedded microprocessors - and real time clocks, and graphical printouts, and... Of course, that was before Reagan was elected and the annual budget for the EPA's Office of Noise Control went to zero dollars. Now, the 68000 boards and related products ARE our business.

The industry has changed too, we think for the better. Now there are lots of folk peddling 68000-based product; one of them (the Stellation 68008 board) costs less than $250 and does not have ANY memory on board. Accordingly, the industry no longer needs somebody out there proving that a super-simple 68000 system can be made to work.

We still favor simple 68000 systems (preferably with 16081 math chips as soon as Nat Semi gets that part finalized and into real production). But the simple systems we now favor are NOT as simple as possible. Just simple as in 'single user, mostly single tasking'. By MOSTLY single tasking we mean that we accept things like print buffers and a few other goodies like timers; the kind of stuff that can be included in a language/operating system without needing expensive and complex memory management.

There are a lot of folk in the computer industry who are using the 68000 in ways we do not approve (shame onthem!). We remain adamantly opposed to the reprehensible practice of chaining more than one user to one microprocessor. If we MUST use a high level language, and most of us must for a variety of good reasons, then let that high level language be written in assembly language, NOT busted in yet another high level language! Unfortunately, it seems at times that we are a lone voice crying in the wilderness over these latter issues. (In Apple land, that is; the word is getting around elsewhere.)

APPLE-OOPS DEPT:

The IIc serial output is about 7% slower than specifications. Well, close counts, doesn't it? NO? How about good intentions?

Page 2, Column 1

HALGOL REPORT:

After writing most of the code required to use floating point arrays, we realized that we could not test the code because we had not implemented integers. So, our full-time HALGOL programmer is now busily implementing integers. To be incautious, it does not seem to us that strings or string arrays will be much of a problem once integers are implemented. So look for release 3 at the end of this summer, with integers, strings and arrays but without a DOS. DOS will be in release 4, which should fill out the major requirements of a computing language and operating system.

We are in the process of negotiating with at least one outside contractor to implement needed parts of HALGOL such as a PRINTUSING statement. This will accelerate the development of the language (or make up for the fact that one of our two resident HALGOL programmers has not been holding up his end of the job lately). The HALGOL object code is now larger than 21K and will pass 32K before the year is out.

Now that we have a couple of year's experience developing (or supervising the development of) a high level language, we know another reason why interactive compilers have not been common up to now: it is twice as much work to develop one! It is much more difficult to write a compiler which permits listing expressions in their original form, which is necessary if the language is to be interactive. Don't worry, though. We're much too far along to quit now!

Other favorable developments include the fact that the Phase Zero cross-assembler has been adapted to the CBM 8032 (PET) and is being adapted to the CBM 64 (we hope to have more details elsewhere in this issue). In a separate development, somebody else has successfully interfaced a Grande and a CBM 64. All he needed were two connectors and some wires!

We get repeated inquiries as to when we are going to tie DTACK to the IBM PC. Once again: to do that, we need SOFTWARE. Translation: we need HALGOL. It looks like we are getting close. The CBM 64 looks even closer because some outsiders are working on that.

We are not working on adapting HALGOL to Mack because Apple has not yet shown the way to disconnect that horribly complex exclusively north-seeking operating system. And because HALGOL is not complete yet. Oh? You want US to disconnect those icons? Who pays the bills while we do all this work, hmmm?

A SUMMARY: HALGOL is progressing as fast as possible (if not as fast as could be desired). The language is past the halfway point and everything that has been written works and works FAST!

Page 2, Column 2

SON OF VDHR REPORT:

The VDHR board set represents high technology at a low price. A VERY low price. We think it is safe to assert that nobody else out there is pushing pixels every 14.7 nanoseconds (67.888MHz) using TTL technology and two-sided boards! Now, if only it were possible to use those boards without spending a quarter of a million dollars...

We know there are some folks who would like reasonably high resolution graphics, preferably color, but who do not happen to have a quarter million dollars' pocket change. For those persons, we have designed what is probably the lowest-priced 7220 board in existence. (While the price has not been finalized, it will be within hailing distance of $700.)

In addition to that 7220, let us tell you about all the high-tech features our new graphics board has. It has, ah... on the other hand... then there's, er, no... um. Uh, it seems that this new board does not represent very high technology, folks. That's why we have named it 'POTBOILER'.

The board layout was completed on 11 July and we hope to have prototype boards in hand by the time you read this. Later this fall you will be able to add color graphics to your DTACK board and have $248,500 pocket change left over!

TRUTH vs. FICTION:

When a semiconductor company tells you what it did in the past, that is sometimes the truth. When one tells you what it will do in the future, it is almost always fictional (on the optimistic side). The WSJ story about Motorola's release of (information on) the 68020 chip was a curious mixture of truth and fiction.

TRUTH: A Nat Semi spokesperson asserted that Nat Semi had shipped, since last October, about 500 chip sets for the 32032 at $600 a set. That is an entire $30,000, guys! WHOOPEE-DOO! If you had asked Nat Semi last October how many chip sets it planned to ship by the end of June the next year, it may have replied "$30 MILLION worth!" A lousy 500 chip sets in nine months? Geez! (A 'chip set' includes some peripheral chips; the 32032 itself is $220.)

FICTION: Motorola is going to ship 100,000 68020s in 1985. Sure it is. Well, at least Motorola is not claiming that they will be in production in 1984. Thank goodness for small favors, we guess. (EVERY semi house lies in its teeth about future schedules because a company which told the truth would be at a severe competitive disadvantage.)

Page 3, Column 1

But we think that Motorola probably WILL ship production 68020s in the second half of 1985. Motorola has announced that the initial price will be $487 each, about $12 less than we had expected. (Initial production is always in small quantities and so the price is selected politically. You DO understand that it does not matter a rat's hindquarters whether Nat Semi got $60 or $600 or $6000 per chip set for the 500 sets it shipped the last nine months, don't you?)

$487 is a very moderate initial price for the 68020. We paid almost that much for the first 12.5MHz 68000 we bought. And never regretted it. Now, if we only had a time machine so we could duck into the future about 15 months or so...

A 68020 PRIMER:

This is the short course on the 68020: It comes packaged in a grid array with 114 pins. Bye, bye two-sided boards, hello six-layer (and up) circuit boards. Sensible persons designing a board set will use 256K DRAMs even though they will be in short supply and hence expensive in the second half of 1985. 32 DRAMs will be required so the minimum memory configuration will be a megabyte. 16MHz 68020s, when they become available, will be about 4 times faster than a 12.5MHz 68000. Look for those initial production 68020s at $487 to be 10MHz or 12.5MHz units.

The 68020 uses three levels of pipelining and so executes many instructions faster than they can be fetched - and remember, they are fetched 32 bits per memory cycle! Therefore, the 68020 incorporates a substantial instruction prefetch queue, 64 32-bit words or 256 bytes. This means that it is very difficult to predict on paper the time required to execute a certain series of instructions (bye, bye software timing loops). Minimum and maximum timings are given for the instructions. For some instructions, the minimum time is zero. (A natural consequence of having a cache and three levels of pipelining.)

You will find it hard to believe the speed of the 68020 while executing loops which fit entirely within the cache...

Certain instructions operate MUCH more swiftly than on the 68000. For instance, a barrel shifter is incorporated, so any number of shifts (0 - 31) can be executed in 1 to 4 clocks (static) or 3 to 6 clocks dynamic. (Static means the shift count is in the instruction field, dynamic means the shift count is in a data register.) 32-bit signed and unsigned multiplies and divides are included; a 32-bit multiply takes 41 to 44 clocks. A 16-bit multiply takes 25 to 28 clocks.

Page 3, Column 2

Some new instruction types have been included: string manipulations, compare & swap, trap on condition code... Another stack has been added, this one for exceptions. Another trace bit has been added to the status register; this one performs trace only on conditional branches.

A bunch of new addressing modes have been added. One is what we have been calling 'double-indirect' but Motorola calls 'memory-indirect'. This means that an address register points to a location in memory; the location in memory contains the address of the operand or destination. This means, for instance, that the innermost kernel instruction of HALGOL (and FORTH) can be reduced from two instructions to one. Another dynamite instruction is our old friend 'indirect with index' to which has been added yet another index register, this one scaled by 1, 2, 4 or 8! The neat way this can address a numeric array is, we hope, obvious.

So much for the short form introduction. The 68020 is a logical successor to the 68000 and 68010 and will extend our interest for awhile (and, we hope, yours too). It is going to be extremely interesting to see Intel's benchmarks against the 68020! We think Intel will discreetly ignore the '20'. Now, if it wasn't a year or so before we could buy one of these goodies...

LET'S DESIGN A PRODUCT:

What product? Why, a simple 68020 attached processor, naturally. And since we did something very much like this over three years ago, we can build on our previous experience. This time we know in advance that Motorola is going to favor enormously complex systems exclusively, so we shut up until the day we can buy a couple of samples over the counter at Hamilton Electro Sales. When we did that in Nov '80 with the 68000, we stupidly expected assistance from Motorola and what we got was hindrance. So we wasted seven months before striking out on our own by grounding DTACK (you know the rest).

So this time we will have experimental printed circuit boards ready in advance of actually receiving the parts. We have at least 6 months to prepare, so this should not be a problem. What kind of printed circuit board? Not a two-layer board like the ones we use now. Motorola talks about six-layer boards and then muses that six layers may not be enough. In addition to that very tight pin-grid-array package, we have thirty-two address buffers (34, really - A0 and A1 get replaced by four lines, just as the 68000 replaces A0 with LDS and UDS). We have 32 data buffers and several more output buffers such as WE, AS... what we are saying is, when all of these buffers turn on at once during a write memory cycle, the ground and VCC lines are going to go

Page 4, Column 1

SPROINNGGG! That's why we need more than two layers: we need a ground PLANE and a VCC PLANE! And Motorola is now speculating whether one ground plane will be enough. First conclusion: the CPU circuit board for a 68020 system is going to be expensive!

This immediately brings up the question of whether we try for a minimum system analogous to the half-gallon GRANDE, which is a single-board system not counting the six-IC interface. That means a largish six-layer board, and largish six-layer boards are more expensive than smaller six-layer boards which contain only the essential CPU circuitry. On the other hand, breaking a minimum system into multiple boards entails buffer expenses, connector expenses etc. On the third hand, an expandable system will have those buffers and some of the connectors already. We say a decision could go either way, but a prototype system is more easily checked out - and modified - if we isolate the memory from the six-layer CPU board. A multi-board 68020 system would be large 3-dimensionally where our 68000 boards are large 2-dimensionally.

What about the expansion bus? The existing DTACK boards use two 50-pin expansion connectors, where every other line is a ground line. So we have two connectors with up to 25 active lines each. The 68020 needs at least 16 more data lines and at least 8 more address lines. In fact, about 32 extra active lines are going to be needed. It seems likely that one additional 50-pin connector is not going to be enough. In any event, the 6.5 inch wide DTACK boards cannot mount a third such connector on the narrow end where the expansion bus is now. So look for a memory expansion bus on the WIDE side of the circuit boards, just like the (ugh!) S-100 systems and the Multibus systems.

It should by now be obvious to the most casual observer that a 68020 system cannot be compatible, hardware-wise, with the existing DTACK 68000 boards.

What bus do we use? A 'standard' bus, or a custom one like DTACK and SAGE now use on the 68000? Well, we can choose between four 32-bit bus structures: VME, Futurebus, Multibus II and TI's NuBus. All of these buses are intended for use in incredibly complex minicomputer systems. And we DO mean INCREDIBLY complex. Since we believe in simple systems, it does not make sense to fool around with any of them. And we're smart enough to realize that any company which makes cost-effective boards which conform to one of those buses PHYSICALLY but does not fully conform ELECTRICALLY (handshakes, DMA protocols etc.) is just setting itself up for a constant torrent of abuse. Nope. Custom bus, here we come again. You disagree? It's a free country, chum. Go build your own enormously complex and expensive minicomputer. Write occasionally?

Page 4, Column 2

(You may want to know that the VME bus restricts you to too-small boards and to connectors which implicitly require multilayer boards to be used, including memory boards. Motorola's nice $10,000 68010 VME minicomputer uses a CPU/memory board which does NOT conform to VME standards - it is not even close! - only the peripheral slots are VME.)

We wind up with a product which is physically incompatible with others BUT - halleluja! - is SOFTWARE COMPATIBLE WITH THE DTACK BOARD! [Also see p.9]

HOW MANY SHEEP?

While every other 68020 system designer in the world has his/her eyes lit up like a pinball machine at the thought of how many sheep can be shackled to a 68020 machine, we still believe in the concept of one person, at least one 68020 and mostly one task (we are now willing to include print buffers and keyboard type-ahead buffers and such under our loosening definition of single-tasking). That means that our simple 68020 system will not need memory management to prevent users or tasks from colliding. And that makes our simple 68020 system even simpler and cheaper.

But that does not mean the 68020 system will be cheap, initially. Neither was the DTACK system, initially. Remember $1695 8MHz 92K 68000 systems?

WILL 32 BITS REPLACE 16?

Of course! The question is, when? And even a simple question like that needs to be qualified: replace in what market? DTACK sales or general personal computer marketplace sales? Let us first talk about the general marketplace, and review some past history.

The 8086 arrived as a purchasable product in 1979, the Z800X in 1980 and the 68000 in late 1981. When did these 16-bit processors replace the old 8-bit Z80/8080 and 6502? General observation suggests that more 6502-based personal computers were sold in 1983 than in any other complete year to date and they are STILL selling like gangbusters in 1984! We cannot judge whether 1982 or 1983 was the peak year for Z80/8080-based personal computer sales (which do you think?). Although these sales appear to be easing off, the basic Z80 machine is being supplanted by 8088-based machines and a lot of folks, including Intel, think the 8088 is an 8-bit CPU. In other words, it is now five years since the commercial availability of 16-bit CPUs and 8-bit sales are still going strong!

As we peer into our cloudy crystal ball, it appears to us that it will be three more years before 16-bit personal computer sales surpass 8-bit sales. That's EIGHT YEARS after 16-bit CPUs became commercially

Page 5, Column 1

available! Now, let's see: 32-bit CPUs have been commercially available for less than a year now...

