DTACK GROUNDED, The Journal of Simple 68000 Systems
Issue # 6 January 1982 Copyright Digital Acoustics, Inc
The detestable device illustrated below is the ONLY, repeat ONLY hardware device which prevents the use of our 68000 board with the following potential host computers: TRS 80 I, TRS 80 II, TRS 80 III, TRS COLOR, XEROX 820, IBM Personal Computer, KIM 1, ELF II, AIM65, Heath H89, Osborne 1, Toshiba T-2XX, Sinclair ZX81, ACORN ATOM, HP-83, HP-85, HP-125, Compucolor (various), Compucorp 6XX, Sord MXXX, Intelligent Systems 3650, Superbrain QD 64K, Atari 800, NEC PC-8xxx, LNW80, SB-80, OSI (various), DEC LSI 11/2 and 11/23, TI-99/4, VIC20, STD BUS (various), S50 BUS (various), and the ubiquitous S100 systems (8 bit cpu varieties) and others!!
We have included the 16 bit TI-99/4 in that list because, as everyone knows, it is quite slow. We have included the LSI 11/2 and 23 because, although most people are unaware of the fact, they are quite slow.
If you own one of the computers listed above, that rotten parallel interface below (with all of six standard LSTTL integrated circuits) is the ONLY hardware which needs to be changed to hook a 92K DTACK GROUNDED 68000 board, with (eventually) 128K expansion boards and the Intel 8087 math processor to your computer!!
If the prospect of hooking a 68000 board to your computer doesn't excite you, just why are you reading this newsletter? Oh, because it is called "The Journal of Simple 68000 Systems"? Well, don't go away. We have something for you on the next page.
DIRTY, ROTTEN, LOUSY OBSTACLE TO 68000 USE
HOW TO BUILD A REALLY SIMPLE 68000 SYSTEM: You may have noticed that the 68000 comes in a very large 64 pin package. Even discounting the four pins used for power and ground, that leaves 60 active pins to worry about. Here is how we go about simplifying things:
After making the changes above, we have eliminated 15 pins. So we are down to 45 active pins and we can STILL address 16.7 megabytes (we have just described the production DTACK GROUNDED boards)!
THAT'S NOT SIMPLE ENOUGH? You want to drop a small, VERY SMART circuit board into an intelligent instrument? Is 32K bytes enough for ROM, RAM and I/O? Then ignore pins 43, 44, 45, 46, 47, 48, 50, 51 and 52. Now you are down to 36 active pins, about the same as a 6502 or Z80! By the way, with only 32K you will not need address buffers or data buffers. Our original prototype 68000 board didn't use these buffers and it worked fine. That, sir, is a SIMPLE 68000 system.
It IS necessary to separately decode the upper and lower 8 bits of the 16 bit data bus. This will require one (1) extra 74LS138, and lets you read and write 8 bit bytes with your 16 bit data bus when needed.
If you want to run at 8 MHz, plan on using LS chips such as the 244 and 374 for interfacing to the outside world. The various NMOS interface chips (PIAs, ACIAs etc.) can't keep up. But LSTTL interfaces are easier to trouble shoot anyhow.
IS THIS INFORMATION FREE? Heck, no. If you build a smart 68000 board for an intelligent instrument using the above information, you must send a six-pack of Heinecken Special Dark or $5 in small, unmarked bills to NEWSLETTER EDITOR, C/O DTACK GROUNDED.
If you are a hardware hacker, just the information above may be worth MANY times the cost of your subscription. FOR SURE, you are NOT going to get that information from Motorola!
However, Motorola DOES provide useful application information on building COMPLEX 68000 systems. For example, they will on request provide application note AN-819, "PRIORITIZED INDIVIDUALLY VECTORED INTERRUPTS FOR MULTIPLE PERIPHERAL SYSTEMS WITH THE MC68000" (no, we did not make that up).
We are not sure, but we think that Motorola Ap note is related to using the 68000 to monitor a keyboard and game paddles plus maybe a 300 baud serial port (Whoopee!). We aren't nearly as smart as Motorola, so we would prefer to use a 6502 or 8048 to monitor keyboards, etc. and let the 68000 concentrate on USEFUL tasks. Remember, we said in Newsletter #1 on page 6: "the 6502 is very good at monitoring keyboards".
The architecture of the future (and of the recent past in many instances) uses a separate processor or processors to handle I/O. There are lots more peripheral devices available for the 6502/Z80/68XX than for the 68000. You want to use a 68000 to monitor a KEYBOARD? That's INSANITY!!
