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.

Post 49 — “The Attack of the Killer Micros” — Biographical Research

Remit: the PEOPLE behind the vector→MPP transition (~1988–2005), MPI standardisation, and the NWP model migration. Birth/death dates & places, career arc, key papers/talks with years, usable direct quotes with sources. Living people handled cautiously; anything unconfirmed marked FLAG / NOT VERIFIED.

Status: COMPLETE (first pass). ~12 web fetches/searches.

Cross-link slug verification (checked against /_posts/)

  • Post 42 (Cyber 205, “The First in Bracknell”): /weather/hpc/history/2026/05/15/The-first-in-Bracknell.html — file 2026-05-15-The-first-in-Bracknell.md CONFIRMED
  • Post 39 (spectral transform, “Thirty-eight Years on the Sphere”): /weather/hpc/history/2026/05/14/Thirty-eight-years-on-the-sphere.html — file 2026-05-14-Thirty-eight-years-on-the-sphere.md CONFIRMED
  • Post 47 (4D-Var / Fujitsu VPP700, “Eleven Years from the Adjoint”): /weather/hpc/history/2026/05/30/Eleven-years-from-the-adjoint.html — file 2026-05-30-Eleven-years-from-the-adjoint.md CONFIRMED
  • Post 48 (satellite radiance / “enabling machine was software”, “The Hemisphere That Caught Up”): /weather/hpc/history/2026/07/20/The-hemisphere-that-caught-up.html — file 2026-07-20-The-hemisphere-that-caught-up.md CONFIRMED
  • Cray lineage (optional, verify slug before linking): Post 26 Cray-1 is almost certainly 2026-05-04-Serial-number-one.md; the CDC 7600 freon-cooling post is 2026-05-07-Freon-and-wire.md. Seymour Cray already appears in Posts 26/30/32/42 per the task brief — cross-link him rather than re-introduce.

FRAME — the “killer micros” and the vector-industry collapse

Eugene Brooks (LLNL) — coiner of “the attack of the killer micros”

  • Physicist / computer scientist at Lawrence Livermore National Laboratory.
  • Coined the phrase in a talk titled “Attack of the Killer Micros” delivered at Supercomputing 1990 (SC90). NOTE ambiguity: several secondary sources also cite SC’89; the term “first entered the Jargon File in 1990.” Recommend phrasing as “around 1989–1990, crystallised at Supercomputing 1990.”
  • Thesis: plot supercomputer peak FLOPS vs. commodity microprocessor peak FLOPS over time; the curves cross ~1990 with micros pulling ahead. Ganged by the hundreds/thousands (MPP), commodity RISC micros riding Moore’s law would overrun the expensive custom bipolar vector CPU on price/performance.
  • Usable direct quote (Jargon File / New Hacker’s Dictionary, widely cited): “No one will survive the attack of the killer micros!” — described as “the battle cry of the downsizers.” Also rendered “Nobody will survive the attack of the killer micros.”
  • Jargon File 2002 update vindicated him: since 1990 the minicomputer vanished, the mainframe went into terminal decline, and the supercomputer business contracted into a niche — all displaced by micro-based machines.
  • Title likely a nod to the cult film Attack of the Killer Tomatoes (1978).
  • FLAG / NOT VERIFIED: full name commonly given as “Eugene D. Brooks III” — plausible and widely repeated, but I did not confirm the “III” against a primary LLNL source in this pass. Birth date NOT found. Living person — treat carefully.

Seymour Cray — the vector paradigm’s founder, killed as MPP won (cross-link Posts 26/30/32/42)

  • Born 28 September 1925, Chippewa Falls, Wisconsin. CONFIRMED.
  • Died 5 October 1996 (one week after his 71st birthday), of complications from massive head injuries. CONFIRMED.
  • Accident: 22 or 23 September 1996 (UPI/Post-Bulletin say Sept 22; Wikipedia says Sept 23 — FLAG the exact day, give “late September 1996”). His Jeep Cherokee was struck while he was merging onto Interstate 25 near the U.S. Air Force Academy, north of Colorado Springs, Colorado; the car rolled over. Police cited the other driver for careless driving (the driver attempting to overtake). CONFIRMED across Washington Post, UPI, Post-Bulletin, Computer History Museum.
  • Cray Computer Corporation (his post-Cray-Research venture, moved to Colorado Springs): the Cray-3 (500 MHz, gallium-arsenide) — only one ever delivered, to NCAR on 24 May 1993, with no paying customer. By then MPP machines had arrived “at price/performance ratios the Cray-3 could not touch.” The Cray-4 (~$360 million cited development cost) never found a buyer; unable to raise the ~$20 million to finish it, the company filed for Chapter 11 bankruptcy on 24 March 1995. CONFIRMED.
  • Founded SRC Computers (Colorado Springs) in 1996, intending — pointedly — a massively parallel machine emphasising communication/memory bandwidth; design had barely begun when he was killed. The irony for the post: even Seymour Cray’s own last vector machine was killed by the killer micros, and his final company conceded the MPP argument.
  • Vector-industry collapse timeline for the FRAME (from brief; treat as scaffold, verify each in write-up): Thinking Machines Corporation bankruptcy 1994; Cray Computer Corp bankruptcy March 1995; Cray Research absorbed by SGI 1996; Seymour Cray dies Oct 1996. TOP500 first published June 1993.

