THE WJUNN

The Open Channel is exactly what the nameimplies: a forum for the free exchange oftechnical ideas. Hold your contributions toone page maximum in the final magazine for-mat (about 1000 words-less, if you want toinclude illustrations).We'll accept anything (short of libel or ob-

scenity) so long as it's submitted by a memberof the Computer Society. If it's really bizarrewe may require you to get another member tocosponsor your item.Send everything to Jim Haynes, Applied

Sciences, UC Santa Cruz, CA 94064.

Stack Overflow Dept. Personals

Several months ago a man calledme from SDC in New York wantingto know more about the articles wehave run on interrupt-less systems.Not having anything readily at hand,I took his name and address forentry into the write-only memoryon top of my desk and promised tosend something later. The name andaddress are now gone forever; so,whoever you are. I hope you read this.The first article, "Please Don't

Interrupt Me While I'm Computing,"appeared in the December 1973issue of Computer. Ornstein's letterwas in the June. 1975 issue. Ratt-ner's letter was in the March 1975issue. Ornstein et al had a paperin the 1975 NCC Proceedings. Ratt-ner referred to a paper on the PlesseySystem 250 in Proceedings of the1972 IEEE International SwitchingConference. I mentioned a paper onthe Honeywell Series 60, Level 64,that was printed in the terminal issueof Honeywell Computer Journal.

Pat Skelley and I discussed thepossibility of putting on a workshopon interrupt-less systems (he thoughtplenty of people would come, if onlyto denounce the concept). Wecouldn't think of anyone who wouldbe gullible enough to do all thework of organizing the workshop,so that's where it sits.

B-1700!Brian Converse

WANTED: algorithms that runon pocket calculators. Wouldlike to start a newsletter orsomething to circulate them in.Send to Ira Chayut, 3030Brighton 12th Street, Brooklyn,NY 11235.

Exclusive! COMPUTERMagazine's Predictionsfor 1976!

In 1976 we predict that severalcompanies will begin delivering sys-tems made up of networks of micro-processors. These will allow eachmajor element of a system to bemore complex in function than hasbeen typical in past systems. Func-tions which have been performed bysoftware modules in operating sys-tems will be distributed among thesystem elements, to be performedlocally and concurrently.These systems will work very well

until 1978 (one of our consultantswill only allow until 1977). Then,one by one, they will begin tocollapse. Nobody will be able to fixthem. The system manufacturer will

say, "But microprocessors are socheap that we can just pull out thebad one and throw it away!" Theservice technician will say, "Whichone of these 19,572 microprocessorsshould I throw away?" The customerwill reply, "All of 'em."

Late in 1978 the company thatbrought us the 512 x 8 static RAMon one chip will introduce a 512 x 9bit model. (Look who's.discoveredparity now!) Nine-bit-wide ROMs willsoon follow; and within a year therewill be a new microprocessor havinga matrix of LED's on the cover thatspell out, in appropriate red andgreen,

I AM

.ee... S.... SO..

Another COMPUTERMagazine First!(and probably last)

PRODUCT TEST REPORTS

Microcomputer Associates Inc.,who market the JOLT system, sentus one of their CPU kits. This con-sists of a double-sided PC board, aMOS Technology microprocessorchip (MCS6502), a peripheral inter-face chip (AMI or Motorola 6820),512 bytes of RAM (on four 2111chips, each 256 by 4), a MOS Tech-nology multifunction chip (MCS6530, containing a 64 by 8 RAM, a1024 by 8 ROM, a programmablecounter, and some I/O port hard-ware), some ordinary IC's (two linears,and one each of TTL, CMOS, andlow-power Schottky TTL), and asmall handful of discrete components.Sockets are furnished for the three40-pin chips and the 2111 RAMs.The ROM portion of the 6530 chipcontains a (presumably proprietary)utility program which drives a serialASCII terminal and allows the userto examine and modify the contentsof registers or memory, begin execu-tion of a program at any address,initialize certain RAM locations,handle interrupts, punch a memorydump tape, load memory from sucha tape, punch a control tape for aROM programmer, and use a high-

January 1976 83

speed tape reader. A user programcan call subroutines located in theROM to get a character from theterminal, print a character to theterminal, print a space, print a carriagereturn and line feed with an arbi-trary delay, type a byte in hexadeci-mal, or get a character from a high-speed reader. The user's serial termi-nal can have a current-loop (20 ma.),EIA (polar voltage), or TTL voltageinterface and operate at a speedbetween 10 and 30 characters persecond. (The program, using theprogrammable counter, measuresthe terminal's speed on the first inputcharacter and takes care of serializ-ing and deserializing thereafter.)