GOOD READING DEPT: The Dec '81 issue of IEEE SPECTRUM has an EXCELLENT article on page 30 regarding present and future trends in high performance microprocessors. Let us get the two things that are WRONG with the article out of the way first: 1) The title does not accurately represent the content of the article and 2) The author accepts uncritically Intel's unproven performance claims for the iAPX 432 (more on this later). If you like articles which force you to THINK, be sure to catch this one!
The sub-title of the article summarizes the subject matter very well: "The trend is toward using ever-cheaper VLSI circuits in place of costly software - but some would reverse the trend". The article treats two diametrically opposed approaches to high performance computing in a fair and even-handed manner. By "ever-cheaper VLSI" read Z8000, 68000, iAPX 432. "Reversing the trend" refers NOT to any extant or pending commercial hardware but rather to a proposal by Bell Lab's David Ditzel and David Patterson of U.C. Berkeley (and others). This proposal suggests using very simple (and therefore low cost) 32 bit microprocessors. These processors would have very simple instruction sets that would execute as swiftly as possible.
One sentence from the article is relevant to the brief review of the T.I. 99XX series microprocessors which appeared in Newsletter #4: "The TI systems are, indeed, in a class by themselves; they do not have general registers, accumulators, or stack pointers, and only by looking deeply into their architecture is it understood how they do any computing at all." Now, what competitor of TI do you suppose said that? Actually, that statement was attributed to Harvey Cragon, senior fellow at Texas Instruments!
DATA PATH ANALYSIS: Back in 1976, with two successful 6502 designs under our belt, we got a brilliant idea. We would build a system using FOUR 6502s which would (naturally) calculate transcendental functions four times faster than a SINGLE 6502. So, we sat down with a sharp pencil, some nice clean paper and a happy smile to figure out how to do this. The happy smile lasted about three minutes. It turned out that our four CPU system would run only twice as fast as one CPU in that application! Adding even more CPUs would not improve the situation, either.
Although we did not know it at the time, what we had run into is what the computer scientists refer to as "data path analysis". Briefly stated, data path analysis is the rigorous, scientific study of the fact that, when climbing a ladder, one cannot climb from the sixth to the seventh rung unless one has previously climbed from the second to the third rung! THIS, by the way, is the reason why the Cray-1 supercomputer is NOT built using a phalanx of 10,000 6502s.
So much for data path analysis. The point being made here is that one does NOT ordinarily increase the operating speed of general programs N times by adding N processors. And NOW we are ready to talk about the Intel iAPX 432.
ABOUT THE INTEL iAPX 432: What follows here is NOT an introduction to the 432. If you have no interest in this device, we suggest that you skip past the next page. If you are interested, you will have (or should have) already read Intel's promotional literature on this device. What we are going to discuss here is information about the 432 which you cannot get from Intel or from other publications.
It may have come to your attention that Motorola's insistence that the 68000 was suitable for use in very complex systems EXCLUSIVELY turned out to be false. Certain claims by Intel for the 432 appear to be even further off the mark. Since the rest of the microprocessor media seems to be regurgitating Intel's press releases verbatim, WE will be the bad guy and raise the awkward questions.
WHY DISCUSS THE iAPX 432 AT ALL? Because the 432 is the only candidate on the horizon to challenge the present and future supremacy of the 68000 as the best available GENERAL PURPOSE computing device. Also, one of the two 'execution unit' chips is about 98% identical to the 8087, another Intel product in which we have a keen interest!
iAPX 432 QUIBBLE #1, THE DATA PATH PROBLEM: Going back to an article in Electronics Magazine, Nov. 6 '80, we find an early introduction to the 432. In this article, it is reported that "processing power can be incremented solely by plugging in additional (execution) units, without affecting the software". The accompanying graphic shows a NEARLY LINEAR increase in system performance as the number of 'execution units' goes from one to five.
Specifically, the 'one unit' system has a 0.5 MIPS rating and the 'five unit' system has a 2.0 MIPS rating (MIPS= millions of instructions per second). Apparently, this nearly linear increase in performance is attained by simply hanging more of those 32 bit execution units on the 16 bit, repeat 16 bit, 'packet bus' which carries ADDRESS AND DATA FOR ALL FIVE 32 BIT EXECUTION UNITS!
