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 26 - The Machine That Looked Like Furniture

Post: /_posts/2026-04-27-The-machine-that-looked-like-furniture.md Subject: Cray-1, Seymour Cray, NCAR serial #3, ECMWF, Warren Washington Verification date: 2026-05-17

Status legend

  • VERIFIED — confirmed by primary or strong secondary source
  • CORRECTION NEEDED — post contradicts source evidence
  • AMBIGUOUS — sources mildly conflict or post phrasing is misleading but defensible
  • UNVERIFIED — could not be confirmed (no easy primary source) but not contradicted

CRAY-1 HARDWARE FACTS

VERIFIED (primary source: Russell 1978 CACM PDF)

  • 12.5 ns clock cycle [Russell 1978 Table I]
  • 8 vector (V) registers, 64-bit elements, 64 elements per register [Russell 1978 Table I]
  • 16 memory banks, 50 ns chip cycle, 4 clock period bank cycle [Russell 1978]
  • 1,662 modules in 113 module types [Russell 1978]
  • 200,000 ECL gates (Wikipedia; not in Russell directly but consistent with chip count)
  • 115 kW power consumption [Russell 1978, Wikipedia]
  • 5.5 ton weight [Wikipedia, NCAR/CISL]
  • “world’s most expensive love-seat” — Russell 1978 cites this directly via footnote [5]; coined by editor Steve Callahan at Auerbach (per HPCwire, Kent State)
  • 12 functional units, pipelined into single clock segments [Russell 1978]
  • Freon refrigerant cooling with Freon/water heat exchange [Russell 1978]
  • 1,048,576 64-bit words memory (max) [Russell 1978]

VERIFIED (secondary sources)

  • Cray-1 announced 1975 [Wikipedia]
  • 80 Cray-1s total sold (all variants combined) [Wikipedia, Cray-History.net] — NOT “~85” as in task brief; post says “eighty Cray-1s” which is correct
  • 12.5 ns = 80 MHz [Wikipedia] — consistent with Russell

AMBIGUOUS — peak performance

  • Russell 1978 (primary): “computational rates of 138 million floating-point operations per second (MFLOPS) for sustained periods and even higher rates of 250 MFLOPS in short bursts”
  • Wikipedia: “Peak performance 160 MFLOPS”
  • Task brief said “80 MFLOPS peak” — appears to be confused with clock rate (80 MHz)
  • Post says “around 160 million operations per second” — matches Wikipedia (theoretical peak with all units running) but conflicts with Russell’s measured 138 sustained / 250 burst
  • VERDICT: 160 MFLOPS peak is the standard textbook number; Russell’s 250 MFLOPS represents true peak via chaining; post’s claim is defensible

UNVERIFIED but plausible

  • 12 i/o channels (Russell says 24 i/o channels organized in 4 groups; post doesn’t explicitly state)
  • “230 kW total power including cooling” — Russell only gives 115 kW for the machine; post claims half-half split (computation/cooling), not directly verified from Russell

CRAY-1 DEPLOYMENTS

VERIFIED

  • Serial #1 to Los Alamos in March 1976 [Cray-History.net SN1 page]
  • Six-month evaluation trial at Los Alamos [Cray-History.net, multiple secondary sources]
  • NCAR serial #3 arrived July 11, 1977 [NCAR/CISL primary page]
  • NCAR price $8.86 million total ($7.9M system + $1M for 16 DD-19 disk drives) [NCAR/CISL primary]
  • NCAR delivery in “two refrigerated electronic vans” [NCAR/CISL primary]
  • Over 30 workers needed to move the machine [NCAR/CISL primary]
  • NCAR acceptance December 1977 [NCAR/CISL primary]
  • NCAR decommissioned January 27, 1989; powered off February 1, 1989 [NCAR/CISL primary]
  • ECMWF first supercomputer installed October 24, 1978 at Shinfield Park [ECMWF news, Wikipedia secondary]
  • Initial Cray-1 used at ECMWF was SN1 from October 1977 (housed at Rutherford), replaced October 1978 [Cray-History.net FAQ-3]

AMBIGUOUS — ECMWF Cray-1A serial number

  • Task brief claims serial #9 went to ECMWF Oct 24, 1978
  • Post DOES NOT claim a specific serial number for the ECMWF machine — so no error
  • Could not confirm the serial #9 claim from any source