All right, what about the DTACK market? It is, after all, for persons interested in very high performance and the 68020 will feature very high performance when it is commercially available in a year or so.

THE ENVELOPE, PLEASE! (Just a moment here...) TADAA! The answer is, 'WE DON'T CARE'! We will offer both 68000 boards and software-compatible 68020 boards and let our customers take their choice. And just as the GRANDE was the best dynamic RAM board we knew how to make, rather than being busted somehow to protect our static RAM board sales, the 68020 board (system?) will be the best we know how to make and it will not be busted in any way. We have no desire whatever to handicap any one of our products in any way to protect the sale of any other product. Some companies think - and act - differently.

68020 boards are initially going to be VERY expensive. And at about 10MHz for initial units, the 12.5MHz 68000 will be more cost-effective (if slower on an absolute basis). You pays your money and you takes your choice.

MATH CHIP REPORT:

We have NOT abandoned our plan to make a multi-68000 and multi-16081 system. We are simply waiting to see when Nat Semi is going to finish messing around with the design (the rev E mask design is currently under way). Making a significant financial and engineering commitment to a part whose design is not finalized is foolhardy to say the least.

Unless you know something we don't, the 68881 mask layout is not yet begun and WILL NOT be begun until the 68020 mask is finalized. The question is not what year the 68881 will arrive but what decade. Remember, Motorola originally promised the 68881 for 1982.

To give you an idea of how far off the 68881 is, the Motorola folks are no longer unhappy with us for revealing that the 16081 works well with the 68000. To the alert observer, that fact speaks volumes!

The Zilog 8070 exists only as a model number. Forget it. Whoever wrote us about the AMD math chip: huh? WHAT chip? Surely you cannot be talking about that long-obsolescent AMD 9512 which Apple Computer designed into LISA I? Open up a LISA I and you will find a nice vacant 24-pin socket. If you check what signals go to what pins, you will find it is wired to accept an AMD 9512. Apple had originally negotiated to buy lots and lots of AND 9512s at five bucks each so LISA I would have hardware math chip support. Then they discovered that the 68000, even crippled with a slow clock and

Page 5, Column 2

sharing memory cycles with the video, was a lot faster in software... [Also see page 24]

DTACK ADVERTISING:

As some of you have noticed, we haven't been doing any. In the previous two months, we were running at absolute peak production capacity and, more important, absolute peak PARTS capacity. It would bore you to discover the numerous parts which we just barely got in time to keep production running. (We can assure you that WE were not bored when we were out chasing those parts!) We had anticipated that crunch and had stopped advertising accordingly.

But we are not operating at peak capacity right now, and will not be for another 2-3 months. And we have the case and power supply that we wanted to have the next time we ran an ad. So, now is the time to run an ad! Er - what ad? Just what are we selling? WHO IS OUR LOGICAL CUSTOMER? Ads are generally written 4 months before they appear. Do we promise that HALGOL will then be available? (We are more careful in these matters than some others. The 'Artificial Intelligence Laboratory promised for spring '84 in the early PDQ ads is not available and no one knows when it WILL be. UNIX has been promised for the SAYBROOK board since before Noah's flood.)

Will an ad pay for itself? Should we logically advertise if we believe an ad will NOT pay for itself? How do we write an ad which keeps 50,000 rock-shooters from checking the bingo card at a cost to us of $3 per pop but which attracts the dedicated hacker who is our logical individual, as opposed to corporate, customer?

The cost of the ad itself is not an issue at this time. We can easily pay for four consecutive full page color ads in BYTE magazine if we felt that would do us some good. (If the ads even paid for 50% of their cost in added business, we could run EIGHT consecutive ads, hmmm?) We do not know the answer to these questions, and when we don't know how to do something, our general reaction is not to do it. That's why it took us so long to develop a case - sheet metal experts we ain't!

Since we no longer regard suggestions of how to modify HALGOL with favor, perhaps you could suggest how we should run an advertising campaign? You see, we think we know how to write HALGOL but we DON'T know how to run an appropriate advertising campaign.

MEMORY SYSTEM PRICES:

The recent revelation that A T & T is selling memory for $2400 per gallon (Jul '84 BYTE, p.305) was very interesting. We have some other system level memory pricing to share with you:

Page 6, Column 1

First is US, of course. The difference between a 1/8 gallon GRANDE and a full gallon is $1200. $1200 per 7/8 gallon works out to $1371.43 per gallon. We also sell half-gallon memory expansion boards; bought separately, they cost $795. That's $1590 per gallon. High-end UNIX systems, the ones which cost $35,000 for the first user, generally price memory expansion at $3300 per gallon. The $100K+ engineering workstations such as VALID and DAISY price memory expansion at $6000 to $7000 per gallon. One major U.S. mainframe manufacturer prices memory expansion for its FORTRAN machines at $40,000 per gallon. You can find some IBM PC memory expansion cards with prices per gallon equal to ours but they use 300nsec floor sweepings while we use full spec 150nsec parts. (Incidentally, we read somewhere that the WE32000s used in the A T & T 3B2 systems are not full spec 70C units but fallout 55C units. We wonder whether this applies to the A T & T-made DRAM as well? And we wonder what speed those A T & T memory modules run at?)

So there you have it, folks! You can pay anywhere from appx $1600 to appx $40000 for a gallon of RAM at the system level - meaning ready to use. That is a 25-1 price range!

A FEW FIGURES:

In issue #30, p.22, we figured that three personal computers alone were accounting for 73,000 megabytes/mo of 64K DRAMs at the (then) current production rate. Since then IBM has made 256K the standard RAM and they are predicting at least 2 million PCs to be shipped in 1984. Since RAM disks are still popular on the IBM PC after they have been in the field a few months, these three computers will probably account for an average of 100,000 megabytes/mo. That's 12.8 million 64K DRAMs a month or about 150 million DRAMs per annum. And that is just three of the personal computers.

It seems reasonable that personal computers will account for 250 million DRAMs in 1984. Let's rub that estimate up against predicted production this year. All of the following figures are taken from Electronic's special issue on national and international semiconductor production (14 Jun '84). Please do not be surprised if some of Electronic's figures do not match others - one of the points we hope to make is that it is tough to get accurate figures.

According to Electronics (p.133), 64K DRAM production tripled in 1983 to 340 million units, worldwide. Forecasts for 1984 are in the range of 750 to 800 million 64K units and 20 to 25 million 256K units. Translation: measured in bytes, about 86% of the DRAM produced in 1984 will be 64K units and 14% will be 256K units. Presumably most of that 256K stuff will be produced in the second half of 1984.

Page 6, Column 2

If our guess of 250 million 64K DRAMs to be used in personal computers this year is accurate, then about 30% of all DRAM being manufactured this year is disappearing under the hood of personal computers. As recently as two years ago that figure was under 10%. (In addition to the fact that personal computer sales have more than doubled each year, the amount of RAM usage per personal computer has increased enormously. We did say that 256K is now standard equipment on IBM PCs, didn't we?)

When semiconductor industry observers are accustomed to seeing semi product disappear into consumer goods, autos, instrumentation, mainframes and minicomputers they can easily misjudge something rapidly exploding on the scene such as personal computers. The IBM PC alone accounts for over 250 ICs per unit, including 8 or 9 LSI parts such as microprocessors and peripheral adapters (do not forget that keyboard). So on p.125 an observer notes the "phenomenal growth of the personal computer industry" and then asserts that personal computers "will gobble up 10% of the worldwide semiconductor market". Just 10%, guys? C'MON!

THE JAPANESE PC MARKET:

A lot of folks think that the personal computers which the Japanese try to sell in the U.S. are typical of what is selling in the Japanese marketplace. BIG, BIG MISTAKE! First, a very quick observation of national character: an American will eat hamburger and drive a used car happily if that is needed to permit home ownership. A German will happily live in an apartment if a Mercedes is in the garage. An upper-middle-class Japanese generally owns a gangbuster stereo, camera - and personal computer!

In 1983 the typical upmarket Japanese home computer had an 8MHz 8086 (NOT an 8088), dual 5 1/4 inch drives with a megabyte formatted EACH, 128K to 256K RAM and superb graphics. Why superb graphics? The Japanese language REQUIRES superb graphics. And the megabyte 5 1/4 inch drives, such as the Shugart 765, which are beginning to appear in this country this year were in mass production in Japan last year.

What is selling in Japan this year? Our information is not that current. Infoworld has reported an 8086-based LISA look-alike (with an operating system which is almost certainly not busted in PASCAL). In fact, observers of the personal computer scene have returned from Japan shaking their head in wonderment for the past three years. Why can't the Japanese translate that into success in the American marketplace? THE LANGUAGE! The Japanese (and Chinese) use of ideographs to represent words means that the keyboards, software and (to a lesser extent) the displays are unsuitable for the U.S. market.

Page 7, Column 1

Still, certain trends (e.g. mass-produced high-density floppies, mass-produced 3.5 inch floppies) now start in Japan, not the U.S. So this might be an interesting quote: "...market researchers say 7 out of 10 equipment designs in Japan are being won by 68000 chips. That is a major shift from a year ago, they say, when Intel's 8086/88 chips [8088 = for U.S. market] were winning designs on the strength of their use in IBM's PC. Motorola believes its second-source strategy is beginning to pay off in a big way. It has a half dozen second sources - one of which is in Japan..." (14 Jun '84 Electronics, p.133)

If that paragraph above surprises you, you should be ashamed of yourself. The Japanese superb graphics require a superb processor and in 1984, the 8086 ain't superb. (Graphics is an almost infinite sink of computing power.) This brings us to COMPUTERGATE, in which the Japanese feel they were unfairly entrapped. (Remember Hitachi/IBM? The Japanese do!) The Japanese are not friendly-inclined toward IBM and things IBM right now. One of the "things IBM" is Intel!

Like everybody else, the Japanese know that Intel is producing limited numbers of microprocessors (the 80186 and 80286) which are superior to the 8086. Like everybody else, the Japanese can see that lots of folk who have designed in those chips are (inevitably) being hurt badly by Intel's allocation policies. Like everybody else, the Japanese can see that Intel probably won't be able to make enough 186s and 286s in 1985, either. AND EVEN MORE THAN EVERYBODY ELSE, THE JAPANESE WORRY ABOUT IBM INFLUENCING INTEL'S PRODUCTION ALLOCATIONS!

If you had been paying attention, you would have realized that it was inevitable that the Japanese would turn away from Intel for their upmarket personal computer products. The only question is whether they would turn to the Z800X, the 16032 or 68000. AHEM! Can someone tell us which of those is made in Japan?

We've already predicted that the 68000 will zip past all those warehouses full of 8080 code three years from now. That's in this country. Looks like that's going to happen in Japan a year or more sooner. Well, in the personal computer marketplace LOTS OF THINGS ARE HAPPENING IN JAPAN A YEAR OR MORE SOONER!

(We're presenting COMPUTERGATE as the Japanese see it. And we went graverobbing to get our picture of the typical upmarket Japanese home computer of 1983. Right, Julain?)

INTEL, XENIX, GATES:

We have in front of us a two-page advertisement by Intel. It features a very large blow-up of the 80286

Page 7, Column 2

chip-carrier package and a very small photo of Bill Gates. After praising the "XENIX 286 Operating System" and its "UNIX enhancements from both A T & T and U.C. Berkeley" (but without ever revealing that XENIX is based on obsolescent UNIX System III, not System V), the advertisement concludes:

"And don't be surprised if you hear shouts of 'Eureka!' coming from your product development team."

Let's see how the other company departments react to the news that an Intel processor has been selected: Ah! The purchasing department is over there: Hi, guys! Congratulations! Your product development team has settled on the Intel 286! What do you mean, "Oh, dreck!"? Look, will you turn the other way? You're going to splatter some of that stuff on our shoes! Oh? You say it's easier to get actual shipments of hens' teeth than Intel 186 and 286s? Hmmm.

Well, it does seem that the purchasing department is not overjoyed. Let's check the comptroller's office here... What in the world are you doing? Yes, we can see that you're jumping out the window. But you're on the ground floor! What's that? You figure jumping out the ground floor window 25 times is equivalent to jumping out the 25th floor window once? Come on, it can't be that bad! Why, your company's development team... Oh, you heard already. Surely you exaggerate, sir! There must be SOMEBODY besides Intel and IBM who is making money off the 186 and 286 CPUs?

...That comptroller must be nuts! Let's check into the production engineering department here. Uh, you there, sir... what do you mean, go away? Surely you can take a moment to chat? No? Just what IS it that you are working so furiously on? Oh, updating your resume. ALL RIGHT! WE'LL GO! You needn't be so rude... Gee! It seems that not everyone in the company is happy about the Intel 286 being selected by the product development team.

In all seriousness, folks, it is not clear to us why ANY company (excepting IBM and perhaps A T & T) would select an advanced Intel processor for an important product after a thorough and thoughtful review of what has transpired in the past six months.

(A T & T can afford as many antitrust lawyers as IBM.)

BAIT AND SWITCH:

This is a disreputable and mostly illegal sales technique. The idea is to offer something desirable for sale at a devastatingly low price. When you arrive at the store, it turns out that the last of that item

Page 8, Column 1

was just sold ten minutes ago - but while you are there (with money in pocket, naturally) the store just happens to have on hand this other incredible bargain...

Just plain "SWITCH" is more common and much less illegal. You walk into a store and the salesperson tries to switch you to another product, usually because there is more profit in the other product. Just the other day a salesman at the Home Computer Center in Tustin made a strong effort to switch us from an F-10 daisywheel printer at $988 to an Apple badge-engineered daisywheel at $1349. We finally shut the guy up by saying "I would like the F- 10 printer, please" using identical words and intonation during three successive pauses in his sales pitch.

Until recently, personal computer retail outlets could not get enough IBM personal computers to satisfy demand. So if you really insisted on a genuine IBM PC they had to add your name to a waiting list. Nobody likes waiting lists, least of all the retailer, because the retailer cannot get his hands on your money (the object of the game, after all!) until much later. And for every additional name on the list the wait gets longer and less desirable for both parties.

It should come as no surprise that, until recently, if you were looking for an IBM PC you would get a "SWITCH" spiel that would make the one we got over a printer look half-hearted and tepid. If the switch was successful, the waiting list grew no longer, the retailer got his hands on your money NOW and you, of course, got your hands on a personal computer (PC clone) NOW. Maybe the $499 Lotus 1-2-3 package you bought the next week wouldn't run, but what the heck, you saved 10% and didn't have to wait, right? And you helped the retailer pay his bills that month.