Far be it from us to state that the above performance claim is impossible. However, the 432 has not, to the best of our knowledge, been publicly demonstrated with more than one execution unit in place. In the absence of such a demonstration, we reserve the right to question whether such performance can be attained. Aside from the data path or scheduling problem, there is the 'bus bandwidth' problem of getting both address and data to FIVE '32 bit execution units' over ONE LOUSY 16 BIT DATA PATH. We are from Missouri!
QUIBBLE #2: The 68000 has a 16 bit external data bus, a SEPARATE 24 bit address bus, 32 bit registers and a 16.7 Mbyte linearly addressable memory space. The 68000 is referred to by Motorola as a 16 bit processor. The 432 has a 16 bit external bus which is multiplexed; part of the time this bus carries addresses and part of the time it carries data. The maximum linearly addressable memory space is ALSO 16.7 Mbytes. The size of the 432 registers is irrelevant since the user can't get at them. The 432 is described by Intel as a 32 bit processor!
QUIBBLE #3: Intel has made the 432 MUCH easier to program (or so they claim) by restricting the programmer to high level language constructs. It is NOT POSSIBLE for the programmer to directly manipulate registers and such. Predictable result: rumor has it that the 8086, when programmed in assembly code, will outrun the 432 in specific applications. Consider how useful your Apple or Pet would be if you could not peek, poke or run machine language!
QUIBBLE #4: Intel executives have trumpeted the 432 as the answer to the "programmer crisis". The 432 will be so EASY TO PROGRAM (they say) as to actually reverse the trend towards more and more programmers. At the SAME TIME, Intel states that the 432 is "intended for projects in which there is a large initial software effort (10-person team for 2 yrs) followed by considerable effort to maintain and enhance the software. Gentlemen, WE CALL YOUR ATTENTION TO THE FACT THAT THOSE TWO CLAIMS ARE MUTUALLY CONTRADICTORY!
Intel CANNOT have it both ways. Either the 432 is vastly easier to program, in which case a 20 man-year INITIAL programming effort is ABSURD, or a 20 man-year initial programming effort is really needed, which means the 432 is probably HARDER to program than most other CPUs.
Intel is in the position of a carnival barker claiming that "in this tent over here is a 60 foot tall gorilla". Now, no member of the public has SEEN this gorilla. NOBODY HAS EVEN HEARD HIM ROAR. But the rest of the microprocessor press simply reports, "Yup. There's a 60 foot gorilla in that there tent."
SON OF DTACK GROUNDED? We think it would be fun to get hold of a 3 chip iAPX 432 set and hook it up to a little RAM, some PROM and a few LSTTL interface chips. We would then apply an initial software effort of 20 moron minutes and see what happens! Would you like to buy a SIMPLE iAPX 432 board?
MOTOROLA 68000 REPORT: The first shrink of the 68000 mask is about to take place. This will result in a 30% smaller chip area and result in improved yield plus a higher typical clock rate. It is anticipated that the typical speed of parts built with the new mask will be 10 MHz with some yield at 15 or 16 MHz!
RUMORS: Motorola is also excited about OTHER pending developments which they are unwilling to discuss specifically. Coincidentally, Charles River Data Systems has just started a big advertising campaign for THEIR $20,000 68000 system. This advertising campaign has started with a two full page advertisement. The left hand page features a large green frog holding a 68000 chip, period.
The remainder of the advertisement claims that the 68000 system is a FULL 32 BIT SYSTEM. Now, anybody who has studied the architecture of the 68000 can see that the intent is to eventually turn the device into a true 32 bit microprocessor, a VAX on a chip. The sticking point is that the additional circuitry would result in a device so large that there would not be an economical yield.
You don't suppose that the folks at Charles River Data Systems know something we don't, and that the impending 33% shrink will make it possible to go directly to a full 32 bit processor in 1982? (WE don't expect to see a 32 bit 68000 before mid-1983, but that big frog could be smarter than us. And Motorola IS excited about something coming up.)
68000 PRICING: The various 68000 vendors are maintaining a HARD LINE in their published price lists and a SOFT LINE across the counter. When we called the Hitachi distributor for pricing on the 8MHz part, we were told it was $169. Our buyer stated that we were getting the part from Motorola for $144. We received a call back within the hour offering $140 parts. We needed two that particular day, so we called the Motorola distributor back and got an immediate offer of $138, which we took. If you have lots of time to bargain you can surely do better. We expect the 8MHz price to drop below $100, quantity one, by mid-1982.