CORRECTION NEEDED — ECMWF operational forecasting timeline

  • Post: “Operational forecasting on an experimental basis began on August 1, 1979… Full operational seven-day-a-week forecasting began on August 1, 1980”
  • ECMWF primary source (2019 anniversary): “On 1 August 1979 ECMWF produced its first operational medium-range forecast for Member States” — this is FULL operational start, not experimental
  • ECMWF history page: “ECMWF has been producing operational medium-range weather forecasts since 1 August 1979. At first, forecasts were made five days a week. From 1 August 1980, they were made seven days a week”
  • The post’s characterization of Aug 1, 1979 as “experimental” is inaccurate; it was the operational start. Aug 1, 1980 was the move from 5 days/week to 7 days/week

SEYMOUR CRAY BIOGRAPHY

VERIFIED

  • Born September 28, 1925, Chippewa Falls, Wisconsin [Wikipedia]
  • Father also named Seymour, civil engineer [Wikipedia]
  • Mother Lillian [Wikipedia]
  • BS Electrical Engineering, University of Minnesota, 1949 [Wikipedia]
  • MS Applied Mathematics, University of Minnesota, 1951 [Wikipedia]
  • Joined ERA in 1950 [Wikipedia]
  • Designed ERA 1103 in 1953 [Wikipedia UNIVAC 1103]
  • Co-founded CDC in 1957 with William Norris [Wikipedia]
  • Designed CDC 1604 (1960), CDC 6600 (1964), CDC 7600 (designed; released 1967) [Wikipedia]
  • Left CDC in 1972 after CDC 8600 cancellation [Wikipedia, Cray Research Wikipedia]
  • Norris invested $250,000 startup capital [Wikipedia]
  • Stepped down as CEO of Cray Research in 1980 [encyclopedia.com, Cray Research Wikipedia]
  • Died October 5, 1996, from injuries in September 22, 1996 rollover on I-25 near US Air Force Academy [Wikipedia]
  • Founded SRC Computers earlier in 1996 [Wikipedia]

VERIFIED CDC reliability quote

  • Post: 7600 “famously failed at least once a day in normal operation, four or five times a day at the worst sites”
  • CDC 7600 Wikipedia: “LLNL and NCAR reported that the machine would break down at least once a day, and often four or five times” ✓

VERIFIED 6600 team size

  • Post: “the 6600 team was thirty-four people”
  • IBM CEO memo quote: “only 34 people, ‘including the janitor’” [Wikipedia CDC 6600] ✓

UNVERIFIED but in widely-cited Cray lore

  • “Activity is progressing satisfactorily” quote (post’s footnote 8)
  • “We have elves here” quote
  • Burning sailboat anecdote
  • Tunnel under Lake Wissota property
  • Five-year goal note (“Build the biggest computer in the world”)
  • “Anyone can build a fast CPU. The trick is to build a fast system.”
  • These are commonly attributed but rarely sourced to a primary publication; post acknowledges they are “Rollwagenisms”

WARREN WASHINGTON BIOGRAPHY

VERIFIED

  • Born August 28, 1936, Portland, Oregon [Wikipedia, BlackPast, multiple]
  • Father a Pullman porter on Union Pacific Railroad [Wikipedia, multiple]
  • Mother a nurse [Wikipedia]
  • BS Physics, Oregon State, 1958 [Wikipedia]
  • PhD Meteorology, Penn State, 1964 [Wikipedia]
  • Second African-American to earn doctorate in meteorology [Wikipedia, Penn State, multiple]
  • Joined NCAR in 1963 [Wikipedia, NCAR News]
  • Died October 18, 2024, age 88 [Wikipedia]
  • Chaired National Science Board 2002-2006 [Wikipedia]
  • Tyler Prize 2019 [Wikipedia]
  • Advised six presidents (NCAR tribute says “advised six presidents”) — but his autobiography titled “Advising Five Presidents” (2008). The “six” figure (Carter, Reagan, Bush 41, Clinton, Bush 43, Obama) became valid after the 2008 book.