Construction: I spent abouthalf an hour reading the literaturethat came with the kit, and then anhour of actual assembly time. Theassembly instructions will seem sparseto one who is expecting Heathkit-type hand-holding, but they shouldpresent no problem at all to anyonewho is at all familiar with electronicparts and construction. Fully 10minutes of the construction hourwere spent in trying to coax the three40-pin chips to bed down in theirsockets without curling up theirlegs. We won't count another 10minutes after I discovered that inspite of all the markings I had putthese three chips in backwards.Some more lost time resulted fromthe chronic shortage of 25-pin EIAinterface connectors around myplace; but in this business we learnto be pretty resourceful with a cup-uppaper clip. Since my householdfurniture includes an 8-level Teletypeand assorted power supplies, I wasready for testing within 2 hours ofopening the kit.

Turned power on, reset, and hita carriage return per instructions.Instead of getting the register con-tents typed out I got a continuoustypeout of mostly the same letterover and over. Brought the thingdown to work the next morning andused the oscilloscope to find thatthe microprogram was at least goingthrough the right addresses in theloop where it waits for the firstcharacter to come in so it can meas-ure the speed. After the carriagereturn the program is a lot morecomplicated so I couldn't really tellwhat addresses were coming up on

the bus. Noticed the logic voltageon the signal from the terminal wasabout 1.2 volts. It was necessary toreduce the value of R13 by abouthalf to get the proper input voltagewith a loop-current terminal. Lookedat the signal going to the terminaland found much shorter pulses thanare appropriate for the bit rate. Leftthe thing for the night.

Next morning there was somereasonable-looking print on the Tele-type and a note from a student thatif I would press on certain pins onone of the RAMs it would work aslong as power was left on. Pulledthe chip, looked at the dubious soc-ket contacts, and scraped and bentthe leads. After that it worked OKeven if power was turned off andback on. (Student said he had alreadytried those things, but then every-thing this guy builds seems to haveIC socket problems. Suspect he han-dles the chips while eating a jellydoughnut.) Has worked fine sincethen.

I haven't had time yet to try any-thing other than trivial programs; butthe monitor utility seems to work asclaimed and makes it reasonablyeasy to load and run small programs.The literature that comes with the kitis adequate for using the CPU boardwith a terminal as a stand-alone com-puter; but to do anything with otherperipheral equipment you will needmore hardware documentation. Thishas to be obtained from MOS Tech-nology, or perhaps from Motorola inthe case of the peripheral interfacechip.

Comments: The two-page adin the October Computer claims thatJOLT is the "world's lowest-costcomputer system." Today a claimlike that is likely to be obsolete beforeit gets into print. JOLT is much morethan a handful of IC's as they comefrom the manufacturer, but in itspresent state of evolution it is less ofa system than some of the higher-priced competing products on themarket that offer a variety of per-ipherals. (As advertised, the JOLTfamily is limited to the CPU boarddescribed above, a 4K RAM board,a power supply, a peripheral inter-fact board, and a universal boardfor the user's own circuitry.) Iwouldn't offer such a technicallycomplicated product for homeassembly by hobbyists even at three

84

times the price. It's too easy to geteaten alive by the costs of servicingcustomers whose kits don't workafter assembly.

There are quite a few "kits" onthe market today, at a variety ofprices, from various vendors: ICmakers, independent suppliers to theindustrial market, independent sup-pliers to the hobbyist market, andtraditional parts distributors. Hereare some of the things you can get,depending on vendor.

a. Software: cross-assemblers,microprocessor simulators that runon batch computers or commercialtime sharing services; assemblers,interpreters, compilers, utility rou-tines that run on the microprocessorsystem, if suitably configured.

b. Loose parts: a collection of LSIchips and other parts needed to get amicroprocessor going without end-less shopping around.

c. Packaging hardware: perhaps auniversal PC board where you doall the wiring, or a board that pro-vides the wiring for a small processor-memory-peripheral configuration, ora system that could include consolepanels and cases as well as the meansfor building an arbitrarily-largesystem.

d. Engineering: ranging from aschematic showing how to hook up afew parts to make a minimum systemto a complete system plan that pro-vides for standard peripherals andsoftware support.

e. Peripherals: ready to plug in, inaccordance with a set of packaginghardware and a system plan.

f. Complete microcomputer sys-tems ready to plug in and use.As noted earlier the JOLT system

provides for a one-board minimummachine, a power supply, 4K bytememories, a universal PC board,and a utility monitor implemented inROM. The suggested mechanicalarrangement is to stack the severalPC boards of a system with threadedmetal spacers through holes in thecorners of the boards. Inter-boardconnections are made with flat cableand connectors. There is presentlyno case, chassis, or front panel withlights and switches. The flat cablebus system and the PC cards save alot of time in assembly, since it takesjust one operation to put a connectoron a piece of flat cable and youhave 40 wires connected. I suppose

COMPUTER

the roughly one-hour job of assem-bling the CPU board would havetaken all day if I had been forced touse hand writing, either solderingor wrapping.