RAM PRICING: We also expect the 2K X 8 static RAMs to drop under $6, quantity 1000, by mid-1982.
******************* SPEAKING OF MONEY! *******************
YOU WILL SEND US MORE MONEY: Most of your subscriptions expire with this issue. If you would like to continue to receive this elegant and sophisticated newsletter you will have to bribe us with an additional $15 (U.S. and Canada) for 6 more issues. That's $25 elsewhere. Yes, we know we didn't ask your permission, but we are going to continue this rag anyway.
AMATEUR DEPT: As a bonus (?), everyone who renews their subscription will receive a copy of a letter written on June 28, 1981. This letter describes the trials and tribulations of developing our prototype 68000 board. The letter makes it clear that we had NOT originally planned to offer it as a commercial product! THAT decision was made shortly after the July 4 weekend, during which the 68000 floating point package was developed.
This letter was sent to 135 magazines and computer clubs. It was printed in its entirety in Oct. 81 COMPUTER WORLD (a British magazine). Excerpts have been printed elsewhere, including in the Kilobaud letters column. Most of it was printed in FWAUG (if you don't know, don't ask!) including an editorial correction which is wrong! With this letter, you will have the complete history of DTACK GROUNDED, including what we may call the prenatal period!
IGNORANCE IS NOT BLISS: We would like to tactfully point out that a great many people, specifically including many involved in microprocessor industry publishing and editing, are ignorant of how a modern general purpose 16 bit processor addresses memory. They seem to think that a 16 bit machine is not usable as a text processor since half of each word would be unusable!
True, in the past there were some crummy minicomputers which could NOT address at the byte level, but the 68000, 8086 and even the late Z8000 CAN. And all three do it A WHOLE LOT BETTER than any of the 8 bit machines. This word badly needs to be spread in certain editorial offices, starting with Creative Computing.
Please do not write to us about the T.I. 9900 series. We know, we know!
GOOD OLD BOY DEPT: All persons and events in the following story are fictitious and any resemblance to living persons is purely coincidental.
As Richard Pretty drove his famous mauve and puce stock car into the pit area, his demeanor belied the stately pace with which the car was moving "Senior!" he shouted at his chief mechanic, Senior Johnson. "Senior, what the bleep have you done to this here car? I was passed on the backstraight by everbody in the bleeping race plus two shopping carts and a baby buggy! Didn't you get that new carburetor installed right?"
Senior Johnson replied, "I done saved you over a hundred bucks, Richard. Instead of that big $144 carburetor you sent me after, I got a littler $21 job made by the same company. The salesman told me the one I got is the logical replacement for that obsolete model we got rid of."
"Logical my fanny!" screamed Richard. "Some of those cars passing me were going ten times as fast as I was! You also saved the bleeping gas and tire money because the stewards done black-flagged me for imitating a parked car!"
The moral of this story is that sometimes you can't AFFORD to save a hundred bucks. The next time you have Senior Johnson slide a new carburetor under the hood of your stocker, make sure it's the right one. By the way, Senior Johnson isn't Richard Pretty's chief mechanic anymore.
BURTRONIX DEPT: When we arranged for Don Burtis of Burtronix to do the interface between our board and the Apple II, part of the deal was that we would acknowledge Burtronix's work. We did, on page 7 of issue #1. But since his design works so well, with no bugs whatever right from the start, we'll mention Burtronix here again. We hope Don reads this THIS TIME!
PRICE AND DELIVERY: The price of the 68000 and 6809, ceramic package, quantity one, is $144 (68000 8MHz) and $21.60 (6809 1MHz). The 68000 is in stock at your local Motorola distributor, such as Hamilton Electro Sales in the Los Angeles area. That is, provided you want the 8MHz version, because that is the version that's moving at the distributor level. You want a different speed? They're in stock at Motorola Phoenix, so it will take a couple of weeks to get the part.
THE VALUE OF AN ITEM IS WHAT IT CAN BE SOLD FOR. Motorola is still bravely carrying a list price of $105 on its 4MHz version of the 68000. The sales at that price have been underwhelming to say the least.
ABOUT TANDY: If the new 16 bit Radio Shack computer uses a 68000 as reported by Electronics magazine, we predict that it will use a 4MHz version, from Motorola, and that less than $25 will change hands for each processor. By the way, will that be a Trash 80 IV, a Trash 68 IV or a Trash 68 I?