ENGINE — MPI, MPICH, Gustafson’s law (the “machine is software” figures)

John L. Gustafson (Sandia) — Gustafson’s law / weak scaling

  • At Sandia National Laboratories in the late 1980s, working on massively parallel processing against Amdahl’s-law skepticism.
  • Landmark note “Reevaluating Amdahl’s Law”: written October 1987, published Communications of the ACM, vol. 31, no. 5, May 1988, pp. 532–533 (some summaries say “March 1988” — the CACM issue is May 1988; FLAG the month if precision needed; primary at johngustafson.net confirms). CONFIRMED.
  • Empirical basis: timing results on a 1024-processor nCUBE hypercube (Benner, Gustafson & Montry, “Development and analysis of scientific application programs on a 1024-processor hypercube,” Sandia report SAND 88-0317, Feb 1988). CONFIRMED.
  • Idea = weak scaling: grow the problem with the machine rather than accept Amdahl’s fixed-problem pessimism. Edwin Barsis urged publication and reportedly first called it “Gustafson’s Law”; hence sometimes “Gustafson–Barsis law.”
  • FLAG: birth date/place NOT found. Living person — careful. (Later known for the “unum” number format, but that is out of scope.)

Jack (Jack Joseph) Dongarra — MPI Forum principal, LINPACK/TOP500

  • Born 18 July 1950, Chicago, to a family of Sicilian immigrants. CONFIRMED.
  • University Distinguished Professor Emeritus, University of Tennessee, Knoxville; Distinguished Research Staff, Oak Ridge National Laboratory; Turing Fellow, Manchester. FRS.
  • Numerical linear algebra & parallel computing: EISPACK, LINPACK, BLAS, LAPACK, ScaLAPACK, Netlib, PVM, MPI, TOP500, ATLAS, PAPI.
  • A.M. Turing Award (2021) for pioneering numerical algorithms/libraries for HPC. CONFIRMED.
  • Central to MPI: co-author of the initial 1993 draft proposal (below). Living.

William D. (“Bill”) Gropp — MPICH co-author

  • Argonne National Laboratory, Mathematics and Computer Science (MCS) Division; co-creator (with Lusk) of MPICH, the reference/portable implementation of MPI.
  • Later Director of the National Center for Supercomputing Applications (NCSA), University of Illinois. Co-author of the standard textbook Using MPI (with Lusk and Anthony Skjellum).
  • FLAG: birth date NOT found. Living person.

Ewing (“Rusty”) Lusk — MPICH co-author

  • Kansas native. B.A. mathematics, University of Notre Dame, 1965; Ph.D. mathematics, University of Maryland, 1970. Began as assistant professor of mathematics at Northern Illinois University. CONFIRMED.
  • Joined Argonne in 1982; became Director of the MCS Division. Co-developer of MPICH / MPICH2 (R&D 100 award 2005). CONFIRMED.
  • Living / retired. Homepage: mcs.anl.gov/~lusk.

Marc Snir — MPI Forum principal, IBM SP architecture

  • Ph.D. in Mathematics, Hebrew University of Jerusalem, 1979. CONFIRMED.
  • Senior manager at IBM T. J. Watson Research Center until 2001, leading the Scalable Parallel Systems group behind the IBM SP and later Blue Gene. Then University of Illinois Urbana-Champaign (Michael Faiman Professor; later interim/led Argonne MCS). CONFIRMED.
  • One of the principal developers of MPI. IEEE Seymour Cray Award, 2013 (nice thread back to Cray). Living.
  • FLAG: exact birth date NOT found.

David W. Walker — MPI Forum convener / draft author

  • At Oak Ridge National Laboratory in the early 1990s; Convener and Meeting Chair roles for the MPI Forum.
  • Co-author of the seminal initial draft: Dongarra, Hempel, Hey & Walker, “A Proposal for a User-Level, Message-Passing Interface in a Distributed Memory Environment,” ORNL Technical Report (ORNL/TM-12231), February 1993 — the document that seeded the Forum. Also co-author of Walker & Dongarra, “MPI: A Standard Message Passing Interface,” Supercomputer, 1996; and of MPI: The Complete Reference.
  • Later Professor at Cardiff University (Wales). FLAG: birth date NOT found. Living.