W hether the JOLT machine is thebest for your own needs, and whetherit is a good value for the price, issomething you will have to decide foryourself. If you want to work with theMOS Technology chips and alsodevelop some of your own hard-ware to use with it, then the JOLTwill get you on the air quickly; andthe time may be worth a lot of moneyto you. If you are more interestedin a system with a box and frontpanel, the new Altair kit using theMotorola/AMI chip might be a betterbuy. But in this product you willhave a lot less freedom to diddlewith the hardware design and addthings of your own choosing. (Also,for some reason, Altair seems to runthe chip a lot slower than its adver-tised capability.) If you don't have ause for the JOLT memories, powersupply, or universal PC card, thenyou might get along with the Motor-ola kit quite well. It now includes aPC card to build a minimum sys-tem and sells for only $149. Thelimitation here is that the packagingsystem consists of only the one PCcard. So far as I know you can'tget any others; and while I haven'tseen this one, previous experiencewith Murphy's Law leads me tospeculate that it is probably an oddsize that doesn't fit into any readily-available packaging system.

If you have or want to developyour own packaging, or if you wantto use distributor off-the-shelf uni-versal PC cards and are willing to doall the wiring, there are the Cramerkits. These give you a schematicto go by and save you the agony ofhaving to contact a dozen differentdistributors to get all the necessaryparts. It isn't clear to me, though,what else is in the kit to justify the$500 price tag.These are intended only as exam-

ples; there are many, many thingson the market these days in manydifferent price ranges.My personal reaction is that the

JOLT is a nice product, easy toassemble, and pleasant to use. Thepackaging has some good points,but overall is too primitive for mytastes; and the prices of some items

January 1976

are a little hard to swallow. Like thead says, I thought the price of $415for the power supply was a mis-print-but probably not in the waythe ad writer had in mind. You canstill buy a neat little 5-volt, 3-ampsupply ready-made by Wanlass orseveral other firms for 25 or 30 bucks.(The CPU board needs only a single5-volt supply if you can use TTLinterfaces to the terminal; the othertwo voltages are needed for current-loop and EIA interfaces.) And theuniversal PC card at $25, plus $40for a bag of things to connect itelectrically and physically to itsneighbors seems high-priced in com-parison to what you can get fromVector or Vero or others. In anycase, we can't complain of a lack ofvigorous competition in this particu-

Recently published books and newperiodicals may be submitted forreview to the Book Reviews Editor;Dr. Francis P. Mathur, College ofEngineering, Bioengineering andAdvanced Automation Program,University ofMissouri-Columbia,Columbia, MO 65201.

B76-1 Microprogramming and Sys-tems Architecture, Infotech Stateof the Art Report -23, InfotechInformation Ltd., Maidenhead,Berkshire UK, ($141 individualreport, $94 subscription rate for8 volumes), 1975 644 pages.

Edited by C. Boon, this is oneof the latest volumes in the Info-tech series of State of the ArtReports. It examines the roles ofmicroprogramming in systemsarchitecture and how these rolesare being affected by technologicaldevelopments.The organization follows the

pattern well established in earlierreports: key extracts from invitedpapers, conference presentations,and dicussions are linked in aneditorial framework designed tobring out the essential points andarguments in a systematicallyreadable and djigestible analysis.It is followed by the full text ofconference and invited papers, andan integrated bibliography.

lar market (and in power supplies, too).

All this is simply one engineer'sexperience with one member of anew class of products that are rele-vant to many of our readers. Wewant to thank Microcomputer Associ-ates, Inc. for making the hardwareavailable. It is obviously beyond ourcapability to produce a compre-hensive report on the whole field.But with microcomputers prolifer-ating and starting to become house-hold appliances maybe it won't belong before we can read about themin the publications of organizationssuch as Consumers Union. Mean-while, visit your local supermarketand flip through the pages of someelectronics-hobbyist magazines.You might be surprised.

-BOOK<

The analysis section (155 pages)deals with the roles of micropro-gramming in systems architec-ture, approaches to micropro-grammed control (horizontal andvertical, emulation, high-level lan-guage support, operating systemsupport, system maintenance,microcode production, user micro-programming), and finally the in-fluence of technology.The report observes the since

microprogrammed control waschosen by IBM for their Sys-tem/360 range of machines, micro-

85

-

'l-j

JI