WHAT MAKES A SUCCESSFUL PERSONAL COMPUTER? From the viewpoint of the manufacturer, it is the PROFITABILITY of a particular model which determines success or failure. Not clock cycles. Not reliable disk drives. Just how much money sticks to the manufacturer's hands!
Judged on that basis, the TRS 80 model I is the best personal computer Tandy has made to date. A strong case could be presented that the profitability of the various Tandy personal computers is INVERSELY proportional to the engineering effort expended. Said another way, the personal computers put together by the marketing types have been more profitable than those put together by the engineers. Which should not be surprising in a marketing company like Tandy.
We wonder whether the new 16 bit Tandy personal computer will be an exquisitely engineered failure or a sloppily engineered success?
THE STAR OF COMDEX proved to be the Fortune Systems 68000 system. Cost: $5000. Uses a 6 MHz 68000, and DTACK is NOT grounded, which means the processor probably runs at an average of about 5MHz. If that is a good speed estimate, our board (which runs at 8MHz at all times) is about 1.6 times faster. Still, it's good to start seeing less absurdly expensive 68000 systems.
Also at Comdex, but not the star, was Chuck Peddle's new machine, being marketed under the Victor label. This machine uses an 8088 (probably 5MHz), 128K RAM, 400 X 800 pixel graphics and two floppies, 600K each. Not a bad package for $5000! With that disk capacity it is a far better value than the equivalent IBM configuration, but Victor simply isn't as glamorous a name as IBM. The name IBM conjures up visions of enormous computers attended by dignified type persons wearing white coats. The name Victor conjures up visions of a lowly clerk in a cluttered back room with a noisy mechanical adding machine ratcheting away!
MISSING COMPUTERS DEPT: We imagine a guy running around in the swamp at night with a kerosene lantern, crying out (after a catchy musical phrase) Lisaaa!! (Apologies to Chloe.)
NOT at Comdex was Apple Computer's 68000 machine. With the $5000 Fortune Systems device and soon the (almost certainly less expensive) Tandy device on the market it will be increasingly difficult to justify an $8,000 or $10,000 pricetag on a 68000 machine unless it is a specialty item such as the Wicat graphics system. And specialty computers do NOT achieve large sales volume.
If the Apple 68000 machine is to be successful (as is large sales volume) it is going to HAVE to be price competitive. And, a price competitive 68000 Apple IV will totally wipe the Apple III out of the marketplace. Once again we speculate: is a substantial inventory of unsold Apple IIIs holding up introduction of the Apple IV? How much longer can Apple AFFORD to stay out of the 68000 market?
TAKE A 6502 TO DINNER: All the new introductions are featuring 16 bit processors or 8 bit devices such as the 8088 and 6809 which are REALLY 16 bit devices (marketing says so, right?). So be NICE to your 6502, even though it is getting a bit old and bleary-eyed.
A LITTLE 68000 STUFF: Assembly language programmers are probably familiar with the three 2 byte vectors at the top of the 6502 memory space: NMI, RESET and IRQ. That is a total of six bytes. The 68000 has 63 vectors plus the machine stack pointer initialization value, a total of 256 bytes! Each vector is a four byte address, of course. The 68000 vectors reside in the LOWEST 256 bytes of memory, $000000 to $0000FF.
You will note that although the address registers are 4 bytes wide and each vector is 4 bytes, we only use 3 bytes to designate an address. This is because the most significant byte is, for now, a 'don't care' due to the fact that it is not (yet) brought out to the outside world.
The startup sequence for the 68000 after RESET is as follows: the first four bytes in the address space are loaded into address register 7, the machine stack pointer. The next four bytes is the RESET vector. These bytes are loaded into the program counter and program execution starts at that point. In the DTACK GROUNDED board, the reset vector points at $000100. Since we use bipolar PROM from $000000 to $0003FF, we therefore begin executing code on the monitor PROM.
Also, we have used only eight of the 256 bytes in the vector table. The remaining 62 vectors point to memory locations to which program execution switches after certain things happen. These 62 vectors fall into two categories, the first of which is a spurious vector.
SPURIOUS VECTORS: When the BERR (Bus ERRer) line goes low, the 68000 will fetch a new program counter value from a location in the vector table and begin program execution at that new address. As it happens, the BERR vector is the one immediately following the RESET vector. There is just this one problem: since we have wired BERR to +5 on our board, this vector should never be taken. So we call this a SPUR, for spurious. Since we also have the three interrupt lines wired to +5, the seven interrupt vectors are also SPURs. The vectors which Motorola has not yet assigned are treated as SPURs.
