29:
304:. If one unit was not fabricated properly, it would be switched out through on-chip wiring and another correctly functioning copy would be used. All critical gates would be produced in triplicate, and after fabbing the system would be tested to determine which were working using a "voting" system, if any circuit disagreed with its two partners it would be considered broken. That circuit would then be disconnected using a
296:(WSI). Instead of making many small boards and wiring them together inside the computer, they would all be printed on a single wafer which would carry the connections patterned using the same process. The design called for a computer chip that was 2.5 inch on one side. At the time, computer chips of only 0.25 inch on a side could be reliably manufactured.
263:(ECL) transistors in chip form at scales that had not been accomplished before. Previous systems, like the System/370, managed up to 35 gates on a chip, using Amdahl's design, Fujitsu was able to reliably produce 100 gates on a chip. This made the resulting system smaller and cheaper and able to run at faster clock speeds.
276:
267:
card would fail, so there was pressure to use more, simpler boards. However, doing so increased the number of boards, to the point where inter-board delays made timing constraints very difficult to achieve. Fujitsu decided to move ahead with the concept even though Amdahl himself was increasingly against the design.
377:
Trilogy
Systems was known as one of the largest financial failures in Silicon Valley before the burst of Internet/dotcom bubble in 2001. In describing the company, financial columnists coined the term "crater" as describing companies that consumed huge amounts of venture capital and later imploded to
299:
The downside to this approach was that if any one of the "circuit boards" was non-functional, the entire wafer would have to be discarded. The chance this would happen would approach 100% at the complexity levels involved. As with other WSI projects, Trilogy's chip design relied on redundancy, that
266:
In the late 1970s, Amdahl began the process of designing the replacement for their 470 series. The new machine, the 580, had 50 chips on a board and 77 boards in total. There was an inherent catch-22 in this design; by placing more components on a card they were increasing the chances that a given
335:
plant was damaged during construction by a winter storm. The redundancy schemes used in the design were not sufficient to give reasonable manufacturing yields. The chip interconnect technology could not be reliably manufactured as the layers tended to delaminate and there was no automated way to
327:
insulation that allowed for extremely dense packing of signal wiring. Though overall system power consumption would be lower, the power dissipation would be much more concentrated at the single large chip. This required new cooling techniques such as sealed heat exchangers to be developed.
287:
In 1980, he decided he had lost control of the company and decided to leave. Still interested in the compatible mainframe market, Amdahl formed
Trilogy in 1980. His idea was to solve the complexity problem by producing all of the circuit boards on a single
362:
The new leadership redirected the company to be a technology provider to other computer companies. The only major customer was
Digital Equipment, which paid $ 10 million for the rights to the interconnect and cooling technologies. These were used for its
322:
Alongside the advances in chip manufacturing, advanced chip packaging techniques were also pursued by the company. These included vertical stacking of computer chips and chip-to-chip interconnect technology that used copper conductors and
331:
The company was beset by many problems. Gene Amdahl was involved in a car accident and preoccupied with the ensuing lawsuit. Madden, the company's president, died from a brain tumor. Their
513:
359:
By mid-1984, the company decided it was too difficult to manufacture their computer design. Gene Amdahl stepped down as CEO and Henry
Montgomery was brought in as replacement.
347:
and raised $ 60 million. By this time manufacturing was improving and it was common to get three or four "quadrants" that ran property. But by this time it was apparent that
508:
498:
351:
design would deliver chips with similar performance within a year or two, and that now, after about six years of effort, there was little reason to continue development.
503:
493:
370:
At the end of 1985, Gene Amdahl, as company chairman, decided to stop all
Trilogy development and use the remaining $ 70 million of the raised capital to buy
518:
523:
319:
manufacturing tolerances over the large chip) than standard-size chips. Consequently, logic density and performance were less than had been forecast.
367:
mainframe computers. Years later, the manufacturing difficulties of the copper/polyimide technology restricted DEC's ability to ship its mainframes.
50:
72:
201:
477:
43:
37:
332:
256:
398:
54:
17:
344:
293:
280:
260:
153:
188:, Gene Amdahl was able to raise $ 230 million for his new venture. Trilogy was the most well-funded
180:
was a computer systems company started in 1980. Originally called ACSYS, the company was founded by
16:
This article is about the bankrupt computer systems company. For other companies named
Trilogy, see
241:
221:
185:
224:
that was both cheaper and more powerful than existing systems from IBM and Amdahl
Corporation.
244:. Using many separate boards allowed them to remove malfunctioning circuits and replace them.
316:
189:
184:, his son Carl Amdahl and Clifford Madden. Flush with the success of his previous company,
468:
Wagner, K. (April 1997). "Adventures in the
Mainframe Trade: Inverview with Gene Amdahl".
301:
416:
251:
at lower cost while also running faster. Much of this was due to his collaboration with
248:
193:
487:
233:
213:
92:
374:, a minicomputer start-up company. In 1989, Gene Amdahl left the merged company.
