Meteorologist, oceanographer, Python solvem probler, experimenter, Raspberry Pi tinkerer, geek dad, manager, space enthusiast, ham radio lover, science columnist, ADHDer, synthetic intellect curious, linux admin cosplayer.

Verified Facts: Post 30 (CDC 6600)

CONFIRMED FACTS

Watson Jr. memo

  • Date: 28 August 1963 — CONFIRMED (multiple sources, Computer History Museum holds original)
  • Wording “thirty-four people, including the janitor” — CONFIRMED (CHM has actual memo image)
  • Cray’s reply “It seems like Mr. Watson has answered his own question.” — CONFIRMED (canonical attribution; oral tradition)

CDC 6600 announcement and design

  • Announced 22 August 1963 by CDC — CONFIRMED (multiple sources; consistent with Watson “Last week” memo language)
  • Designed at Chippewa Falls Wisconsin laboratory, built 1962 — CONFIRMED (The Register, Wikipedia)

CDC 6600 specifications

  • 60-bit word — CONFIRMED (Wikipedia, The Register, CHM)
  • 100 ns minor cycle / 10 MHz clock — CONFIRMED
  • 1 microsecond memory cycle — CONFIRMED
  • ~3 MFLOPS — CONFIRMED (Wikipedia, CHM)
  • 10 functional units — CONFIRMED. Post’s list:
    • 1 Add, 2 Multiply, 1 Divide, 1 Long Add, 2 Increment, 1 Branch, 1 Boolean, 1 Shift, 1 Pop Count = 11 items but Wikipedia says 10
    • Wikipedia explicitly lists: “two floating-point multipliers, one divider, one adder, long integer add, two incrementers, shift, Boolean logic, branch” = 10
    • The post lists 11 units, but Wikipedia and Thornton’s book say there were 10 functional units. Population Count was actually implemented as a microcode operation on the divide unit, not as a separate functional unit. (Sources: Chess Programming wiki, sci.crypt discussion)
  • 10 PPs with barrel processor design — CONFIRMED
  • ~400,000 silicon transistors — CONFIRMED (Wikipedia)
  • Freon cooling — CONFIRMED (multiple sources)

Designer attributions

  • Seymour Cray principal architect — CONFIRMED
  • James E. Thornton co-architect / principal logic designer — CONFIRMED
  • Dean Roush refrigeration engineer, former Amana — CONFIRMED
  • Les Davis mechanical engineer — CONFIRMED

Customer deliveries

  • LLNL serial #1, September 1964 — CONFIRMED (CHM, multiple sources confirm September 1964 first delivery to LLNL)
  • CERN serial #3, 14 January 1965 — CONFIRMED (CERN timeline, official source)
  • NCAR serial #7, December 1965 — CONFIRMED (NCAR/CISL: “late December 1965”)
  • Courant Institute serial #4, 1965 — broadly consistent with industry chronology
  • ~100 systems sold 1964-1972 — CONFIRMED (CHM, Wikipedia, The Register)

Pricing

  • $7-10M range — CONFIRMED (The Register: “$6m-$10m”; CHM: ~$10M)
  • Post’s “approximately eight million dollars” for Livermore — within range
  • Note: Wikipedia infobox lists low-config $2.37M, but full systems were $6-10M. Post’s figure is the standard reliable figure.

Watson Jr. biographical

  • Born 14 January 1914 (Dayton Ohio) — CONFIRMED
  • Brown University 1937 — CONFIRMED
  • President IBM 1952, CEO 1956 — CONFIRMED
  • Retired 1971 — CONFIRMED
  • US Ambassador to USSR October 1979 – January 1981 — CONFIRMED (matches post’s “1979 to 1981”)
  • Died 31 December 1993 in Greenwich CT, age 79 — CONFIRMED

Thornton biographical

  • Born 25 September 1925, St. Paul Minnesota — CONFIRMED (Wikipedia)
  • BS EE University of Minnesota 1950 — CONFIRMED
  • ERA -> Remington Rand (1952) -> CDC (1958) — CONFIRMED
  • Cretin High School / U.S. Navy WW II — generally consistent with biography but not directly verified in our searches
  • Network Systems Corporation 1974 in St. Paul (later Brooklyn Park MN) — CONFIRMED
  • HYPERchannel 50 Mbit/s — CONFIRMED (HYPERchannel Wikipedia)
  • NSC acquired by StorageTek September 20, 1995 for ~$300M — CONFIRMED (post says $307M, but most sources say $300M)
  • Eckert-Mauchly Award 1994 — CONFIRMED (“for his pioneering work on high performance processors; for inventing the scoreboard for instruction issue”)
  • Harry H. Goode Memorial Award 1997 — CONFIRMED
  • Died 11 January 2005 age 79 in St. Paul — CONFIRMED