The point, of course, is that the shortage of PCs was an extremely important contributor to the success of whichever clone that store carried. Now there is no shortage of genuine IBM-brand PCs. You want to buy an IBM PC, sir? Very good! How many would you like to take home right now, sir? And now those clones are not doing so well.

We add the recent entry of I T T to the clone market, the very recent entry of the badge-engineered A T & T clones and the very recent second-wave entry of Japanese PC clones and what you see is clones from horizon to horizon. The distinguishing characteristic of each of those clones is that each is 10% cheaper than a genuine PC. In the meantime retailers are so flush with (excess?) IBM PC inventory that 20% discounts are not uncommon...

Page 8, Column 2

AT & T, I T T, NEC, Fujitsu, etc. are going to survive the clone wars because the profitability of their clones is irrelevant, given the scale of the company. Compaq, with a deserved reputation for quality, saw its profit margin drop to 5% in the first (calendar) quarter this year, and they had to cut their prices after IBM introduced its portable. It will be interesting to see whether Compaq can finish the second quarter using black ink - and Compaq has no massive corporate backup. If any of the other 40-plus clones are going to survive, they must be planning to do it with mirrors.

LOOK! UP IN THE SKY!

Why, it's SUPERCLONE! We'd recognize the red underwear and the yellow cape with blue piping anywhere! SUPERCLONE says, belay that sad-sack obsolescent 8-bit instant antique which IBM is pushing. Step over here and see what a REAL computer looks like! SUPERCLONE's sales pitch is not that it is identical to the PC but that it is SUPERIOR! The Tandy 2000 is the most visible of these new boxes, especially if you have seen their full-page ads in the Wall Street Journal. The most recent ad we saw featured SUPERCLONE outdistancing other competitors (including Big Blue) by a wide margin while running SUPERCALC(3) or Lotus' 1-2-3. Yes, the Tandy 2000 now runs 1-2-3. In genuine 80186 assembly language, NOT busted in C the way 'Fat Mack' will 'run' (ooze?) 1-2-3 next year.

Although the number of 80186 SUPERCLONES Tandy can sell are limited by Intel's production, Tandy seems to have a larger allocation of 80186s than anybody else, including poor Convergent Technologies. The Tandy 2000 is genuinely a superior machine to the IBM PC. Its floppies have double the capacity. Its CPU is, on average, about 2.7 times faster. Its color graphics are so superior in so many ways to the IBM PC that meaningful comparisons cannot be made. What the Tandy 2000 needs is software and now it is getting that. It now has (and we now have) MultiMate, the Wang work-alike wordprocessor. It now has (and we now have) Lotus' 1-2-3. In the new marketplace one does not need LOTS of software support, just support by the right packages.

The SUPERCLONES are going to drive the final nail into the coffin of those 40-plus clones (as if another nail was needed). They may also pressure IBM into replacing the PC sooner than IBM would like.

IBM MAKES MISTAKES,

and under pressure from the SUPERCLONES they may make another one to hang on the wall alongside the corpse of PCjr. (You DO understand that each SUPERCLONE represents a lost sale for an IBM PC, don't you?)

Page 9, Column 1

Industry speculation is building over IBM's next move. We agree with most pundits that IBM will move away from an open architecture, as Apple recently has. (Why let other folks make money off hardware add-ons?) We don't know what IBM is going to do but we sure like the idea of IBM being forced into responding to competition!

The above is NOT inconsistent with the paragraph above the headline "BAIT AND SWITCH". Tandy already HAS its allocation (if not as large as it would like). That allocation is relatively generous because Tandy placed its order early. We are talking about a company getting in line for 80186 deliveries NOW! We don't think that would be a smart move at all.

POOR SUPERCLONE:

IBM is not the only company that makes mistakes. A couple of those 80186-based superclones, whose obvious strength is that they are a BUNCH faster than the IBM PC, are being marketed (as we see it) as game-playing machines. One of them, Mindset, has just had the dubious honor of becoming the first maker of 80186-based machines who has more 80186 chips than computer sales, and has had to lay off and cut back. Who wants to pay three-piece-suit prices for a rock-shooting machine, 80186 or no 80186? There is a lesson there...

68020 AFTERTHOUGHTS:

A few pages back we explained how a 68020 system would necessarily be different from our existing 68000-based products, and hence incompatible with all of our graphics boards, etc. An inherent assumption in our arguments was that folks would naturally want to use the 68020 configured for the highest possible performance, meaning as a real, honest 32-bit processor. Actually, the 68020 can also be configured to operate with a 16-bit data bus with some degradation of performance.

Early information accompanying the announcement of the 68020 indicates that only a 12% performance degradation would result. That seems low to us - but it is undeniable that a 68020 with a 16-bit bus would be a great deal faster than a 68000 at the same clock rate. Not only that, but it would be possible to build a 16-bit 68020 board to be compatible with our present 6.5 by 15 inch form factor and with our existing expansion bus (with addressing limited to 16 megabytes). And that would provide a higher performance CPU to support our graphics boards...

Only a dedicated follower of Mammon would be so impure as to introduce a 16-bit 68020 board. Let's see now: do we know such a person?

Page 9, Column 2

BACKING WINNIE:

A couple of issues back we brought you the poignant yet true story of a company "backing up" their Winchester data faithfully only to learn after a disk failure that their backup floppies, for the most recent six months, were unreadable! This suggests that, in addition to having backup, there must be a way of CONFIRMING that the backup is really a backup.

It has already been proved that folks will not back up Winchesters if backup is not easy. The obvious (in retrospect) corollary that folks will not CONFIRM their backup copies if that is not easy is not yet well known in the industry.

Take, for example, the increasingly common practice of using a streaming tape cartridge (or cassette in some cases) to back up a Winchester. Congratulations! You have just backed up your entire Winchester and it did not even take very long! Uh - it IS a good backup copy, isn't it? We mean, the recording head didn't fail six months back, did it?

Hmmm. Now, how do we check out a streaming tape cartridge? One with, say, 10 megabytes of data? We asked that question and it turns out that the answer is, file by file! With a STREAMING TAPE CARTRIDGE? Rewinding the tape cartridge for EACH file? REALLY? Sounds to us like excellent motivation to never check the backup!

To us, backup is a removable disk that is reliable. Unfortunately, all the feedback we have received indicates that the removable Winchester drives are not, uh, 'perfected' yet. That IOMEGA floppy disk with the $70 10 megabyte removable cartridge looks better and better. (We have discovered that the error detection and correction is handled in software by a 6800 microprocessor. Not 68000, 6800. So the disk is not so fast when the error detection and correction is switched on, although it is still faster than a floppy. With the error detection and correction switched off, the IOMEGA is nearly as fast as a Winchester - but these pesky bad bits sometimes crop up...)

What the individual should logically want and need is a cheap, convenient method of backup which also provides convenient CONFIRMATION of backup. A company such as Digital Acoustics should logically want the same thing, only real cheap is not absolutely necessary.

Did you know that Tandon's 10 megabyte half-height Winchesters are now going for $325 each in large OEM quantities?

Do you realize that, absent backup, Winchesters are time bombs?

Page 10, Column 1

WE GET MAIL:

"...about Blue Sky N: I couldn't fully explain to [Diversi-DOS author Bill Basham] why you wanted to embark on this madcap scheme, because I don't understand it myself." Ralph D., Mt. Laurel NJ

"I liked your explanation of the upcoming HALGOL DOS. It would seem that you have all the details worked out and only need to implement them." Robert L., York, PA

"I would like to state some arguments against Blue Sky One... I recommend that you go with ProDOS, which will run pretty fast on a hard disk, and which will provide a satisfactory (and general) solution to the capacity problem." John S., Silver Springs MD

"After reading newsletter #32, I set out writing you a lengthy letter trying to convince you to use ProDOS as a base for your 68000 operating system. However, after two pages or so, I went back to the newsletter to read again what you had said about the advantages of your system.

"Suddenly what you said seemed to make more sense to me than all that I had been writing so far, so I decided to scrap that letter and start over again. I think there just is not much one can say against reading a whole diskette in around ten seconds, compatible or not. I hope it will not be exceedingly difficult to write a transfer program between the two formats.

"I don't know about the timing of the refresh interrupt on the Grande boards, but I think there could be a possible pitfall with it when accessing the disk from the 68000, as far as I can understand your scheme.

"Did you decide on whether to go ahead with Blue Sky One yet, or will you keep it on the backburner for awhile?" Thomas W., Munich W. Germany

We get decidedly mixed responses to Blue Sky One, Thomas. In addition to the speed, consider that we cannot provide source code for ProDOS but we could provide source if we write Blue Sky One. What the heck good does it do to make HALGOL source available if you can't get source on the operating system?

Also, permitting that mentally defective 6502 to perform an important part of a 68000 language and operating system is against our religion. If you do that, you have a 6502 system, not a 68000 system - and most of you ALREADY have a 6502 system! We'll let the 68000 make all the decisions, thanks.

Page 10, Column 2

Next we have a very important philosophical question: in addition to a vanilla host computer, it is necessary to buy what to run HALGOL? A 128K DTACK Grande? Yes. $60 ProDOS? WHY, if we can provide a free substitute that runs faster? Dual Iomega 5 or 10 megabyte "floppies"? At $3500? No. Can we afford the Iomega drives? Yes. Would using them help us develop HALGOL? Yes. It would also inevitably cut HALGOL off, at a practical level, from those #%&*@! 140K Disk IIs. And if we are going to distribute HALGOL source, it is going to have to be distributed on Disk II floppies or our audience for HALGOL will be reduced to five people.

We most certainly prefer 80 columns to 40 but when we mentioned supporting what we think is the best 80 column card out there a reader (naturally) protested that his 'Grumix XJ77' attachment was incompatible with that card, so naturally we cannot support that 80 column card (?). Besides, if we had an 80 column card we would (naturally) USE it - and then none of the HALGOL demos would run on a 40 column Apple or on some other 80 column cards (no, we are NOT going to support YOUR brand of 80 column card). See the problem? Do you see that Blue Sky One is only one facet of that problem?

A summary: 1) Blue Sky One is the fastest option 2) It is the cheapest option 3) It is a 68000, NOT 6502, program 4) We can give away the source code if we want - and right now, we want.

As mentioned elsewhere, we have been locked into hardware design lately. We would LOVE to work unmolested on various aspects of HALGOL software for 2 to 4 months. But since we can't do that (gotta pay the bills), Blue Sky One is on the back burner.

It is absolutely NOT necessary for the Grande to shut down and do a refresh cycle. In fact, it never has to shut down at all provided that each of 128 rows in each DRAM get addressed at least once every 2000 microseconds. And it is NOT necessary to address all of those rows at one time. In addition, the Apple Disk II interface only reads (or writes) one byte every 32 microseconds, an eternity for a 12.5MHz 68000. And the refresh interrupt is a priority 4 interrupt and hence can be masked by software... That should be enough of a hint for you, Thomas, and for some of our other assembly language hackers - FNE

"...I'd like to reiterate again my suggestion that you offer a 6502 add-on processor board for the Mack. Maybe it could be ready next April 1? ...back when I subscribed (#27) Hal Chamberlin assured me this was a tiny, obscure newsletter in an 'out', quiet backwater. If just anybody reads it..." Rafael S., Warminster PA

Page 11, Column 1

Somebody else made that 6502/Mack suggestion but we forget who wrote first. TINY? Obscure? Quiet backwater? Why, we don't even accept new subscribers anymore unless they are listed in either Who's Who or by Dun & Bradstreet! We'll make an exception for a Nobel prizewinner IF the prize is in the physical sciences. Now, about the incredibly complex (and utterly incomprehensible) schematic you sent us which purports to be your attempt to mate the DTACK board to an Atari:

A certain CADD house busted a half-megabyte Grande and gave it to one of their programmers. The programmer un-busted the board (a few hour's work with a soldering iron) and then decided to interface the board to a Commodore 64. He did it by getting one connector to plug the 6-IC interface into and another connector to plug into the 64. Then he wired the two connectors together. Voila! It worked! Honest!

Now, about that miserably complex Atari interface of yours: perhaps you should ask Hal Chamberlin to help you design in a few dozen more TTL chips? "Out", quiet backwater indeed! @%*&#! - FNE

"I will not give you any more 'advice' on However HALGOL as you seem to be getting enough. I can't resist pointing out that when you make your mythical case that it absolutely must be size compatible with my and everyone else's power supplies, while stacking conveniently with our disk drives? You do realize that we will all insist on this yes?" William H., West Linn OR

We appreciate the advice, William. Now, will you go over there and stand in a straight line with Rafael S. and Hal Chamberlin so we won't waste bullets? - FNE

"On the issue of commercial application programs being written in HLL vs assembly: The primary claim of the HLL people seems to be that the program is easier to read and easier to maintain. But just who benefits from these attributes? Certainly not the user... Even if it is easier to develop the application in HLL, I have seen no evidence that the resulting program is any less expensive than one written in assembly. Any price difference now between MBA and 1-2-3 (to name an example) is due to discounting of the former to make up for its poor performance." Hal C., Wake Forest NC

Right, Hal. Now get back in line - FNE

"MICRO is trying to take a giant leap forward into the 68000 world. While not abandoning the 6502 and 6809, we feel that our readers are now ready to move forward

Page 11, Column 2

to better machines. It also seems right in that the number of 68000-based computers available is growing rapidly and the pricing is becoming more 'reasonable' for the average reader.