CORRECTION NEEDED — MS degree subject

  • Post: “a master’s degree in meteorology there [Oregon State] in 1960”
  • Wikipedia: MS in “general science” (1960)
  • Multiple sources confirm MS was in general science, not meteorology

CORRECTION NEEDED — NAE election year

  • Post: “He was elected to the National Academy of Engineering in 2002”
  • Wikipedia: “He was elected to the National Academy of Engineering in February 2009”
  • Citation in NAE: “for pioneering the development of coupled climate models, their use on parallel supercomputing architectures, and their interpretation.”

CORRECTION NEEDED — Medal of Science year (minor)

  • Post: “He received the National Medal of Science from President Obama in 2009”
  • Actual: Medal was for 2009 but presented in 2010 (November 17, 2010 in East Room) — post phrasing implies 2009 ceremony, which is incorrect

CORRECTION NEEDED — Charles E. Anderson description

  • Post: “Charles E. Anderson (1919-1994), a former Tuskegee Airman”
  • Actual: Anderson was a weather officer for the 332nd Fighter Group at Tuskegee in the US Army Air Forces; he was NOT a Tuskegee Airman (the term denotes the pilots and air crew). [Wikipedia Charles E. Anderson, BlackPast]
  • This is a meaningful factual error. He should be called the “first African-American meteorologist” or “first African-American to earn a meteorology PhD” and noted as serving as weather officer for the Tuskegee Airmen’s unit.

ECMWF FACTS

VERIFIED

  • Convention signed 1973, came into force 1 November 1975 [ECMWF history]
  • 18 founding member states [ECMWF, Wikipedia]
  • Operational forecasting began 1 August 1979 [ECMWF news 2019, primary source]
  • Five days/week initially, switched to 7 days/week on 1 August 1980 [ECMWF history]
  • Aksel Wiin-Nielsen first Director, January 1, 1974 - December 31, 1979 [Wikipedia, ECMWF tribute]
  • Wiin-Nielsen Danish, born 17 December 1924, died 26 April 2010 [Wikipedia]
  • Lennart Bengtsson, Head of Research 1975-1981, Director-General 1982-1990 [Wikipedia, ECMWF anniversary]

CORRECTION NEEDED — founding member states list

  • Post says: “ten European member states – Belgium, Denmark, Germany, Spain, France, Greece, Ireland, Italy, the Netherlands, the United Kingdom, plus Switzerland, Sweden, Yugoslavia, Finland, and Austria as associate members”
  • Actual: 18 founding states, all signatories (NOT distinct “associate” tier). Founders: Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, Turkey, UK. [ECMWF official, Wikipedia]
  • Yugoslavia ratified the original Convention and was a founding party (according to one source). Some sources list 18 founders.
  • The post is wrong on two counts: (1) misses founders Luxembourg, Norway, Portugal, Turkey; (2) wrongly creates an “associate” tier for Switzerland, Sweden, Yugoslavia, Finland, Austria (all of whom were full founders)

VERIFIED — initial forecasting model

  • Post: “210-kilometre horizontal resolution with fifteen vertical levels”
  • ECMWF news 2019: “Grid Resolution: 210 km”, “15 levels” ✓
  • Post: “five hours of CPU time on the Cray-1A”
  • ECMWF news 2019: “used about 5 hours of CPU time on the Centre’s Cray-1A computer to produce a 10-day forecast” ✓

MESA LABORATORY AND I.M. PEI

VERIFIED

  • I.M. Pei designed Mesa Laboratory [Wikipedia Mesa Laboratory]
  • Selected 1961 by UCAR committee [Wikipedia, Archives.ucar.edu]
  • Construction began April 1964 [Wikipedia]
  • Completed 1966, dedicated 1967 [Wikipedia]
  • Modeled on Anasazi cliff dwellings of Mesa Verde [Wikipedia]
  • Walter Orr Roberts founding director of NCAR (1960) [Wikipedia]
  • Pei born 1917, was 44 in 1961 ✓
  • Pei went on to design JFK Library, East Building of National Gallery, Louvre Pyramid ✓

CORRECTION NEEDED — Land acreage

  • Post: “to buy 500 acres of mesa-top land”
  • Wikipedia Mesa Lab: “565-acre site”
  • Some sources may differ; 500 vs 565 is a small discrepancy