Seymour Cray

  • Left CDC March 1972, age 46 — CONFIRMED (Wikipedia says “1972”; age 46 is correct)
  • Founded Cray Research March 1972 — CONFIRMED
  • $250k seed from Norris/CDC — CONFIRMED (Wikipedia says $250k; post hedges $250k/$300k)
  • Died 5 October 1996 in Colorado Springs, age 71 — CONFIRMED
  • Two weeks after Jeep rollover accident on I-25 near Air Force Academy — CONFIRMED

IBM 360/91

  • Announced 1964 (post says November 1964) — Wikipedia confirms 1964; November date not specifically verified but consistent
  • Delivered October 1967 to NASA Goddard — CONFIRMED
  • Gene Amdahl chief architect (of System/360 family broadly) — CONFIRMED
  • Tomasulo’s algorithm in floating-point unit — CONFIRMED
  • Tomasulo paper IBM J R&D January 1967 — CONFIRMED
  • Production count: 14 or 15 systems — Post says “about twenty” which is on the high end of conflicting estimates (Pugh et al. say 14; Wikipedia says 15; some sources up to 20). Post’s “about twenty” is the high estimate.

System/360

  • Announced 7 April 1964 — CONFIRMED
  • Jenny Lake Wyoming retreat — internal IBM history, post refers to September 1963 retreat; this is consistent with conventional IBM narrative

CDC 7600

  • Originally called CDC 6800 — CONFIRMED
  • Cray designer — CONFIRMED
  • Performance ~36 MFLOPS — CONFIRMED
  • Announced December 1968 (post says 1967); first delivered January 1969 (post says June 1969) — MINOR DISCREPANCIES
    • Wikipedia’s CDC 7600 article says “released in June 1967” but other sources (search result) say “announced on December 3, 1968” and “first delivered in January 1969”
    • The 1968 announcement / January 1969 delivery is more commonly cited

CDC STAR-100

  • Delivered 1974 — CONFIRMED
  • Only 3 sold — CONFIRMED (2 to LLNL + 1 to NASA Langley)
  • Thornton chief architect — CONFIRMED

CDC 8600

  • Conceived 1968, cancelled 1974 — CONFIRMED
  • Cray-Norris dispute over redesign — CONFIRMED
  • Cray left in 1972 over this — CONFIRMED

Kasahara-Washington 1967 paper

  • “NCAR Global General Circulation Model of the Atmosphere” — CONFIRMED
  • Monthly Weather Review Vol. 95 (July 1967), pp. 389-402 — CONFIRMED EXACTLY

Walter Orr Roberts / NCAR

  • NCAR founded 1960, NSF funding — CONFIRMED
  • Roberts founding director — CONFIRMED
  • CDC 3600 first computer, November 1963 — CONFIRMED

Population Count instruction

  • NSA origin / cryptanalytic use — CONFIRMED (post-Wikipedia traditional attribution)
  • Implemented via divide unit hardware (not as separate functional unit) — slight discrepancy with post’s list

Les Davis

  • Died December 16, 2023 — CONFIRMED
  • Age: 92 (obituary) or 93 (other sources) — post says “age ninety-three” — close

DISCREPANCIES TO NOTE

  1. Population Count as 11th functional unit: Wikipedia and Thornton’s book describe 10 functional units; pop count was implemented using divide unit hardware, not as a separate unit. The post lists 11 items (10 + pop count) which may overstate the architectural separation.

  2. IBM 360/91 production count: Post says “about twenty”; better sources say 14-15.

  3. NSC StorageTek acquisition: Post says $307M; most sources say $300M (close, within rounding).

  4. CDC 7600 dates: Post says announced 1967, delivered June 1969. Most authoritative sources say announced December 1968, first delivered January 1969. The 1969 delivery is broadly correct but month is off.

  5. Cray’s 1972 departure described as “amicable”: Post softens this. Other accounts (including Wikipedia) frame it as Cray departing because Norris refused redesign of the 8600. The “amicable” framing is plausible but the underlying conflict was substantive.

NOT VERIFIED / NEUTRAL

  • “Project Y / ACS” effort at IBM as 6600-killer
  • Specific Cray foreword wording in Thornton (1970) book — likely accurate per post’s quote, but we don’t have direct access
  • Tomwsulcer/The wub/Tukulti65 image attribution for the figures
  • Specific 6600 cordwood module count (~6,000) and hand-wired interconnect length (~100 miles)
  • Dean Roush retirement / death dates (post says retired late 1970s, died early 2000s — plausible)
  • Watson 360/91 quote “an honest attempt that ran out of momentum” — could not directly verify