"I am writing to you in the hopes that you might be able to direct me to individuals, companies, clubs, groups, ..., that would be able to provide articles on building 68000-based systems, including the fundamental software monitors to make them useful. The ideal situation would be a multi-part build-it tutorial in which the reader could build his computer/expansion board step-by-step while developing software and system knowledge on the 68000." Robert M. Tripp, Editor-in-Chief, MICRO

Any of you who would like to help spread the 68000 gospel in a mass-market publication should write MICRO at P.O. Box 6502, Chelmsford MA 01824. Bob, we have looked into the Sinclair QL machine as well as we can without actually getting our hands on one and that machine seems to be ideally suited to MICRO's audience. It is a superb first computer and cheap enough for a lot of your readers to purchase solely as a training vehicle for the 68000 (in this case the 68008). As far as we can see, its BASIC is not busted so it should be faster than an Apple II, unlike the $2500 Mackintosh. We would think Sinclair would want to cooperate with your publication. Good luck! - FNE

[The very next MICRO after this letter had an article comparing the 68000 to the 6809. The conclusion was that the 68000 was VASTLY superior to the 6809 and therefore should not EVER be used in the home! @%*&#! Bob, do you read the articles you publish? - FNE]

"...I am a geophysicist, a member of what I consider an honorable profession and a rigorous natural science. Truth is an ideal based upon solid proof. Any attempt to distort the "nasty old facts" for pecuniary or other reasons is unprofessional and repugnant to me. (Does this mean that I'll never work for the government?)" John W., Austin TX

Not until you change your attitude, John. Question: how does someone with at least 25% of your regard for the truth draft copy for the next DTACK advertisement? GEOphysicist? SOLID proof? - FNE

"...I have the Phase Zero assembler - first version - and the newsletter said that we should return them for updating. However, the material which has been appearing in DSEx indicates that letters, etc., go there and disappear. Is Phase Zero operating in a businesslike manner or no? Is it safe to return my disk for updating?" Rowland B., Huntsville AL

Page 12, Column 1

That's a tough question. We have not received any complaints from anyone who has sent them a check for $95. But we ourselves had grave difficulty getting an updated copy of their assembler. We sent in OUR 'first version' original diskette (for which we paid $95, by the way) with an updated version of our math routines for HALGOL in source code for their benefit (they already had our permission to use it). We then had a real bitch of a time getting the ungrateful bastards to update and return the disk!

We phoned them several times, each time asking for the update. On each occasion Dan Davidson, who is Phase Zero's president, assured us that he would send it right away. He didn't. Finally we got P.O.'d after three highballs one evening and called Dan at his home with our, uh, complaints. We then received (over four months late) our update promptly! Considering the free advertising we have given Phase Zero, we were frankly disappointed with them.

Apparently others (who do not have Dan's home phone # or who are politer than we are, especially after three highballs) have received even shabbier treatment than we have. Whether Phase Zero is businesslike or not we dunno (WHAT business?), but they ain't very humanitarian. Dan: we offer you and Phase Zero free space here for any rebuttal you might like to make.

"DTACK GROUNDED must be the yellowest piece of journalism since the Hearst heydays or PC World, whichever came last. Nonetheless, it's great fun and makes even Doug Clapp sound like a double-E major." Rick B., Editor, Club Mac

Hey, don't knock Doug Clapp! In his latest column (as this is written) he did a superb job of identifying the three most important things about RISC CPUs: 1) they have an awful lot of internal registers 2) if you keep a lot of internal registers working hard then you potentially have a very fast machine 3) paper machines such as RISCs always, repeat always run faster than real machines such as 68000s. Lots of folk with EE sheepskins cannot get those three facts right. Did you notice that we corrected your misspelling of 'heydays'? Think nothing of it, just a professional courtesy from one editor to another - FNE

(We assume that our readers with Macks know about Club Mack, the national Mackintosh user group. If not, call (303) 449-5533 in Boulder, CO.)

"Asked SORD [Japanese outfit] for information on M68 and nearest dealer - 'How about you?' - AWK! Lifetime (so far) Peripheral suggested 'TORO POO-POO' for stronger language." Ron D., Minneapolis MN

Page 12, Column 2

Ron, we have been hearing that for over a year now re the SORD M-68. Can't use 'POO-POO'; some of our readers have very tender ears (we hope they have forgotten the meaning of 'Felgercarb'!). Suggest immediate and sincere apology (+ long-stemmed roses?) to lifetime (so far) Coprocessor. Your lifetime (so far) peripheral is your ahhhh, remember the brand name of your FNE's very first home computer?

"While I recognize (and often admire) your freedom in the use of the English language, as she are spoke in America, I am distressed at an obvious misspelling in issue #33, page 4. This is particularly appalling since the word involved is a terse, much-used anglo-saxon term which you should certainly have learned to spell by now. To wit:

     Your spelling     Correct spelling 
       @%*&#!               $#!+

"Please try to get it right next time. Note that it works best at the beginning of a sentence rather than at the end, e.g. 'Oh, $#!+, I forgot to write-protect it ...'

"(A T & T's) marketing of the 3B series computers is so poor it's laughable. They have vivisected UN*X, their baby, into tiny pieces '...so the user won't have to buy what he doesn't need.' Sounds nice, until you realize that the C compiler package DOES NOT include an assembler or loader. Obvious logic: well, if you program in a High Level Language, you don't need an assembler, right?" Edward N., Austin TX

What's worse, Edward, is that A T & T has un-bundled their own UNIX so far that (as they admit) much of the limited UNIX-based applications software will not run because the needed utilities, considered a necessary part of wonderful UNIX, just aren't there! (Aug '84 UNIX REVIEW, pages 19 and 44) Missing utilities include tar, more, spell, nroff, troff and man. Translation: A T & T is not even providing what most folks call 'UNIX' with their 'UNIX' computers! Isn't that a bunch of @%*&#!? - FNE

"...I would strongly prefer that my name and address not be included on any junk mail lists of any description furnished to other persons or firms, if you follow the disgusting practice of selling the names and addresses of your subscribers." George C, Marietta GA

As usual, the rumors aren't quite right, George. We would NEVER sell our subscriber's names and addresses! Now, to business: how is your health? Do you have good teeth? Are you able to perform heavy labor for 14 hours a day? - FNE

Page 13, Column 1

TEETERING HALF-BITS:

A couple of issues back, we cited MMI ap note 111 as support for our assertion that some partial products are ignored in commercial computing. If you do not have access to that ap note, perhaps you have access to 14 Jun '84 Electronic Design p.314. Note that the figure shows that six of the sixteen partial products generated by a 64 X 64 multiply using 16 X 16 hardware multipliers are ignored! ("The six partial products shown in color make no contribution to the 64 MSBS.")

Let us discuss the errors due to our deletion, in the latest HALGOL 62-bit floating point multiply, of one of the nine partial products: 99.997% of the time the result will be the same. 0.003% of the time the result will be in error by 0.003% OF ONE LEAST SIGNIFICANT BIT (LSB) (as compared to an unrounded result).

Please understand that EVERY floating point multiply has, after rounding, an error of up to half a least bit (50% of a least bit!)! In our opinion, spending appreciable time to eliminate an error of 0.003% of a least bit is absurd!

The error due to ignoring six of sixteen partial products when performing a 64 X 64 multiply can be almost as great as 0.01% of a least bit! Tsk! This brings us to the Floating Point Systems FPS-5000 array processor, which is based on the Weitek single-precision floating point chips. Division (more accurately, the calculation of a reciprocal) results in a least significant bit which is in error half the time (!). In effect, the error is exactly one bit greater than the "error" due to rounding when every precaution is taken to achieve the greatest accuracy.

In exchange for one least bit error (instead of just half a bit) during divide one gets 5 megaflops per chip. That seems to be a reasonable tradeoff to us.

Remember that there is no such thing (in general) as an errorless floating point operation! There will ALWAYS be an error, and that error will be as great as half of a least bit. We assert that it is worthwhile to deliberately increase the maximum error from 50% of a least bit to 50.003% (or 50.01%) of a least bit if a noticeable time savings can be achieved.

(We are aware that some folks will disagree with the above argument. That does not surprise us; we have received several letters asserting that it is perfectly O.K. for a 68000-based machine such as Mack to run more slowly than Applesoft on a 1MHz 6502 [as it does, using Microsoft's Mack BASIC]. There are times when we wonder if we live on the same planet as the folks who approve of slow 68000 software.)

Page 13, Column 2

YATES STRIKES AGAIN!

Jean Yates is the UNIX guru who forecast sales of 600,000 UNIX boxes in 1983. After 1983, when it became obvious that that forecast was, um, a tad optimistic, Yates tried to, er, cover her fanny by asserting that there were a lot more UNIX boxes sold than folks recognized. A reliable source has told us that she claimed, for instance, that 70,000 UNIX boxes were sold in 1983 by manufacturers who sold fewer than 100 boxes! (That would mean about ONE THOUSAND FOUR HUNDRED vendors selling between 1 and 99 UNIX boxes! Can't ANYBODY in the UNIX world do simple arithmetic?)

Jean may have just topped herself. She has predicted that "UNIX will emerge as the only operating system offered by IBM as a prelude to an A T & T buy-out of Big Blue." (Jul '84 Microsystems p.12)

ELECTRONICS STRIKES (AGAIN?

In the 28 Jun issue, p.6, the editor makes an enormous and obvious mistake. Discussing bargains that you can buy, we find "...For less than $1 million, you can pick up an array processor that operates at a screaming 331 megafiops, outpacing the 160-megaflop Cray-1 supercomputer, which sells for from $5 million to $10 million." GROAN! Look, that is a reference to the Floating Point Systems 164-MAX which is going to be a very fine array processor when FPS gets around to making it a year or so from now. It even does double precision arithmetic, something which Electronics fails to note.

But you most certainly cannot purchase the 164-MAX now.

BUNNELL STRIKES AGAIN!

We really should stop picking on David Bunnell. He's finally selected a reasonable likeness to go with his publisher's column in PC World and we even agree with him occasionally. If he would only stop making such obvious mistakes...

In the July issue (pp13-14) Bunnell asserts, presumably with a straight face, that Symphony will be Mitch Kapor's SECOND attempt to grab the brass ring (make a bunch of money), his previous successful program having been 1-2-3. Sigh. Feb '84 (IBM) Softalk, p.28:

"Kapor delivered VisiPlot and VisiTrend in 1980, and the products were introduced in April of 1981. They proved to be Kapor's first big break: In a period of only six months, he raked in almost half a million in royalties."

While you have that issue of (IBM) Softalk in front of you, let us turn to page 32 to find evidence supporting

Page 14, Column 1

our feeling that the 68000 is being deliberately subverted by saddling it with inappropriate high-level languages while inferior Intel processors are awarded assembly language with superior resulting performance:

(Mitch Kapor speaking) "'We decided both to write for the PC and to optimize for it.', he recalls. 'Everybody else had eight-bit software, and it was going to take a long time to redo it.' And even then, it would still only be eight-bit software pasted onto a sixteen-bit machine, certainly NOT the best way to use resources.

"Kapor and Sachs were quite serious about optimizing their new baby. It's a little-known fact that what was later to become 1-2-3's spreadsheet was first written in C; that version was both too slow and too large, so Sachs rewrote it in assembler. (Versions of 1-2-3 for other machines, such as those based on Motorola's high-performance 68000, are written in C.)"

And there you have it, folks: the 68000 has high performance, so it is perfectly all right to write programs in a way that would be too large and TOO SLOW on inferior processors such as the 8088! WHY? WILL SOMEBODY TELL US WHY 68000 SOFTWARE HAS TO BE DELIBERATELY BUSTED IN A HLL?

(Perhaps everyone else is correct and we should rewrite HALGOL in C so it can be slower than Applesoft like all the other 68000 BASICs we have seen?)

APPLE STRIKES AGAIN:

Apple management, that is. Apple is so enamored of its new products, Mack and the cute little IIc, that it neglected to manufacture enough of its cash cow, the IIe. In yesterday's L.A. Times saturday sports section there were very few ads for IIes. That's because most stores were sold out of IIes. Honest.

On the other hand, one can find the Mack and the cute little IIc in stock in any store you might check at random. (There might be a shortage of the cute little monitor that goes with the cute little IIc.)

Judging by the Mack and the IIc, both of which are closed systems obviously intended for non-hackers, Apple management must believe the time has passed for hacker machines like the Apple IIe, which is the latest version of the greatest hacker machine of all time. So they didn't build enough of the machine which is generally recognized to be the company's cash cow.

Does anyone know anybody who has actually purchased a LISA II at a retail outlet? We don't. The few who have second drives for Mack have discovered that all the software assumes you have only one...

Page 14, Column 2

A T & T STRIKES ?

We would love to tell you something about A T & T, but our facts and our opinions seem to be exactly the same as everybody else's. Should we give you the hot news that the sun is expected to rise in the east tomorrow morning? We DID manage to tell you that the Olivetti PC clone that A T & T is going to badge-engineer today (this is the morning of June 26) is in turn a badge-engineered Corona Data Systems PC clone, something the rest of the media has not yet reported. Did you notice? (#33, p.12 under 'PCs FOREVER?')

While you have #33 out, turn to page 25 and, in line #128, change 'STK' to 'S2'.

[LATER: The L.A. Times business section gave the A T & T clone announcements the coverage they deserved - one tiny paragraph!1]

AUGHH

There are about 35 folks who have had their hands on HALGOL release 2. Perhaps they are an informal Apple User's Group of HALGOL Hackers? (Jeff H. hasn't ripped that one off yet!)

PHANTOM 10MHz 16081s:

We aren't the only folks who have been picking on Nat Semi over its 10MHz 16081s, which seem to be commonly available everywhere ELSE except wherever you happen to be! Another outfit that has been picking on them is Rockwell (formerly Autonetics). On the reasonable assumption that a big outfit like Rockwell cannot find its hind end using both hands in broad daylight in less than 18 months, a Nat Semi representative handed a Rockwell type a nice 24-pin package marked "NS16081D-10" and departed, chuckling to himself.

While the Nat Semi type was still chuckling, the Rockwell type climbed into his car and drove over to Digital Acoustics. "Would you check this part out for us, please?" he asked. The next morning we reported to him that the part did not work at 10MHz, it did not work at 8MHz and it did not even work at 6.25MHz. Congratulations!, we continued, you have a MARVELOUS paperweight! (The device was OBVIOUSLY defective - it ran cool!)

A certain Nat Semi type is about to learn that Rockwell can find its fanny about 17.97 months earlier than he had thought...

[We remain convinced that there is a laboratory somewhere which has one or more functional 10MHz 16081(s). We do not know what kind of cooling system is used.] [LATER: See also p.24]

Page 15, Column 1

OLYMPICS:

The good-guy part of the world is holding a half-Olympics in our area pretty soon. That's fair; the bad-guy part held THEIR half-Olympics in Moscow four years ago, remember? About 99.9% of the local populace wish the whole thing would go away. The Olympics planning committee want 99.9% of the local populace to park their cars and stay home so as not to interfere with needed Olympic transportation. For instance, runners will warm up at UCLA's Drake stadium and then be bused to the Memorial Colisseum across the street from USC. With any luck, the trip will take less than a week...