CORRECTION NEEDED — $250,000 land appropriation

  • Post: “Colorado state legislature appropriated 250 000 dollars to buy 500 acres”
  • Wikipedia: does not mention specific $250,000 figure; says state purchased the site
  • This figure could not be verified; may be correct from another source

CRAY-2, CRAY-3 FACTS

VERIFIED

  • Cray-2 designed by Cray after 1980 [Wikipedia]
  • Cray-2 released 1985 [Wikipedia]
  • Four vector processors, 256 MWord max memory [Wikipedia]
  • Fluorinert immersion cooling [Wikipedia]
  • 1.9 GFLOPS peak [Wikipedia]
  • Cray-3 used gallium arsenide [Wikipedia]
  • Only one Cray-3 (S5, “Graywolf”) delivered to NCAR May 1993 [Wikipedia, NCAR/CISL]
  • Cray Computer Corporation bankruptcy March 1995 [Wikipedia]

CORRECTION NEEDED — Cray-2 sales

  • Post: “shipped about thirty units”
  • Wikipedia Cray-2: “25” units sold
  • 25 not 30

CORRECTION NEEDED — Cray-2 “world’s fastest” period

  • Post: “the world’s fastest computer from 1985 to 1987”
  • Wikipedia: Held title until “surpassed by the Cray Y-MP in 1988”
  • Should be 1985 to 1988, not 1985 to 1987

AMBIGUOUS — Cray Computer Corporation spin-off date

  • Post: “In 1989 Cray spun off his next project”
  • Wikipedia: “In November 1988, the Colorado Springs lab was spun off as Cray Computer Corporation”
  • Slight off-by-one; spin-off was November 1988

UNVERIFIED — “world’s most expensive aquarium” nickname for Cray-2

  • Post: “Critics called it ‘the world’s most expensive aquarium.’”
  • Wikipedia Cray-2 mentions nickname “Bubbles” for cooling system; aquarium quote not in Wikipedia article retrieved
  • Likely accurate (widely repeated) but not verified by retrieved sources

OTHER FACTS

VERIFIED — CDC 6600

  • Released 1964 (some sources 1965 for first delivery); fastest computer 1964-1969 [Wikipedia]
  • 34-person team [Wikipedia, IBM memo]

VERIFIED — CDC 7600

  • Designed by Cray, world’s fastest 1969-1975 [Wikipedia]
  • Reliability problems (multiple failures per day) [Wikipedia] ✓

VERIFIED — CDC STAR-100

  • Delivered from 1974 [Wikipedia]
  • Poor scalar performance, vector-optimized [Wikipedia] ✓

UNVERIFIED — ERA 1103 customer claims

  • Post: “shipped to twenty installations including the Lawrence Livermore Laboratory and the Bureau of Ships”
  • Wikipedia UNIVAC 1103 lists customers but does not confirm 20 installations, Lawrence Livermore, or Bureau of Ships explicitly
  • Cannot verify or contradict from retrieved sources

SUMMARY OF CORRECTIONS NEEDED

  1. Charles E. Anderson was NOT a “former Tuskegee Airman” — he was a weather officer for the 332nd Fighter Group. The Tuskegee Airmen term refers to the pilots and air crew, not support officers.

  2. Warren Washington’s MS degree was in “general science” not “meteorology”

  3. Warren Washington’s NAE election was 2009 not 2002

  4. Warren Washington’s Medal of Science was for 2009 but presented November 17, 2010 — post implies 2009 presentation

  5. ECMWF founding states — post lists wrong subset with wrong associate/full member distinction. There were 18 founders, all full members, including Luxembourg, Norway, Portugal, Turkey

  6. ECMWF Aug 1, 1979 was the full operational forecasting start (5 days/week), NOT “experimental” — Aug 1, 1980 was 5→7 day/week transition

  7. Cray-2 sales — 25 units not “about thirty”

  8. Cray-2 fastest period — 1985 to 1988 (succeeded by Y-MP), not 1985-1987

  9. Mesa Lab acreage — Wikipedia says 565 acres, post says 500 (minor discrepancy; possibly resolvable)

  10. Cray Computer Corp spin-off date — November 1988, post says “1989”