All SPURs point at the same location: $001000, which is the start of the DTACK GROUNDED RAM. The first six bytes are initialized on RESET to $4EF9 0000 0122, which happens to be a (long) jump to the place in the monitor PROM where the 68000 idles, waiting for a command from the host. This is as close to "ignore" as we could come. Remember, a spurious interrupt should never occur!
And since this secondary vector is in RAM, the user can change it.
In a future issue we will discuss what we do with the LEGAL vectors which remain. We do not want to turn this newsletter into a data manual! We'll just sneak in a little technical stuff occasionally.
HERE IS A SIMPLIFIED MEMORY MAP FOR THE DTACK GROUNDED 68000 BOARDS: U2, U3= $000000-0003FF U26, U27= $00C000-00CFFF U4, U5- $001000-001FFF U28, U29= $00D000-00DFFF U6, U7= $002000-002FFF U30, U31= $00E000-00EFFF U8, U9= $003000-003FFF U32, U33= $00F000-00FFFF U10, U11= $004000-004FFF U34, U35= $010000-010FFF U12, U13- $005000-005FFF U36, U37= $011000-011FFF U14, U15= $006000-006FFF U38, U39= $012000-012FFF U16. U17= $007000-007FFF U40, U41= $013000-013FFF U18, U19= $008000-008FFF U42, U43= $014000-014FFF U20, U21= $009000-009FFF U44, U45= $015000-015FFF U22, U23= $00A000-00AFFF U46, U47= $016000-016FFF U24, U25= $00B000-00BFFF U48, U49= $017000-017FFF $000400 THRU $000FF7 IS DECODED FOR THE EXPANSION CONNECTOR $000FF8 THRU $000FFF IS DECODED FOR THE ON-BOARD I/O $018000 THRU $FEFFFF IS DECODED FOR THE EXPANSION CONNECTOR $FF0000 THRU $FFFFFF IS DECODED BACK INTO ZERO PAGE ON BOARD
BYTE MAGAZINE: We have just received the Jan. '82 issue, and we would like to call certain items to your attention. On pages 54, 55 and 298 is yet more evidence that 8088 processors, including the IBM machine, are in fact running 8080 code at 8080 speeds. On page 223 is an advertisement for an 8086 processor card for the Apple II. This advertisement claims a performance advantage of 4 over the 1MHz 6502 used in the Apple (not very far from our personal estimate). How do we explain this disparity? Simple! Assembly language programmers are, by and large, not stupid people.
The installed base of 8080 (and Z80 and 8085 etc.) personal computers on 1 Jan. '82 was over one million. The installed base of 8088 machines was under ten thousand. No programmer in his right mind is going to write assembly language programs for the 8088 which cannot ALSO run on 8080 based machines! This is in addition to all the OLD 8080 stuff that is now being translated to run on the 8088.
This means that if you buy an 8088 based machine you can absolutely depend on running 8080 code unless you happen to be a good assembly language programmer yourself. And, if you have the ability to write your own assembly language programs, wouldn't you rather use an 8MHz 68000 than a 5MHz 8088?
HAPPINESS IS SOURCE CODE INCOMPATIBILITY: By a great stroke of good fortune, the indexed address modes of the 6502 are sufficiently different from those in the 68000 that it is absolutely impossible to translate 6502 source code into 68000 code. This means that ANY SOFTWARE WHATEVER which runs on the 68000 will be based on 68000 code! (We know that last sentence sounds ridiculous, but the same is NOT true for the 8086 or 8088.) When we say the indexed addressing modes of the 68000 are different, we mean different as in MUCH BETTER.
PRICING: You have probably noticed that every time you go to the store, the price of milk or hamburger has gone up a few cents again. Although a manufacturer of electronic equipment can't adjust his prices weekly, they DO have to be adjusted. We change the prices of our environmental noise monitors twice yearly, on Jan. 1 and July 1. Last July, we raised our prices 7% and this month we are raising (the noise monitor) prices about 5%. Since the noise monitors are mature products, the price changes reflect the prevailing inflation rate.
The DTACK GROUNDED 68000 boards are definitely NOT mature products. Our manufacturing cost for the Pet/CBM version is slightly higher than we anticipated, plus we now provide an additional 4K EPROM (more on this later). So the base price of the Pet version is going up $45 or 10%. If we subtract $15 for the extra EPROM, the change is 6.7% or about 2.2% greater than the current (6 month) inflation rate.