Oh, yes: the buses. The buses that the warmed-up runners will ride and the reporters covering the Olympics will ride: they are school buses! School buses, for the benefit of our Nome, Alaska readers, do NOT have air conditioning. Have you ever been in a bus loaded with people at 2PM on a hot August Los Angeles day, parked on a jammed freeway? A bus without air conditioning? Or 'convenience facilities'? Those runners are going to be delivered more warmed-up than the Olympic planners expected. And we sure hope the reporters don't drink too much beer...

However, we don't care anymore. The DTACK GROUNDED 51-man squamish team has been disqualified (sniff!) from the Olympics for excessive Heineken consumption. While it is true that the excessive consumption occurred in the wee morning hours, the assertion that the location was the dormitory of a certain women's college is almost certainly untrue. (How many of you remember 51-man squamish? Your FNE is ideally suited for a certain position on a squamish team, yes?)

The greater DTACK community is not completely shut out; we understand that Jeff H. is one of the top seeds in the freestyle claim-jumping event.

COMMODORE WATCH:

When you replace Jack Trameil with a long-time garment worker executive, you can guess what might happen to what had been an enormously dynamic and successful company. When you additionally discover that the new executives chased out (or failed to retain) the highly skilled, tightly knit team which had been running the company you KNOW what is going to happen. Commodore isn't shipping 400,000 C-64s a month internationally because the dealers have cut back their orders. The dealers are ordering fewer C-64s because they can't get enough floppy disk drives. The shortfall is about 275, 000 floppy disk drives a month! You can't just walk into a store off the street and buy 275,000 floppy drives a month, you know. Some garment industry execs just might not have learned how it's done...

Page 15, Column 2

APPLE WATCH:

Apple and IBM are the only really solid, successful players in the game right now. Add Commodore and you have identified the only three players who can control their own destiny. The rest of the industry consists of the various gulls, tick birds and remoras feeding off IBM. Every time IBM sneezes, three of those parasites die of pneumonia. (Perhaps we should acknowledge Franklin, the Apple parasite which has just entered bankruptcy.)

We are therefore pleased to note that Apple is doing all right. Their obvious goofs have been minor - not realizing, for instance, that every purchaser of the cute little IIc would want the cute little matching monitor. Not making enough IIes. Not planning on two disk drives for Mack right from the start.

Apple's less obvious goofs - not permitting Mack to travel in any direction other than north, for instance - have not hurt because the other players in the game have made larger goofs. Commodore's goof is spelled 'no disk drives', IBM's goof is spelled 'PCjr' and everybody else's goof rests on the historically disproved assumption that one can live long and prosperously as an IBM tick bird.

We are glad that Apple is around. We are glad that they have adopted the 68000. We wish they would demonstrate the speed of which that chip is capable.

THE DEAD CALM:

Aside from Commodore and Apple, the personal computer marketplace consists of IBM and its gulls, tick birds and remoras. Suddenly, NOBODY in this market is selling computers in any numbers! If this were a yacht race, we would have a bunch of boats with slack sails, sitting dead in the water. Buy an IBM? WHY, when they are announcing a new, more advanced model soon (the rumor goes). Buy a clone? When the clone-maker's profitability has just been torpedoed by the IBM price cuts and you might wind up with an instant orphan? When 90% of the IBM-compatible software won't run? When the 80186-based makers can't get enough chips?

Suddenly, lots of clone-makers find themselves swimming in a sea of unsold inventory (80186 machines excepted). Look, for example, at Eagle computer's warehouses of completed 'PC-compatible' computers (all clones are 'PC-compatible' until you take one home and try to run some IBM-type software). Filled warehouses. Filled with dusty boxes since they are not moving on to dealers...

Just exactly how many finished PCjrs do you think are sitting in IBM warehouses? More or less than 350,000?

Page 16, Column 1

Just as you do not turn on production of 275,000 floppy disk drives a month like turning on a water tap, you ALSO do not turn parts OFF like a water tap. Right now there are many, many millions of dollars' worth of parts streaming into the production facilities of the PC clones but the completed product is not moving out of the warehouses. So the clone-makers are spending money but not collecting it. The books look great, of course - those parts get turned into completed inventory which is carried on the books at wholesale prices. (Anybody want to review pages 9 and 10 of issue #4 of this newsletter?) Unfortunately, books that look great don't pay the bills. That's why Eagle Computer has a creditors' committee these days.

A lot of computer retailers who had been doing well selling PCs and PC clones have just decided not to buy that BMW after all. And the two-week August Mediterranean cruise is being replaced by a glorious, fun-filled week at the Newark, NJ Club Med.

There are a whole lot of folks out there 'whistling past the graveyard'. Those computers piling up in the warehouses look great on the books, and they are (naturally) going to sell REAL SOON NOW (apologies to J. Pournelle). The alternative - to those folks - is unthinkable. Stop buying parts? Shut down the production lines? Why, that's an admission of failure! As in, let's everybody go look for a new job! HORRORS! We can't do that! So let's stack some more completed inventory in the warehouses...

This can go on for as long as investors are willing to throw good money after bad and as long as suppliers are willing to ship parts without getting paid. After that everybody goes looking for a new job anyhow and the liquidators take over the swollen warehouses. This is going to begin soon, like within three months.

If you can buy clone 'A' at liquidation prices, how can clones 'B' and 'C' sell enough product at a price high enough to stay in business? With clones 'A', 'B' and 'C' all in liquidation, how low will the prices have to drop so that the liquidators of clone 'A' can move their product? After the other two liquidators drop THEIR prices accordingly, how can the other 40 clone-makers stay in business?

Those filled warehouses are going to be proved houses of cards, and they are going to tumble down and tumble down soon. And here we are having dropped some heavy hints in past issues about things transpiring in the three-piece-suit personal computer market THIS FALL! Heck, it's not even July yet...

Anybody want to buy a PC clone today?

Page 16, Column 2

AFTERTHOUGHTS:

Like lots of other folks, we have been watching this shakeout 'shape up' for some time now. What precipitated the above writeup was a conversation we had with our local personal computer store proprietor on Jun 28. The next morning we stayed home from work and wrote the above. The day after that, Sat Jun 30, we had lunch with a friend who is very well connected in the personal computer peripheral business (well, a particular aspect of that business). Over lunch he told us stories of horror, doom and despair. Stories of payables stretched out over 90 days by EVERYBODY in the business. Stories of contracts cut in half or even cut to a quarter. Stories of layoffs of 50% or 60% of companies' production and engineering personnel. And ALL of this (the stretched payables excepted) had happened within the preceding three weeks.

After that lunch we went by our local Software Centre to pick up a Peter Norton book on MS-DOS. The latest InfoWorld (16 Jul) was lying on the counter, so we picked that up too. Back at the office, we turned to John Gantz' column and sonofagun! we read the headline 'SHAKEOUT!'! Sigh. It does seem that some other folk have been holding their ear to the ground...

THREE years ago, when we wrote (issue #4) about auditors and inventory that isn't moving we got a WHOLE BUNCH of flak. Now it's becoming routine. In the second part of the Osborne story, on p. 59, we learn that auditors do not like to change opinions previously issued. Heck, that's old hat! Two years back we told you about the hassle at AM International (formerly Addressograph-Multigraph) and its new management which wanted to restate previously audited results to place some big, big dollar losses on the PREVIOUS management and how they fired their auditors when they refused!

John Gantz' comments are right on target: "My own informal survey of the market tells me that potential customers are simply weary of the noise. Their eyes are glazing over. Many are postponing purchases for fear that what they'll buy will be obsolete by the time they learn to use it."

SPEAKING OF AUDITORS:

IBM has auditors, too. 350,000 PCjrs have a retail value of a third of a BILLION dollars! If IBM has 350, 000 PCjrs in warehouses, then they are probably carried on the books at about $210 million. That's 1.5% of last year's total IBM sales and WAY, WAY too much for even IBM's auditors to overlook. Following Apple's lead with the III, IBM can slide by the auditors once by introducing a new keyboard at audit time. But that can only work once. If the inventory is still there the following year...

Page 17, Column 1

Look, folks, facts are facts. IBM has a large inventory of completed PCjrs, and that is a fact. (The exact size nobody knows. The 350,000 is a reasonable guess based on IBM's plan to sell one million PCjrs this year and also that big companies can't turn off the tap very fast, especially when Teledyne is mass-producing jr under contract.) It is also a fact that IBM's auditors are not going to let that inventory slide past two consecutive audits.

Question: is IBM going to follow Timex' and TI's lead and sell off that inventory at 15 cents on the dollar or are they going to take Atari's lead and bulldoze that inventory into the side of a hill? (Better make that the side of a mountain!)

(The last time we wrote something like the above the folks at Apple got real mad at us for some reason. We believe that illegal photocopies of issue #4 were circulated in Apple's legal department. Let's see now: IBM is about forty times larger now than Apple was then... )

NORTON WATCH:

In the IBM PC world, which is becoming THE personal computer world except for Apple and the CBM 64, Peter Norton is THE Establishment Spokesperson. The dozens of PC-related publications are divided into the anointed (those which carry a regular Peter Norton column) and the great unwashed (those which do NOT). Since we now read several PC-related publications, we wind up reading Peter Norton's columns frequently.

For instance, in 24 Jul '84 PC magazine (purchased on 26 Jun) his column tells us just why most programs which run under PC-DOS on the IBM PC will not run on most clones. He even echos some private thoughts we have had on the subject: "IBM's hammerlock on personal computer sales is self-perpetuating, because the best software is written for the PC first and only sometimes translated to other machines later... If the non-IBM micro makers had any sense, they would get together and fund an independent effort to create such a set of driver software, and then give it away. But that's a pipe dream."

Since Peter is a software person and we are a hardware person it should not be surprising to learn that we can disagree. In this same column (and in other columns at other times) Peter simply cannot understand why computer manufacturers will not donate computers to him so he can adapt his IBM PC programs to their machines so that he, Peter, can increase his profits. Ah, Peter: as a hardware manufacturer we can assure you that the quickest way to go bankrupt is to give away your product. We will not tell you what Digital Acoustics tells folks who demand a free DTACK board so

Page 17, Column 2

they can develop some software because we do not permit such language in these pages.

In fact, for making such a ridiculous (from a hardware maker's point of view) suggestion, we are going to reveal that in the penultimate paragraph of that 24 Jul PC column you wrote "hoards" when "hordes" was clearly intended. Tsk! (Hey, WE get picked on for stuff like that; it's only fair that we get to pick back occasionally. )

MACK CRITICIZED !

We will pass along, courtesy of PC magazine, a piece of Tandy Prexy John Roach's keynote address at Atlanta COMDEX:

"Then there is Apple with its 20-page inserts to try to convince you that a black-and-white monitor on a 16-bit computer with 80K available to the software writer with a single disk drive and no numeric key pad or cursor keys is great! If you think about it very much, you'll laugh!"

WHO'S SUCCESSFUL?

If we restrict our comparison to Apple and Commodore, which company has been most successful recently, say in the past nine months? What's that? You say your leg jumped straight out in a knee-jerk reaction as you immediately replied, "Apple!"? Why, of course! It's Apple that is fighting the IBM hordes, right? (That's 'hordes', Peter, not 'hoards'.) It's Apple which is producing the darling of a part of the press media, Mackintosh. And it is Apple which has rung up profits of $20 million in the past nine months while Commodore's profits have been restricted to a mere, uh, $110.7 million... Hmmm...

TRAMEIL WATCH:

We learned, on Sat Jun 30, that Jack Trameil was negotiating to buy Atari, negotiations between Warner and Philips having proved fruitless. (We understand that both Warner and Philips made a sincere effort to reach an agreement.) Our reaction was, betcha Jack buys Atari for half what Philips offered and got turned down!

This is the morning of 3 Jul and according to the morning L.A. Times business section, Trameil just bought Atari - without a single cent changing hands! We think we just figured out why Kindly Uncle Jack has over $60 million in cash AND Atari while we have this newsletter.

Now we will sit back and watch the heads roll as Kindly Uncle Jack reduces the overhead (and deadwood), there

Page 18, Column 1

being no chance that Atari can produce new computers in time for the Xmas selling season. Oh, yes: the dispute which caused Trameil to leave Commodore will not arise at Atari because Jack's son Sam, age 34, is already the president of the new company. Vendetta? Who said vendetta? (Kindly Uncle Jack's last vendetta cost T.I. over $700 million.)

Let us make this unmistakably clear: Irving Gould blocked KUJ from promoting his sons to high levels of management (there was a presidency vacant, remember?) so KUJ now intends to drive Commodore and Gould into the ground using a company (Atari renamed Trameil Technology Ltd.) whose president is one of KUJ's sons. Please note the new name of the company and that its president is Sam Trameil.

According to Jim Strasma, Irving Gould is aware of Kindly Uncle Jack's intention of competing with Commodore and is not worried even a little bit. Gould apparently hasn't noticed that the last several folk who bet against KUJ went home wearing a barrel.

CHRISTMAS vs XMAS:

We were severly criticized when we used the term 'Xmas' a while back. "Christmas", it was explained to us, "stands for Christ. To use the spelling 'Xmas' is to take Christ out of Christmas." We have given the matter some thought and here is our conclusion:

There is indeed a religious aspect to a part of the Christmas holiday, which is a modern successor of the old pagan winter solstice celebration. However, the Xmas SELLING season is clearly related to Mammon, not religious sensibilities. It is now our editorial policy to use 'Christmas' to refer to the religious holiday and 'Xmas' to refer to the accompanying commercial selling season. The alternative, after all, is to exhort the merchants to "get out there and sell one for you-know-who"!

If you disagree with our reasoning, at least give us credit for considering the matter?

ON THE WIND:

If the wind is just right, listen closely and you will hear the death-rattles of 40+ IBM PC clone-makers. The wailing and tearing of garments you hear is from the widget-makers who supplied those 40+ clone-makers and whose accounts receivable are out past 100 days and who have just realized that they will NEVER be paid for all those widgets! The slow, muted sloshing you hear is Digital Acoustic's bank balance, Digital Acoustics not having supplied any widgets to the clone-makers. (Hey! We work HARD on this newsletter! It's only fair that we get to brag occasionally.)