312:
209:
197:
181:
168:
247:
At Amdahl
Corporation, Amdahl was able to produce systems compatible with the
237:
232:
Large computers of the 1960s and 70s were physically constructed using small
337:
324:
364:
252:
205:
98:
480:/ Fortune Magazine article on Trilogy's history, September 1, 1986
371:
305:
289:
274:
143:
348:
275:
311:
The large chip size demanded larger minimum dimensions for the
217:
22:
440:
438:
436:
434:
212:and others. The plan was to use extremely advanced
162:
149:
138:
123:
105:
90:
514:Defunct computer companies of the United States
8:
419:. Computerworld. 15 Jun 1981. pp. 11–12
85:
417:"ACSYS"&pg=PA11 "Trilogy Systems Corp"
84:
509:Computer companies disestablished in 1985
499:American companies disestablished in 1985
73:Learn how and when to remove this message
36:This article includes a list of general
18:Trilogy (disambiguation) § Business
387:
196:history. It had corporate support from
504:Computer companies established in 1980
494:American companies established in 1980
456:
444:
393:
391:
478:GENE AMDAHL FIGHTS TO SALVAGE A WRECK
216:manufacturing techniques to build an
7:
399:"In Memoriam: Gene Amdahl 1922-2015"
519:Defunct computer hardware companies
470:IEEE Design & Test of Computers
524:Defunct computer systems companies
42:it lacks sufficient corresponding
14:
378:leave nothing for its investors.
27:
171:, Carl Amdahl, Clifford Madden
1:
202:Digital Equipment Corporation
343:In 1983, the company had an
178:Trilogy Systems Corporation
111:; 44 years ago
540:
283:attempt by Trilogy Systems
236:populated with individual
15:
403:Communications of the ACM
333:semiconductor fabrication
257:semiconductor fabrication
345:initial public offering
294:wafer scale integration
281:wafer-scale integration
57:more precise citations.
284:
192:up till that point in
278:
261:emitter-coupled logic
259:lines to produce the
255:, who used their own
154:Cupertino, California
405:. November 12, 2015.
228:Amdahl leaves Amdahl
242:integrated circuits
87:
300:is replication of
285:
222:mainframe computer
186:Amdahl Corporation
175:
174:
83:
82:
75:
531:
473:
460:
454:
448:
442:
429:
428:
426:
424:
413:
407:
406:
395:
317:photolithography
302:functional units
190:start-up company
134:
132:
119:
117:
112:
88:
78:
71:
67:
64:
58:
53:this article by
44:inline citations
31:
30:
23:
539:
538:
534:
533:
532:
530:
529:
528:
484:
483:
467:
464:
463:
455:
451:
443:
432:
422:
420:
415:
414:
410:
397:
396:
389:
384:
357:
273:
240:or small scale
230:
165:
157:
130:
128:
115:
113:
110:
95:
86:Trilogy Systems
79:
68:
62:
59:
49:Please help to
48:
32:
28:
21:
12:
11:
5:
537:
535:
527:
526:
521:
516:
511:
506:
501:
496:
486:
485:
482:
481:
476:Myron Magnet,
474:
462:
461:
449:
430:
408:
386:
385:
383:
380:
356:
355:Changing focus
353:
272:
269:
249:IBM System/370
234:circuit boards
229:
226:
194:Silicon Valley
173:
172:
166:
163:
160:
159:
151:
147:
146:
140:
136:
135:
125:
121:
120:
107:
103:
102:
96:
91:
81:
80:
35:
33:
26:
13:
10:
9:
6:
4:
3:
2:
536:
525:
522:
520:
517:
515:
512:
510:
507:
505:
502:
500:
497:
495:
492:
491:
489:
479:
475:
471:
466:
465:
459:, p. 10.
458:
453:
450:
446:
441:
439:
437:
435:
431:
418:
412:
409:
404:
400:
394:
392:
388:
381:
379:
375:
373:
368:
366:
360:
354:
352:
350:
346:
341:
339:
334:
329:
326:
320:
318:
314:
309:
307:
303:
297:
295:
291:
282:
277:
271:Trilogy forms
270:
268:
264:
262:
258:
254:
250:
245:
243:
239:
235:
227:
225:
223:
219:
215:
214:semiconductor
211:
207:
203:
199:
195:
191:
187:
183:
179:
170:
167:
161:
158:United States
155:
152:
148:
145:
141:
137:
126:
122:
108:
104:
100:
97:
94:
89:
77:
74:
66:
63:December 2015
56:
52:
46:
45:
39:
34:
25:
24:
19:
469:
452:
447:, p. 9.
421:. Retrieved
411:
402:
376:
369:
361:
358:
342:
330:
321:
310:
298:
286:
265:
246:
231:
177:
176:
150:Headquarters
142:merged into
101:: TRILF
69:
60:
41:
457:Wagner 1997
445:Wagner 1997
313:transistors
292:, known as
238:transistors
220:compatible
210:Sperry Rand
198:Groupe Bull
182:Gene Amdahl
169:Gene Amdahl
55:introducing
488:Categories
423:4 December
382:References
164:Key people
38:references
338:soldering
325:polyimide
279:An early
139:Successor
93:Traded as
365:VAX 9000
340:errors.
315:(due to
472:: 5–13.
336:repair
253:Fujitsu
129: (
124:Defunct
114: (
106:Founded
51:improve
206:Unisys
99:Nasdaq
40:, but
372:Elxsi
306:laser
290:wafer
144:Elxsi
425:2015
349:VLSI
131:1985
127:1985
116:1980
109:1980
218:IBM
490::
433:^
401:.
390:^
308:.
208:,
204:,
200:,
427:.
156:,
133:)
118:)
76:)
70:(
65:)
61:(
47:.
20:.
Text is available under the Creative Commons Attribution-ShareAlike License. Additional terms may apply.