Our manufacturing cost for the Apple version is LESS than we expected, so the price drops $5. This is a price drop of 5.5% in constant dollars! We have also dropped the prices slightly for the 68000 and for extra RAM.
These price changes are consistent with the predictions on page 10 of issue #2 of this newsletter.
DTACK GROUNDED PRICING effective 15 Jan '82 RAM WITHOUT 68000 WITH 68000 4K $495 $695 12K $567 $767 28K $710 $910 60K $995 $1195 92K $1195 $1395
The above prices are for both the Pet/CBM and the Apple II compatible version. Note that we have added a 12K model. The Apple II version is electrically and physically compatible with the Apple III version, but WE do not plan to offer software support for the III.
THANKS, COMMODORE: We now have written permission to provide modified versions of the Pet/CBM BASIC ROMS for the purpose of improving the performance of the computers! Therefore, the Pet/CBM version now includes a replacement for the BASIC ROM which contains the floating point routines. This replacement, a 4K 2532 EPROM, contains the 'hooks' which link to the 68000 floating point package.
EQUIPMENT STATUS: We now have three (main) computer systems. The Wang 2200 it now entering its tenth year (!) of service, thanks to a certain 5 1/2 year old attached processor. (We KNOW that an updated attached processor can extend the useful life of an obsolescent computer because WE HAVE BEEN THERE!). The attached processor is a 56K 2MHz 6502 machine. We have two Shugart 8 inch floppies, 256K each, plus a Teletype Model 40 printer. The printer kicks out a printed page every 15 seconds when needed, and in fact is so fast that we drive it with the 6502 rather than directly from the Wang 2200. Talk about BIG print buffers!
Our second computer, obtained about a year ago, is a CBM 8032 with the 8050 disk drive. The 8050 has MORE than 1/2 megabyte per single sided 5 1/4 inch disks, more than a megabyte on line. An Epson MX100 is currently shared with our third computer. Oh, yes: the 8032 has a 92K 68000 attached processor!
We bought our third computer last month. This is a 48K Apple II plus with a 16K ram card, two DISK IIs, an MX100 printer and a 68000 attached processor.
(IN)COMPETENCY STATUS: If we don't know how to use the Wang by now we sure aren't going to admit it! However, we are still learning Microsoft BASIC and trying to learn the disk operating systems of the CBM and Apple systems. Trying slowly and with little success, we might add. Look, have you ever had to use three different keyboards and three different operating systems all in the same hour? Not for fun (?) mind you, but because the job requires it?
On the other hand, we know quite a bit about the Microsoft BASIC interpreter, including the fact that the CBM and Apple interpreters are essentially identical. It helps that parent company Digital Acoustics has lots and lots of experience with 6502 assembly language, and that we have been able to disassemble ALL of the 8032 interpreter - that's 18K of source code - and annotate the parts we're interested in the most, like the floating point package. Did you know that the CBM and Applesoft floating point packages are absolutely, completely identical?
SIDEBAR: There are lots of Pet assembly language experts and lots of articles in print written by those experts. Ditto for the Apple. These experts, and their readers, pass silently in the night, assuming that their respective computer is the only one in (their) world. As a matter of fact, almost EVERY useful assembly language utility or extension to the BASIC interpreter that works on one machine can be readily adapted to the other!
For example, in the Oct. '81 issue of Commodore magazine, there are some neat machine language utilities beginning on page 56 which can be readily adapted to the Apple. It will probably come as a shock to Apple diehards to find that Pet stuff is useful. If so, be advised that it will shock Pet users to find something useful FOR THE PET in an official Commodore (U.S.) publication!
SINCERE THANKS DEPT: Like we admitted above, there is lots we don't know about the Apple operating system. After we got the MX100 printer, we had to call one of our newsletter subscribers (in a different area code, yet) before we were able to list our hand assembler's helper program! The fact that we now have the 68000 linked into the Applesoft interpreter is due more to the very useful assistance of several friends and/or customers. Without that help, we would still be spinning our wheels come summertime, more than likely.
The Apple compatible hardware was up and running for five weeks before we were able to generate enough documentation and demonstration software to be able to ship boards to the general computer hacker (we shipped boards earlier to certain industrial type customers who did not need that support).
EDITOR'S REPORT: This is our first attempt at publishing a newsletter. If we had known how many interesting persons we would meet (in person, over the phone and by mail) we would have done this long ago!