Page 18, Column 2

CIRCUIT CELLAR BACK !

Yep, a genuine Circuit Cellar with a genuine cellar product: chimes which will doubtless play "the Yellow Rose of Texas"! Steve's regular article has recently featured commercial products which were more likely to have been developed in NORAD headquarters under Cheyenne mountain near Colorado Springs than in Steve's cellar. That MPX-16 PC clone and the Trump card (with BASIC compiler) must have each cost as much as the annual budget of Zimbabwe to develop. And then Steve and his presumptive silent partners get tons of free advertising for those commercial projects in BYTE! Yes, we ARE jealous.

The chimes are nice.

UNFORTUNATE FOUNDERS:

The last two of the three founders of Fortune Systems have just departed the company. Tch. Managing the "tenth largest financial placement [$110 million] in United States business history" (Dvorak) doesn't get a guy much respect these days, does it? The departures were described by a Fortune spokesperson as being "by mutual consent". Uh huh.

DVORAK: PURE GOSSIP?

As you all know, this respected journal is a crystalline fountain of knowledge, an alabaster monument to objectivity and a burnished bastion of transcendent Truth! Geez. We explain how to make chips run faster and the attendant problems. How to interpret the 68000 spec sheet timing diagrams. Why Convergent Technology isn't shipping many N-GENs. We explained to Nat Semi that their 16081 wasn't asynchronous like their spec sheet said (they did not know - or believe - that at the time!). We printed a schematic showing how to interface the 68000 and the 16081 and a prose description of the function of all the pins on the 16081. We showed you how to calculate a logarithm using the 68000/16081 combo taking advantage of instruction and operand prefetch. We even warned you, a long time in advance of anybody else, that you should steer clear of UNIX unless you are a full-time computer professional. You will have the Devil's own sweet time locating most of this information in other publications. So what happens?

John Dvorak, for the SECOND TIME, calls our respected journal a "pure gossip rag"! @%*&#! This is in the 23 Jul issue of InfoWorld, purchased on 5 Jul. Oh, well; the rest of the issue contains nearly comparable errors. For instance, the two-page story on Convergent Technology only mentions its vitally-important N-GEN once and NEVER mentions a critical shortage of 80186s...

Page 19, Column 1

HOT NEWS (?) DEPT:

A reader breathlessly asserts that Apple plans to drop LISA II next year. Gadzooks! Just because nobody is buying LISA II? Just because nobody is supporting LISA II with software? Just because LISA's screen is forever incompatible with Mack? Just because Apple is coming out next year with 'Fat Mack'? Zounds! Whatever made that reader think that Apple's dedication to LISA II is less than total?

Meanwhile, the WSJ, 10 July p.6 quotes Steve Jobs as asserting that "sales have gone through the roof"! He declined to give specific sales figures but said current demand for LISA II exceeds supply. Let's see, now: just where is the LISA II production line? Er - there IS a LISA II production line - isn't there?

FIGHT! FIGHT?

We have become involved in a dispute with Jim Strasma, who is well known in the Commodore world as editor of the Midnite Software Gazette, which merged with the Pet Paper to form Midnite. Jim recently told us Midnite is the one-hundredth largest personal computer-related magazine in the U.S. as measured by circulation.

The undisputed facts are that Terry Peterson adapted MICROMON, widely believed to be a public domain program, for use with the CBM 64. This necessarily required SOME original work on Terry's part; there is a dispute between Terry and Jim as to just HOW MUCH original work was involved. Our dispute with Jim is not identical to their dispute, although there is some overlap. We disagree with Jim over what constitutes appropriate use of public domain software and over the validity of HES's copyright of HESMON.

PUBLIC DOMAIN:

Here is how our Webster's New World Dictionary, Second College Edition, defines 'public domain': "1. public lands. 2. the condition of being free from copyright or patent and, hence, open to use by anyone."

Read those last five words again: "OPEN TO USE BY ANYONE".

COPYRIGHT STUFF:

We have had occasion to study copyright law several times during the past twelve years at Digital Acoustics. Our motivation was financial; we wanted to protect Digital Acoustic's property rights. While we do not claim to be declaiming ex-cathedra about copyrights, we are a moderately well informed layman.

The copyright law is a federal law. State and

Page 19, Column 2

municipal courts have no jurisdiction. Because of the enormous body of material being copyrighted daily, congress wrote the law to be as idiot-proof as possible. It is perfectly permissible to copyright something without actually registering it and in fact most copyrights are done in this way (to the enormous relief of the copyright office!). Such a copyright expires in five years.

Anything so copyrighted can be registered within the following five years; doing so extends the life of the copyright. Such registration is REQUIRED before filing a civil lawsuit in a federal court charging copyright infringement.

The point is that the copyright process is idiot-simple and results in a bona fide, bullet-proof valid copyright. The copyright remains valid until it expires or until a federal court (after appeals are exhausted) rules that the copyright is invalid. To repeat, only a federal court can rule a copyright invalid.

The next thing to realize is that there is no such thing as a criminal prosecution under the copyright act. The copyright act provides only for civil suits. In fact, only a particular civil suit is allowed: a civil suit for copyright infringement.

In a civil suit, the suit is initiated by a PLAINTIFF. The plaintiff must have STANDING, that is, a stake in the suit. Example: If DTACK GROUNDED reprinted the centerfold from the issue of PLAYBOY we just bought, PLAYBOY could bring suit against us for copyright infringement but Jim Strasma could not - Jim would have no 'standing'.

Suppose that Digital Acoustics made extensive modifications and additions to Applesoft and then copyrighted the result. Suppose it were to be shown in court that substantial original work was embodied and yet that substantial proportions of the original Applesoft remained. Is the Digital Acoustics copyright valid? YES! ANY COPYRIGHT IS ALWAYS VALID - until a federal court rules otherwise.

Suppose that the holder of the copyright on Applesoft (Apple or Microsoft?) brought suit charging copyright infringement. How would the ruling go? Well, there is no provision under copyright law which states that XX% of original material constitutes a new work. However, as court decisions accumulate so do guidelines. According to Jim, his lawyer friend asserts that 30% new material is the rough guideline in this country (but get a good lawyer anyhow, that is only a guideline!). We have read that the guideline in Europe is 25%. (It should be emphasized that we are talking about modified COMMERCIAL, COPYRIGHTED software.

Page 20, Column 1

Suppose that Terry Peterson modifies a public domain program, MICROMON, so that it can run on the Commodore 64. Suppose that Terry signs a contract with HES transferring his rights in the modified program to HES and that HES then copyrights the program. Is the HES copyright valid? YES! ANY COPYRIGHT IS ALWAYS VALID - until a federal court rules otherwise.

To invalidate the HES copyright a civil suit would have to be brought against HES charging copyright infringement. REMEMBER, THE PLAINTIFF MUST HAVE STANDING! Question: WHO, in this very specific instance, has STANDING to bring suit against HES for copyright infringement? NOBODY! No person on the face of the planet earth has standing to bring suit to invalidate the HES copyright! Why? Simple. It is agreed (for the purposes of this argument) that MICROMON is a public domain program. A public domain program, by dictionary and legal definition, is a program which is devoid of copyright protection and hence open to use by anyone!

If the original authors of a program choose to place the program in the public domain they are renouncing their rights under the copyright law! We would like to be present when Wozniak and Seiler, or Seiler and Cochrane, walked into a federal court and complained to the judge that HES (and Terry Peterson) had violated their rights under the copyright law of a program which they had placed in the public domain!

Thus, Jim was out of line to question the validity of the HES copyright - twice - in Midnite #17, page 12.

COPYRIGHT SUMMARY:

1. Reference is made to the U.S. copyright law.
2. Copyrights are very easily made and are ALWAYS, INSTANTLY, valid.
3. Only a Federal Court can render a copyright invalid.
4. A Federal Court can act only on a civil suit. There is no provision for any criminal prosecution.
5. The only permissible civil suit is one charging copyright violation.
6. The plaintiff in a civil suit must have standing.
7. It is impossible to infringe the copyright of a public domain program, since by definition such a program is devoid of patent or copyright protection.
8. Therefore no copyrighted work which is a combination (in ANY proportion) of original work by the

   Page 20, Column 2

   holder of the copyright and of public domain software can EVER be ruled invalid since no one has standing to bring civil suit charging copyright infringement. The issue of the percentage of original work becomes moot.

APPROPRIATE USE:

We believe we have proved that the use of public domain software is LEGAL. (In fact, such legality is INHERENT in the dictionary and legal definition of 'public domain'.) This leads us to the question of morality, since Jim evidently believes that Terry's (and HES's) use of MICROMON is/was immoral. This is trickier, as there is no bedrock of morality where one can say with confidence, "This IS (or is NOT) moral."

Our own moral standard is twofold: has the person (company) done something illegal? If not, has he hurt someone? If the answer to those two questions is no, then we cannot bring it upon ourself to accuse someone of an immoral or unmoral act.

Given Terry's assertion that he contacted the only author of Micromon that he was aware of (Bill Seiler, at his then workplace at Victor Technology) and asked for permission to use it in a commercial work, and given that Jim has not taken issue with that assertion, we simply cannot find any evidence of any immoral activity. In fact, if Micromon is in the public domain, we could not find any evidence even if Terry had not asked Bill. (It should be noted that Jim has Bill Seiler's HOME phone # in his possession and does chat over the phone with Bill about once a year.)

We can only conclude that Jim believes that the original authors of a work in the public domain continue to have rights in/to that work. A quick check of the dictionary definition of 'public domain' (which is the same as its LEGAL definition) proves that the authors' rights DO NOT continue.

To be fair to Jim, we should report that there is a lot of confusion in the personal computer industry over what constitutes 'public domain' . One constantly sees folk placing their software 'in the public domain for individuals but not for companies or for commercial purposes'. Unfortunately, there is no provision for such a form of QUASI-PUBLIC DOMAIN in the federal copyright law. A work is either in the public domain or it is not; there is no middle course.

(Hint: we copyright software and then make it readily available - for instance, by publishing it in REDLANDS. Nobody, not even the federal court system, expects folks to protect their copyrights by suing individual infractors - and yet we retain full rights under the copyright law in case a corporation tries to rip us off!)

Page 21, Column 1

FIGHT! ROUND 2:

Next month we will continue with a new subject: is Steve Wozniak (and Allen Baum) one of the original authors of MICROMON? Remember, nobody in the Pet world has EVER given Woz credit - even though we sent Jim Strasma a copy of the Wozniak/Baum disassembler source code in Interface Age. Did Jim check to see if the code and comments are the same?

These are important questions because Jim has severely criticized HES for not clearly acknowledging "the original authors of MICROMON". Can it be that Jim Strasma HIMSELF is EQUALLY guilty of the VERY SAME OFFENSE? (After all, Jim asserts in issue #17, p.12, par 2 that he is bothered by 'moral inconsistency'!) Tune in next month, same time, same station.

(Believe it or not, Jim - and Ellen - Strasma and your FNE have had many pleasant conversations over the past two or three years, including one quite recently. We think that Jim has made several errors of judgement, all centering on the question of whether original authors of public domain programs have continuing rights in or to those programs, and we think the opposing view should be published.)

By the way: we wouldn't DARE accuse Jim of being like an archbishop with a mistress! Not because we are afraid of Jim or his editorial pencil but because his wife Ellen would KILL us!

PLAYBOY WATCH:

Proving that we will leave no leaf unturned in our tireless dedication to our readers, we picked up the Aug '84 issue of Playboy. Why? To do some research on heavenly bodies, of course. For instance, on page 145 the moon is drawn improperly, being shown illuminated from above and to the right. If that were the case the sun would be in the sky above and it would be daytime, not night. The illuminated portion of the moon should have been drawn, for instance, in the reverse direction. Then the sun would (properly) be behind the earth and it would indeed be nighttime. But in whatever direction the moon is pointing, the artist should NOT have drawn a dozen stars in close proximity to the moon. The moon is a very bright object, being illuminated by direct sunlight. It will 'wash out' all but the very brightest stars in its (apparent) vicinity. You don't believe us? Next time the moon is up, count the stars in its (apparent) vicinity.

Still, the subject matter of that cartoon on page 145 might provide an answer to the question on page 75. When the men amongst our readers are finished checking out the heavenly bodies, they might try reading Colorado Governor Lamm's prediction of 10 years from

Page 21, Column 2

now. Lamm seems to think the third world countries will be broke by 1994; we thought they were broke NOW. Or haven't you been following the fun & games the IMF and the international bankers have been playing with Argentina et al?

To prove that we are not a male chauvinist pig, we read the article (we resist saying 'piece') on successful women executives beginning on page 76. To prove that there is still a tattered shred of dignity left in this rag, we will not question the whereabouts of the mouse in the two-page spread on pp78-79, which pictures a naked female executive with two-thirds of a Mackintosh.

TOO MUCH CUCUMBER ?

We have spent some time reading various articles, reviews, etc. on Mack and Mack-related stuff. You will remember that anything we write is suspect on account of we 'compete' with Mack - right? For once we will set the quotations and the attributions aside and talk about impressions. The impression we get is that there are an awful lot of folk who share our reservations over Mack's needless complexity and restrictiveness. Of course, most writers and most publications are not as outspoken or candid as this one.

Briefly, the reaction is resignation that Mack is designed for those described by Cyril Kornbluth as "The Marching Morons" in his early 50s novelette. Perhaps the problem is that these morons do not write articles and reviews? We know several readers and other acquaintances who own Macks and we know of quite a few other owners second hand through friends and acquaintances. You know what? There is not a moron, marching or otherwise, in the bunch! (There ARE a few computer unsophisticates - but ONLY a few.)

We are told over and over that the Mack purchaser does not WANT to write programs and is not ABLE to write programs. Is this really true? What percentage of Mack buyers, 120 days after purchase, have NOT purchased Microsoft's MackBASIC? If Apple is right, Microsoft's sales of MackBASIC must be VERY disappointing.

