1. Chapter 6 1965 - 1975 The Chip & Its Impact First Microprocessor Chip 1

2. Grosch’s Law A big system that costs twice as much as a small system actually gets you 4 times computing power 2

3. Why Grosch’s Law? Memory & Circuitry costs  cost / bit higher for small computers Connecting of small computers not efficient Held thru 1960’s 3

4. The Chip- 1959 Drove minicomputer IC replaced discrete transistors, resistors, core memory Next step in “modules” Printed Circuits Cheap materials & Mass Produced 4

5. Integrated Circuit - 1959 Jack Kilby, Texas Instruments  Nobel Prize, Physics, 2000 Robert Noyce, Fairchild Electronics, CA Solved “Tyranny of Numbers” problem Driving forces  Military - smaller, reliable missiles & rockets (many failures)  Civilian - cheaper, less errors, hand work 5

6. Molecular Electronics - USAF Goal: New device of substances whose individual molecules did the switching 1959 – Westinghouse – grants of $2 mil & $2.6 mil For rockets, ballistic missiles 2 years later, quietly dropped USAF - “clean” rooms, early 1960’s, for minuteman missiles, detailed records Demanded high reliability (Hi-Rel) from all suppliers 6

7. Jack Kilby 1957- Centrallab  Printed circuit boards  Germanium to silicon  Did not have resources for development 1958 – TX Instruments, Dallas  Microminiaturization  Make all components of silicon or germanium  More expensive than carbon, ceramics  All components same material as transistor - silicon Would allow one set-up not many Nobel Prize 2000 7

8. Jack Kilby Built circuit of silicon, then germanium  Work as oscillator  Built by hand  Resistors, capacitor Applied for patents - 1959  Awarded 1964  Drawings Pg. 184 -185  “Solid Circuit” 8 First IC

9. Robert Noyce Fairchild Semiconductor, CA Heard of Kilby’s design Designed same idea on silicon Photo-etching process (Jean Hoerni)  Flat transistor, “planar process” Silicon  best insulator  Layers isolate devices Applied patent - 1959  Shared credit, Noyce’s process most significant 9

10. I.C. Progress Neither “inventor” had connection to USAF Military & Aerospace industry provided market  “advanced” Minuteman II missile  2K IC’s + 4K DC’s vs. 15K discrete circuits  Flew 1964  Established acceptability of IC & volume production lines  TI, Westinghouse, RCA - 4,000 / week  Noyce “military stifles research” Military use: 1963 – 100% IC, 1964 – 95% IC 10

11. Apollo (NASA) Fairchild provided IC’s Apollo computer  5,000 simple chips  Single-type  75 guidance computers built During project: chip price dropped from $1,000 to $20 - $30 11

12. Chip into Commercial Computers IBM 360 – solid state ceramic circuitry SDS and RCA announced silicon IC computers IBM 370 – silicon IC’s Quickly used in Minicomputers 12

13. 2 nd Generation Mini’s DEC dominant, not like IBM 1968-1972 - 100 companies offered minis  Did not need excessive capitol - see Pg. 192 Increase in performance; decrease in cost Driving forces  Standardized, Inexpensive chips  TTL- transistor- transistor logic Today’s “caterpillar” packaging  Printed Circuit Board (standardized) Pioneered by Globe-Union “stuff” chips in board; molten solder on back 13

14. Minicomputer Developments DMA (early) - DEC 16 bit word length; 8- bit byte ASCII (not IBM’s EBCDIC) CCC-DDP116 - 1 st 16- bit mini  Redesigned after IBM 360 announced  1966- Honeywell bought; withered 14

15. ARPA Defense Dept. – Advanced Research Projects Agency (ARPA) Funded projects 1967 - meeting on how to link computers in a network  across the U.S. To share resources among funded agencies 15

16. ARPANET Problem! All different computers Honeywell DPD- 516 @ each node as the interface between mainframe and N.W.  IMP- Interface message processor Dec. 1969 - 4 nodes West of Rockies 1970 - 10 nodes, spanned the U.S. 1971 - 15 nodes with 23 host computers Oct. 1972 - Demo’d in D.C.  30 nodes Dismantled in 1988 No longer need IMP 16

17. Data General Edson DeCastro @ DEC  Design for 16 bit mini, rejected by DEC  Formed D.G. – to build 16-bit mini 1968 – DG Nova Ken Olsen - DEC- claimed Nova was “copied”  PDP-X designed at DEC Used all newer innovations MSI - Medium Scale Integration 1971- Super Nova  IC for memory (RAM) - no core  1970 - in Illiac-IV, U of Ill., non-von Neumann (computer was failure)  Established viability of semiconductor memory (RAM) 17

18. In the Meantime… 1968 - Robert Noyce & Gordon Moore left Fairchild; also Andrew Grove Company to focus on memory Intel => “integrated electronics” 1970 - announced the 1103 chip  1024- bit dynamic RAM This + Nova  end of magnetic core 18

19. PDP-11 1969 - DEC not maintaining market (12-bit word, limited instructions)  PDP-8 was likely being replaced by competitors DEC decided to try 16 bit, again  Engineers and Carnegie - Mellon  Abandoned a years worth of work  Adopted an alternate design - McFarland Deliveries began early 1970 PDP-11 44 19

20. PDP 11 (cont.) New Innovation - BUS  “set of wires to serve all major sections of computers in common and standard way”  UNIBUS - 56 lines Earlier BUS (BUSS)  Whirlwind, Mark I, NOVA, PDP-8 20

21. PDP 11- Success From 5,800 employees in 1970 to 36,000 in 1977 Sold 170,000 in 1970’s 1970’s Recession in computing  DEC & Data General survived 21

22. Direct Access Computing 1950’s- IBM RAMAC  But still used tape for 10 years Time sharing was developing  370; GE/ Honeywell; Dartmouth 1960’s - 70’s  Cost dropped twenty fold  Capacity increased 40X 22

23. Magnetic Disks 23

24. Direct Access Computing (cont.) 1980 - fixed disk packs  IBM - Winchester- 2 x 30 mbytes  DASD CICS - IBM (Generic)  Allowed direct query from database Transformed retail sales by phone  Customers wrote own applications 24

25. Computer Science Education Batch prevailed for many years “load-and-go” compilers; MAD @ Michigan Waterloo U., Canada - Dept.-1962  WATFOR (version of FORTRAN for 360) 6,000 student jobs per hour Reduced cost- $10 to 10 cents per job For IBM 360/75  WATBOL 25

26. Dartmouth University John Kemeney - Math professor - 1963  Wanted system to teach interactive programming to all students  With Thomas Kurtz - developed BASIC & system to use it on GE 235 Slowly went to some other Universities  Schools began charging fees for time  Still used punch cards into 1980’s 26

27. RSTS-II on PDP-11 (Resource Sharing Time Sharing) Time sharing; no memory protection RSTS implemented in BASIC Modified BASIC  PEEK & POKE (Individual Bits)  Took less memory  Ready for PC’s “The mini generated the seeds of its own destruction by preparing the way for personal computers” (Ceruzzi) 27

28. Chapter 6 1965 - 1975 The Chip & Its Impact 28