Of course, it is the VERY interesting hardware - the 68000 and its friends like the 8087 - that is responsible for whatever popularity this journal has. For that reason, your editor has chosen to remain, within these pages at least, anonymous. Personalities just don't count when you have sixteen 32 bit registers!
We had been afraid that DTACK GROUNDED, which is itself a maverick technical publication, would attract an excessive number of weirdos, but this has not proven to be the case (maybe they are all busy writing newsletters).
NO PHONE NUMBER: There are many more people looking for information about the 68000 and its use with the Apple than there are people who HAVE that information (and have time to chat on the phone). Hence, this newsletter, which you can read at YOUR, not OUR, convenience.
We will shortly begin our second round of advertising, this time with a photograph of our boards. The advertisements will list a phone number. The phone number will connect to an order desk.
TWO PLUS TWO IS FOUR (SURPRISE!): A local area customer of ours timed an Applesoft program which draws concentric circles in HIRES graphics at ten minutes. The compiled version of the program runs in three minutes. When he aided the one line of code to his BASIC program that hooks the floating point into our 68000 board and re-compiled the program, it ran in 55 seconds!
Note that the time saving due to the 68000 is not particularly impressive in this case, a saving of just over two minutes out of a total of ten. But when this saving is added to that due to the compiler, the result was spectacular! We sure hope you aren't surprised.
A PROSPECTUS: The nature of this newsletter is going to change considerably over the next six issues. We have up to now mostly discussed potential performance for the 68000. Henceforth, we will mostly report on ACTUAL performance and applications of the 68000. Nuts and bolts stuff. We will continue to cover OTHER high performance processors.
PIONEERS: Pet/CBM compatible DTACK GROUNDED board serial number 00001 went to Gordon Smith of Rockwell. Gordie is a noted Forth addict. The Apple compatible serial number 00001 board went to Jack Fester of the (original) Computer Store in Santa Monica.
We use five digits on the Pet version as an inside joke and five digits on the Apple version because we may need them.
SOFTWARE STATUS REPORT: We have managed (with a little help from our friends) to hook our 68000 board into the Applesoft interpreter! Oh. You heard already?
PRODUCTION STATUS: We are now practically current for both the Pet/CBM and Apple versions of our boards. In other words, all of the boards which were ordered prior to 5 Jan. have been shipped as of 15 Jan. (Okay, so this issue IS two weeks late. You have maybe heard of holidays?)
We expect to have no trouble keeping up with orders during the first three months of 1982. That's as far ahead as our crystal ball works!
HARDWARE STATUS REPORT: During the past month we have not done a darned thing on the expansion board. The expansion board has very definitely NOT been dropped or even delayed because of technical problems. Its just that all of our time has been spent on software and documentation for the Apple compatible board.
However, the front page of this issue can be construed as a very broad hint to OTHER hardware hackers who have OTHER computers. We will soon (maybe issue #8?) announce a kit for those who want to tie into other computers. This kit will contain schematics and other such hopefully useful stuff.
ACKNOWLEDGEMENTS: Apple; singular, II, III and soft are trademarks of the Apple Computer Co. Pet and CBM are trademarks of Commodore Business Machines. TRS 80 is a trademark of the Tandy Corp. and DTACK GROUNDED is OUR trademark. On the front page of this issue we are infringing on so many people that it will be simpler to let them sue us than try to acknowledge all these trademarks
Now did the human race ever survive before lawyers were invented?
SUBSCRIPTIONS: $15/6 issues U.S. and Canada, $25 U.K. or Germany. We use strictly first class mail. Payment should be made to DTACK GROUNDED. The subscription will start with the first issue unless otherwise specified. The address is:
1415 E. McFADDEN, St. F
SANTA ANA CA 92705
FAREWELL, MAYBE: In the past six issues of this newsletter we have tried to provide candid information on the performance, potential and limitations of our 68000 boards. We don't know, right now, exactly how many of our subscribers have considered this information judiciously and decided that DTACK GROUNDED is not their 'cup of tea'.
For these subscribers we hope the substantial amount of information which was not directly related to our commercial endeavour was worthwhile. If we have not met even that standard of performance, well, you pays your money and you takes your chances. If this is the only fifteen bucks you have ever wasted, you are ahead of the game!
If you are one of the three people who has gotten his money's worth, your reward is you get to send us another check.
REDLANDS: Like the recent Miami-San Diego football game, this issue has gone into overtime. REDLANDS will resume next month.