The other night we watched John Ford's old (1939) western Stagecoach. Just as the passengers were in extremis the cavalry arrived, heralded by bugles sounding the charge. It reminded us of the Queen's messenger saving 'Mack the knife' in the cliched nick of time and of a giant eagle with "Deus ex Machina Airlines" markings saving Frodo and Samwise after they (with a little help from Gollum) returned the Ring of Power to the volcano's depths.

But hark! Bucky, can you hear? Is it the distant sound of trumpets? Are they sounding the charge,

Page 22, Column 1

Bucky? Will we be freed from the tyranny of forever traveling north? Will we be permitted to write a program? Who is the instrument of our salvation? Why Bucky, it's, it's...

Hello there, Kindly Uncle Jack! If you will just unlock these shackles here... thanks! Boy, is THAT a relief! Let us and Bucky stomp around here for a little bit and get rid of the kinks... hey! Kindly Uncle Jack! Where are you going?

Well, folks, it seems that KUJ saw the gleam of a silver dollar up there in the hills and you know about KUJ and dollars, right? So we guess we will have to explain to you what properly should come from KUJ:

You see, he just bought Atari. Atari does not own a silicon foundry and it does not have a pact with Zilog which permits it to make its own Z800Xs. KUJ has cast a beady eye on the marketplace and discovered that 1) there is a hole in the personal computer market in the $400 to $1000 price range big enough to drive a billion-dollar company through, and 2) there aren't any 68000 machines out there in the traditional sense, Sinclair's QL not being in this country and seriously behind schedule at that. By 'traditional sense' we mean as in Pet, Apple, even Atari 800. Why, Bucky here once wrote a checkbook-balancing program!

We have no idea whether the above scenario is a good (or even reasonable) prediction. KUJ has proved that he can outguess lots smarter people than us. But we have to make SOME sort of prediction about KUJ's plans or lose local face. Our smart-alec (bowdlerized) project engineer asked us what we would do if we were KUJ and we immediately reacted, 'duhhh...' Actually, the bit about KUJ making a Trash-68 is undoubtedly just wishful thinking. We really would like to have a mass-market vehicle for HALGOL and it increasingly appears that Mack is not that vehicle.

Oh, yes: Atari's 1983 sales totalled $1.12 billion.

PURE GOSSIP:

It seems that this outfit which we will call company Z had a promotional videotape made of its forthcoming very high performance processor (something like the "Z eight hundred million"). These videotapes were prepared six months in advance of their intended release and contained, until that scheduled release, privileged and proprietary information.

The audio-visual outfit which produced the promotional videotape for company Z were proud of their work and fond of its associated profitability. So they

Page 22, Column 2

approached company M with a copy of the videotape they had produced and asked, "Would you folks like us to produce something like this for you?" The company M folks accepted the sample videotape (over 4 months in advance of the scheduled release date) and "evaluated" it for two weeks.

Then company M returned the sample videotape, stating "Nope! Can't use it." Company Z is, of course, unaware of the events which have transpired. And we, of course, have forgotten the true names of the companies involved.

Inter68 PASCAL UPDATE:

"Version 1.2 of Inter68, the Apple Pascal compatible P-code interpreter, has been completed. It features a true RAM disk instead of the "segment cache scheme" used in earlier versions. A 92K board is still sufficient, but to really use the RAM disk you should have more memory. With a 512K Grande you get appx. 850 blocks, and happy owners of a full gallon get appx. 1880 blocks! K(runching and B(ad block scanning happen almost instantaneously. The RAM disk comes configured as volume #10, but this can be changed to any other number the user might desire. In addition, the RAM disk can be designated the root volume. If you place all system files on this root volume you can work all day without ever leaving the RAM disk. If the RAM disk is the root volume, it is loaded from volume #4 whenever Inter68 is cold booted.

"Inter68 v1.2 contains a double precision floating point package, an engineer and scientist's delight! The Pascal compiler accepts a new standard type DOUBLE which can be used just like the standard type REAL. Up to 15 significant digits can be displayed by the WRITE statement, and the exponent ranges from -308 to +308. DOUBLEs, REALs and INTEGERs may be freely mixed in expressions, everything will be converted to the highest precision involved.

"The double precision package comes with a complete set of transcendental routines (all written in 68000 assembly language, of course): SIN, COS, TAN, ARCTAN, LOG, LN, LOG2, EXP, EXP2, EXP10 and SQRT. PWROFTEN is not included because EXP10 does a more general job, i.e. the argument is not restricted to type INTEGER. All transcendental routines return a value of the same type as the argument; INTEGER arguments are converted to REALs in order to maintain compatibility with previous versions. The package is currently implemented in software only, but will use the 16081 math chip Real Soon Now (both for single and double precision arithmetic). 16081 owners will be able to upgrade for $10 as soon as I have written the math chip routines.

Page 23, Column 1

"Inter68 v1.2 allows (relocatable) Phase Zero code files to be linked into Pascal programs. Such code files are first transferred to UCSD Pascal by means of a utility program and then processed by another utility program that generates enough linker information so that the SYSTEM.LINKER can link the assembly code file to a P-code file. The user supplies the number of arguments and the name of the external procedure/function. Only one procedure/function is allowed in a Phase Zero code file, but several Phase Zero files may be linked into the same P-code file.

"Inter68 v1.2 allows machine language procedures to call P-code procedures. The user specifies the segment and procedure numbers of the P-code procedure to be called; the parameters must be correctly pushed onto the stack. This scheme even works recursively.

"I have included five additional procedures in the unit APPLESTUFF which I felt would be useful. They allow you to communicate with the 6502 at a very low (and thus very efficient) level; included are PEEK, POKE and block transfers in both directions.

"The price for Inter68 v1.2 is $95 for newcomers and $50 for buyers of previous versions. Send international money orders to:"

Ulrich Schmidt
An der Junkersmuhle 33/35
5100 Aachen W. Germany

AVE IMPERATOR! (OOPS!)

You peasants will be pleased to learn that your FNE is now a very important executive. Important executives are easy to spot; they have a personal computer of the kind where the CPU is out of sight and there is just a keyboard and a cute little monitor pushed way back on the desk. Remember the DEC ads? Did you ever ask yourself where the CPU and disk drives were? Did you ever ask yourself where the CABLE to the CPU and disk drive was?

However, it is just IMPORTANT executives who have that DEC/Wang-like setup. WE, as we have just explained. are now a VERY important executive. VERY important executives have hidden CPUs which are superior to the common variety made by IBM, as our Tandy 2000 is. However, that is not the principal distinguishing characteristic of a VERY important executive! The principal characteristic is, uh - step a little closer and let us whisper this...

A VERY important executive has never used any of the software - MultiMate and 1-2-3 - that his company has purchased for use with that hidden CPU!

Page 23, Column 2

Now, about the OOPS!: as every Apple person knows, one commonly and routinely turns the Apple II on and off with the floppy disks mounted in the drives. Turning a Tandy 2000 either on or off with a floppy disk in place is absolutely guaranteed to destroy the portion of the diskette that happens to be under the recording head! We managed to prove this beyond all doubt, several times, before we learned what was going on. Fortunately, $250 MultiMate or $500 1-2-3 were not the disks we learned on. Talk about having to learn a brand new set of reflexes! (Write protect tabs do NOT help!)

When you use that $150 sideways mounting platform so the Model 2000 CPU can disappear around to one side of your desk, Tandy lets you turn the TANDY logo 90 degrees so it still reads right side up. It's nice that Tandy took care of the important stuff. Sigh.

LATE PCjr STUFF:

According to PC WEEK, IBM had been scheduled to announce the PCjr changes on 17 Jul but some last minute glitches got in the way. It seems the 'entry-level PCjr', the one which had been planned to sell only 10% of the up-market model (and hence represents a much smaller completed inventory stockpile) is being kissed off, i.e. is being left unchanged, chiclet keyboard included.

The model which currently lists at $995 and which represents 90% of the completed inventory stockpile of PCjrs is being split into two models, both expandable to 640K RAM and both with a new, genuine 62-key keyboard (shades of Mackintosh!). The original jr power supply is inadequate to handle these add-ons so there will be another box, an 'auxiliary power supply', to power the extra goodies.

IBM has given up trying to keep these boxes in the home so as to avoid competition with their own PC. The new boxes are said to be "nearly 100% compatible with all PC software".

And now we get to see how the revised PCjr sells. Doug Clapp of InfoWorld seems to think that IBM will uh, goof up somehow - he suggests by retaining the name 'PCjr', which is now indelibly associated with a disastrous loser (almost typed 'looser', Pete - you'd think we would have learned by now!). Gosh, with only 62 keys, how can they er, foul up the keyboard for touch typists? Golly!

COMETH THE 80286-BASED PC:

The blue-suede-shoes boys are drooling over the prospect of chaining lots of sheep to the forthcoming IBM super-PC. Is that why we feel nauseous?

Page 24, Column 1

 ------------------------------------------------------- 
                    The Halgol Review 
                 DTACK Software Exchange 
      4326 Congressional - Corpus Christi, TX 78413 
                      512-852-5208 
        "The Pretty Good 68000 Software Exchange" 
 ------------------------------------------------------- 
           "DSEx - An Institution Since 1983" 
 ------------------------------------------------------- 
                  DSEx Current Catalog 

 All disks are $3 each postpaid (overseas airmail $4 
 each postpaid). Document prices as listed, postpaid. 
 MAKE CHECKS PAYABLE TO "DTACK Software Exchange" 
 if you know what's good for you. 

 Disk 1 - DAS, DTACK linked Applesoft by Pete Soule 
 Disks 2/3 - Sensenig BASIC, sample programs, manual 
          and index.  For Grande/Grounded boards, >32K 
          memory (60K recommended) (must be ordered 
          together) 
 Disk 4 - Pascal Stuff 
 Disk 5 - Assorted assembly stuff and miscellany (not in 
          official distribution, order at own risk) 
 Disk 6 - Sensenig BASIC for the 16081 math chip, sample 
          programs and manual addition 
 Disk 7 - FigFORTH for static DTACK boards 
 Doc  1   -  Printout of Sensenig BASIC documentation 
          (including math chip extension) ONLY for those 
          without boards or with no convenient way to 
          print it from the disks. $6 postpd. (No 
          overseas orders on Doc 1 please, unless you 
          have Compelling Reasons) 
 ------------------------------------------------------- 
The HALGOL Review is the independent newsletter of 
 DTACK board applications and programming, published 
 every once in a while (pseudomonthly) by Jeffrey W. 
 Hull. Subscriptions both domestic and overseas are free 
 to any exchange member, board owner, or for the time 
 being, anyone interested in the DTACK boards or the 
 68000 microprocessor. Letters and calls are welcomed. 
 Continued existence is always problematical and entirely 
 dependent upon your correspondence. 
 ------------------------------------------------------- 
         If you can't get it from DSEx, you can 
             probably get along without it. 
 -------------------------------------------------------

MATH CHIP REPORT II:

Some very strange things are going on with the latest Nat Semi production run of 16081D math chips. They do not work in our experimental math board, for instance, at either 10 MHz or 6MHz. And the local Nat Semi folk seem to have no way to verify that the chips are good.

Page 24, Column 2

As reported earlier in this issue, Rockwell got a sample 16081D-10, which we tested. It didn't work. Since then, Nat Semi has handed Rockwell four more 16081D-10s and none of them work either. So we grabbed another $200 plus $12 to pay the governor and bought yet another 160810-6 from Avnet, our local authorized Nat Semi distributor. It doesn't work, despite the fact that the part number is identical to a bunch of others that DO work, right down to the D as in dog mask revision.

All right, we said to ourselves, we have a critical timing situation here. So we used a highly preliminary - make that a spectacularly preliminary - version of the 68000/16081 ap note which Nat Semi has advertised heavily. We changed some timing around, modified the *SPC pulse width, changed the pullup resistor on the *SPC. No matter what we did, the old chips continued to work and the new ones continued to imitate a paperweight. We have begun to suspect that all five chips are duds, including the one we paid $200 for.

So we called the factory. First, the schematic in that spectacularly preliminary ap note has been changed, to our vast relief (it made no sense to our project engineer). Next, we were advised to contact our local Nat Semi types, who could use their DB-16000 breadboard to determine whether the five parts were good or not.

So we called the local office. The locals were not at all certain (in fact, they were pessimistic) that their DB-16000 could determine whether the chips were good. They specifically asserted, as we recall, that it was unclear whether the old DB-16000 would work with the new 16081Ds! THAT WAS NOT REASSURING, SINCE THIS IS THE EXACT POINT WE WERE TRYING TO DETERMINE!

As of today, 19 July, that is how matters stand. We have five chips, one of which we paid for. None of them work in a test circuit (our experimental math board) which has proven highly reliable when used with earlier date code 16081Ds. The local's assertions that new 16081Ds might not work in a Nat Semi test board in which earlier 16081Ds work just fine makes us nervous. What is going on? Who's on first?

Can any of you readers cast a light on this situation?

Those of you who have recently purchased an experimental math board for software development purposes should try to get hold of an OLD date code 16081D part until this situation becomes clarified. Please do not demand that we resolve the situation in the next 17 microseconds? Obviously, we are concerned and in fact more so than you - we are all set to go on a major design project when Nat Semi (IF Nat Semi) stops screwing around, freezes the 16081 design, and puts it into production.

Page 25, Column 1

EXCITING TIMES:

As we mentioned in this rag before, "may you live in interesting times" is an old Chinese curse. Right now, times in the personal computer industry are not just interesting but downright exciting! Unsuccessful software companies disappear overnight; the successful ones have layoffs ranging from 50% to 90% of their personnel. Magazines which formerly were so thick it was dangerous to drop one on your foot are now so thin you can cut yourself on them if you are not careful. About 35 to 50 IBM PC clone-makers are dead; everybody knows they are dead; and yet they continue to imitate the signs of life for a little longer. Feverish speculation over IBM's future plans abounds when all you have to do is turn to page 442 of June BYTE to see what IBM products will be featured in Sept BYTE!

Everybody who visited NCC in Vegas had a chance to see Titan Technologies showing Analytical Engines' Saybrook card. You will recall that Titan Technologies took over Number Nine's Accelerator card and also some other folks' Z80 and 8088 cards, all under-the-hood boards for the Apple. So we called Analytical Engines late on a Friday afternoon to ask about that and the UNIX system which AE had promised for Spring '84. Results:

1) AE is firmly committed to really developing UNIX for their board someday soon. They have figured out that UNIX requires a hard disk, which may have cost them a board sale. (We had intended to buy a Saybrook/UNIX/floppy disk combination for use on days when we needed a good laugh!) 2) The deal, or whatever potential deal, with Titan Technologies was not finalized, hence we were instructed not to write about it. 3) We were told that the AE folk had been displeased with what we had written in these pages about them in the past, especially when the price of the Saybrook was rapidly escalating. Sigh.

We asked the AE spokesperson, what have we ever written about AE that is incorrect? "Nothing," the AE spokesperson admitted, "we just didn't like some of the stuff you wrote." AE obviously misunderstands the purpose of the media. It is not our job to just write stuff they like to read, else Nixon would still be President, hmmm? (Well, you get the idea.) If AE wishes to read stuff they like, they should stick to MINI-MICRO SYSTEMS or COMPUTE's Gazette. (AE does not read our newsletter regularly, so they don't know that the last time we picked on their UNIX for being late, we ALSO picked on our HALGOL for being late - in the same paragraph!)

The significance of Titan Technologies demonstrating the Saybrook board at NCC is that in the past they have purchased complete and exclusive rights to boards such

Page 25, Column 2

as the Accelerator - and Analytical Engines is a one-product company.

A bloody battle is shaping up this Xmas selling season between Commodore's C-64 and Trameil's Atari 800. We foresee a glorious Gotterdammerung; we wonder who, if anyone, will survive? As a couple of industry pundits have noted, Trameil does not actually have to destroy Commodore, just weaken its stock relative to Trameil Technology/Atari so that Trameil can purchase Commodore out from under Gould - the sweetest of revenges!

The head of Western Design Center is continuing his Napoleon/Mad Scientist act. He continues to rave about 8MHz samples which he is really, no lie, running. Meantime, would you like to buy a 1MHz sample for $95?

8MHz? In CMOS? (One clock is a complete memory cycle, unlike the 8086/Z8000/68000/16032s, which use 4 clocks per memory cycle.) Why, you'd need 50nsec memory if you didn't use address and data buffers. With LS244 address buffers and LS245 data buffers we are looking at 30nsec RAM. REALLY? And the only samples one can buy run at ONE megahertz?

MORE: In a brilliant move, Western Design Center has changed its assembler mnemonics from SET/CLR, three letters each, to SET/RESET. And if you have the nerve to correct that lousy mnemonic (and others similar), they will very nastily withhold your right to correctly identify the part number made by Western Design Center! So there! EVEN MORE: WDC brilliantly "fixed" the indirect,Y addressing mode. You 6502 assembly experts will recall that that addressing mode performed a read cycle before the write cycle. Chips so "fixed" by WDC will not run lots of Apple software, including the DOS. But it seems that Woz hardware design counts on that read cycle. So the Mad Scientist at WDC is now fixing the fix...

(We wonder if the AE folk understand why it is not our job to write stuff that they like? And that it has nothing to do with being competitors? Assuming that us and AE really ARE direct competitors, something which we are not ready to acknowledge.)

InfoWorld quotes Trameil: "A couple of years ago, I predicted that in 1987 the personal computer market would be 50 million computers a year. This year, there might be less computers sold than the prior year. The only reason is a shortage of components."

And some of you thought we were kidding - or exaggerating - when we told you that Digital Acoustics was having fits getting parts. Right now, as we get

Page 26, Column 1

ready to go to press with this issue, our production of the VDHR board sets - which has recently accounted for a large part of our dollar volume - is just plain shut down for lack of a critical part (a 67.888MHz xtal). This condition is expected to continue for another two or three weeks.

The post-election recession is widely expected to cure the parts shortage. Hooray? (Actually, Digital Acoustics did better during the last recession than during the post-recession recovery. For several reasons which we will not go into at this time, we think the situation is going to repeat. We seem to be the opposite of an olive grove - when a recession hits, the first thing folks do is quit buying olives!)

We now read an awful lot of IBM PC-related publications. More than Apple-related, in fact. If we had wanted, we could have BURIED this issue in quotations from those PC-related publications supporting our viewpoint that FAST wins and SLOW loses! Most of our readers (see Hal Chamberlin's letter in this issue, for instance) have come over to our point of view (if they were not here ahead of us!).

Even the folk at InfoWorld have caught on (probably because they follow the IBM publications maybe closer than we do). Read Doug Clapp talk of MBA's lack of success vs. 1-2-3 and of the Context folk looking for employment. Read InfoWorld's story about Son of Valdocs (ah, make that Valdocs 2) which is a lot faster than its daddy. Even George ('you'll see me dead before you see Morrow Microsystems fool with the 68000') Morrow has proudly introduced his company's new machine, which is based on the uh, 68000...

Why do we still have a couple of pointy-headed readers who think it is a good idea that Mack runs MackBasic slower than Applesoft? Huh?

Every time somebody does something worthwhile, as in Motorola's introduction of the 68020, you can count on a bunch of yahoos grabbing publicity on the coattails. In this case, both NEC and Hitachi took advantage of the 68020's introduction to showcase their own latest paper designs. Like all paper designs, these are truly wonderful. Hitachi's paper design is a superset of the 68020 with shrunk design rules (1.3 microns). NEC's paper design is a proprietary superset of the 8086 (WONNERFUL, WONNERFUL!) with "as many as 700,000" transistors on board. And please excuse us: the NEC part is not even, yet, a paper design! ("Although architecture for the device is not yet on paper, NEC plans to have the part in production by late 1986.") Of COURSE this not-yet-paper design will be in full production in late 1986! Who could doubt it?

Page 26, Column 2

The quotations above are from Jul MMS, p.86. Meantime, we already are reading in other publications about companies "perhaps using the NEC 32-bit CPU" in their new fall line of personal computers. Hoo boy, look who just fell off the turnip truck!

PC magazine published our little arithmetic correction for Adam Osborne as a letter. You know, correcting his "7K to 12K" per month sales for Lotus' 1-2-3 to their actual 37K+ per month in the first quarter of 1984. But we haven't received a thank-you note from Adam yet. Perhaps he is too busy bowing to the applause accompanying his new book (with John Dvorak) explaining how he was personally blameless in the Osborne Computer debacle?

If UNIX succeeds beyond its boosters wildest dreams, or if its boosters suddenly compare actual UNIX sales to actual MS-DOS sales and all commit hari-kari, we (meaning both Digital Acoustics and FNE) do not stand to gain or lose a cent. That makes us a disinterested observer of the UNIX scene, which is not the same as uninterested.

Then there is John Little of Multi Solutions, Inc. John's polemics against UNIX make our own seem mild, Unlike our opinion, which is that UNIX is unsuited to the mass marketplace but well suited to certain applications involving full-time computer professionals, John asserts that UNIX is unsuited for any use whatever. The problem is, John is NOT disinterested. The company which employs him offers for sale an operating system which competes with UNIX and which (naturally) is (ahem!) VASTLY SUPERIOR TO UNIX!

One can therefore clearly detect an ax being ground behind his vitriolics against UNIX. Still, a lot of his criticisms of UNIX are valid; it is not hard at all to find legitimate complaints about UNIX. We would be more impressed if the competing operating system in which he has a financial interest actually existed as advertised. Judging by a 'review' of that system ("S1") in David Fiedler's UNIX newsletter Unique, it does not even properly exist!

But we are definitely not UNIX' severest critic. Question: are we UNIX' severest DISINTERESTED critic?

As hard as it is to believe, new IBM PC clones continue to appear on the scene. One is minded of a Mr. Lemming, on seeing the mass-migration passing on its way to the sea, calling to Mrs. Lemming: "Mavis, you'd better hurry or we'll be late!"

Page 27, Column 1

PEARL HARBOR BOMBED!

Yes, folks, we always have the latest and hottest news in this rag! Uh, make that 'sometimes'. This newsletter is the effort of one person, and it is written at a pace of about a page a day. When events pass something written earlier in the month we do not have a horde (not 'hoard', Peter) of editors to do a re-write. Today is just the 8th of July and it is apparent that the story of the industry slowdown and shakeout is going to be decidedly old hat by the time you read this. Well, when we did that writeup just 9 days ago, that news was NOT old hat! Honest.

It goes like this: IBM announced that it was going to triple personal computer sales in 1984 over 1983. The entire industry is aware that IBM does not lie. It might make a mistake, but it does not lie. Accordingly, software houses and peripheral manufacturers and other tick birds budgeted for a massive increase in sales. And the retailers stocked their warehouses with inventory to prepare for that massive demand.

But IBM had planned for the PCjr to account for a third of those new sales. You have maybe heard about how jr is selling, right? So that brings IBM's forecast down to a doubling of sales (IBM does not lie but it makes mistakes). What IBM did not emphasize - and what a lot of folks overlooked - is that IBM planned to achieve a lot of that doubling of sales of the regular PC by stepping up international production and sales. They have in fact done that and are selling a LOT more PCs in the Common Market (Europe) this year than last.

The point is, although IBM's prediction of a tripling of sales was entirely honest IBM never intended or planned to triple domestic PC sales. In fact, IBM did not even plan to double regular PC sales domestically; 60% is more like it. But the folks planning their 1984 budgets did not take the time to analyze IBM's plans and went merrily along believing that the industry was going to triple sales in 1984! (The situation is precisely analogous to Fortune System's gearing up to meet the 600,000 UNIX system sales that Jean Yates forecast for 1983.)

Well, it is now apparent that even IBM's planned DOMESTIC sales increase of the regular PC (which we guess was 60%) may be slightly optimistic for a variety of reasons. So we have here an industry (IBM-related software houses, IBM-related peripheral manufacturers and other tick birds) which began the year primed for a massive increase in sales. The massive increase having failed to transpire, cutbacks and layoffs are the order of the day. The companies with thin margins and no cash hoard (not 'horde', Peter) ain't gonna make it.

Page 27, Column 2

(Damn! Since writing the last sentence of the above paragraph, we have seen almost the identical words published in at least four publications. Question: did we unconsciously copy those words from some place else? Or are they so obvious that lots of folk can come up with them independently? Nobody is copying US - we haven't gone to press yet!)

Most observers are bundling the ongoing problems at Franklin and Eagle in with the above. Uh, uh. Those two companies did well in 1983 because they were selling something they did not own - the Apple and IBM BIOS ROM, respectively. Now that the legal system has caught up with them and they are restricted to selling stuff they DO own, they aren't doing so well. It could not happen to a more deserving bunch of guys! (Actually, Eagle has the double-whammy of also being a clone-maker when clones aren't selling.)

Well, there really were more UNIX boxes sold in 1983 than in the previous year. Nearly double, in fact. 1983 was a poor year for UNIX only for those folks whose expectations were aligned to Yates' 600,000 forecast. And there really will be more PCs and PC clones sold in 1984 than in 1983. Even domestically. But the number is not going to triple and it is not even going to double. 1984 will be a good year for everybody except those folk whose expectations were not aligned with reality.

BUT - that is on average! And on average, jumping out of a first floor window 25 times equals jumping out of a 25th floor window once, right? Here is another factor that is tearing up the clone-making industry: the retailers are among the folk who prepared for that massive onslaught of sales that was gonna transpire in 1984. So they stocked up their warehouses.

Right now, the warehouses are full and the retailers are desperately trying to reduce inventory. Trying to reduce inventory to bring it into line with the current, lower than expected sales rate. AND - AND trying to reduce their inventory of low-tech 8088 stuff before IBM changes the rules with a new, higher-tech PC. The net effect is that the retailers have, momentarily, just plain stopped buying. This is happening AT THE SAME TIME THAT THE CLONE-MAKERS ARE ADJUSTING THEIR PRODUCTION LEVELS TO REALITY!

Result? The completed PC-clones are piling up as completed inventory in the manufacturer's warehouses and the clone-makers do not have enough cash flow to buy a box of paper clips! Sure, the whole year is going to average out, BUT RIGHT NOW NONE OF THE CLONE-MAKERS CAN PAY THEIR BILLS! Maybe that's why everybody can figure out independently that the companies with no cash hoard to carry through this trying time ain't gonna make it?

Page 28

This figure courtesy of Alan Kalker of Microsculptor in Berkeley, CA

Numbers Crunched on 68000 by DTACK
Graphics Made Possible by FONTRIX

20 Hgr2: For Z=-120 to 120 Step 3: Xlimit = Sqr(14400 - Z*Z)
30 For X = -Xlimit to Xlimit : Rem Step .25 for more data
40 X0 = Sqr(X * X + Z * Z) * (3.14159 / 100)
50 Y = (Sin (X0) + 2.5 * (Sin(X0) * Sin(X0))) * 20
60 Gosub 70 : Next : Next : End
70 X1 = 130 + (X + Z/4) : Y1 = 100 - (Y - Z/4)
80 Hcolor = 3 : Hplot X1, Y1
90 Hcolor = 0 : Hplot X1, Y1+1 to X1, 191 : Return

Page 28, Column 1

There was a pretty good story in this morning's (Sunday's) L.A. Times business section about the above problems. We especially liked the euphemistic phrase "practicing cash conservation". That means you don't pay your bills. There is a lot of that going around these days; we understand that payables and receivables are uniformly over 90 days any which way you look (except for us and a few really big outfits).

We have been using an outside contractor to stuff and wave-solder our circuit boards for about seven months. Although we have regular 30-day commercial credit with this contractor, we always pay his invoice immediately when the product walks in the door. A lot of his other corporate clients are "practicing cash conservation". This contractor seems to give us real good service, even though we are much smaller than some of his other customers. We wonder why?

Too bad that a lot of this newsletter is going to be ancient history even before it goes to the printers. Pearl Harbor bombed indeed!

THE REJECT PILE:

"The 'case' should be named 'Justin'!" "The Pure Gossip Rag was written by Scott Joplin and published by Ink Spot in 1902." "...simulation program for a

Page 28, Column 2

supersonic helicopter. I know it will work since I've seen it on TV - but since the results keep saying the blades will fall off there must be a bug in your board." "You can't expect us hardware types to have orderly habits like those goddam left-brained programmers can you?"

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 184th 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:

DTACK GROUNDED
1415 E. McFadden, Ste. F
SANTA ANA CA 92705

REDLANDS has turned up missing this month, as has part II of the WEITEK story. Maybe next month?