Rolf Hempel — European message-passing (PARMACS), MPI draft author

  • German parallel-computing researcher; associated with GMD (Gesellschaft für Mathematik und Datenverarbeitung / German National Research Center for Information Technology) and later head of the DLR (German Aerospace Center) Simulation and Software Technology lab.
  • Key figure behind PARMACS, a pre-MPI portable message-passing macro library; co-author of the 1993 ORNL MPI draft proposal (with Dongarra, Hey, Walker), bringing the European PARMACS experience into MPI. Also worked on PARMACS→MPI migration and MPI on the NEC SX-4.
  • FLAG: birth date NOT found. Living.

Tony (Anthony J. G.) Hey — MPI draft author, from particle physics to parallel computing

  • DPhil in particle physics, University of Oxford; research at Caltech and CERN.
  • Professor of computation, Department of Electronics and Computer Science, University of Southampton, from 1986; Head of School 1994, Dean of Engineering & Applied Science 1999. CONFIRMED.
  • A pioneer of distributed-memory message-passing in the 1980s; co-wrote the first draft of the MPI standard with Dongarra, Hempel and Walker (the 1993 ORNL proposal). Later Director of the UK e-Science programme; then Microsoft Research (VP); then STFC Chief Data Scientist. Living. (Honours: CBE — verify before stating.)

Steve (Steve W.) Otto — Zipcode, MPI Forum

  • Developer of Zipcode, a portable communication library built atop the Caltech Reactive Kernel — one of the pre-MPI dialects the Forum unified.
  • Associated with Caltech (Caltech Concurrent Computation Program lineage) and later the Oregon Graduate Institute (OGI, near Portland).
  • Co-author of MPI: The Complete Reference (with Snir, Huss-Lederman, Walker, Dongarra) and of early MPI Forum drafts (“A Message Passing Standard for MPP and Workstations,” CACM 1993, Dongarra/Otto/Snir/Walker). FLAG: birth date NOT found.

MPI standardisation timeline (from brief; scaffold — confirm in write-up)

  • Pre-MPI Babel of dialects: PVM, Intel NX, Thinking Machines CMMD, Express, p4, PARMACS, Zipcode.
  • MPI Forum convened from 1992 (Williamsburg workshop → BOF at Supercomputing 1992) → MPI-1.0 in May 1994 → MPI-2 in 1997.
  • MPICH (Gropp & Lusk, Argonne) = reference implementation. See Gropp & Lusk, “Sowing MPICH: A Case Study in the Dissemination of a Portable Environment for Parallel Scientific Computing,” Int. J. Supercomputer Applications, 1997.
  • Programming model = domain decomposition + halo exchange; Gustafson’s weak scaling is the theoretical justification.

SCAFFOLD — NWP migration people at ECMWF

George Mozdzynski — IFS parallelisation / spectral-transform scalability

  • ECMWF computing/model-infrastructure scientist; long associated with parallelising the Integrated Forecasting System (IFS) and specifically the hard problem of the spectral transform (global Fourier + Legendre transforms needing all-to-all “transpose” communication).
  • Work on the transposition strategy, the IFS transform library, the reduced Gaussian grid, and a fast Legendre/spherical-harmonics transform enabling very high resolution (e.g., the first T7999 global forecast, ~2.5 km). Relevant paper: Wedi, Hamrud & Mozdzynski, “A Fast Spherical Harmonics Transform for Global NWP and Climate Models,” Monthly Weather Review, 2013. Also authored/co-authored early “IFS: a parallel production weather code” material.
  • Good bridge to Post 39 (spectral method) and Post 47 (VPP700 / IFS). FLAG: birth date NOT found. Living.

Geerd-Rüdiger (“Geerd-R.”) Hoffmann — head of computing at ECMWF, later DWD

  • Head of the computing/computer division at ECMWF during the vector→parallel transition; organiser/editor of the ECMWF “Workshop on the Use of Parallel Processors in Meteorology” series and editor of The Dawn of Massively Parallel Processing in Meteorology (Springer) — a directly citable primary artifact of exactly this post’s theme.
  • Later at / director of Deutscher Wetterdienst (DWD), Offenbach, Germany. Publications: “Weather Forecasting and Parallel Processing: a View from ECMWF”; “High-Performance Computing and Networking for Numerical Weather Prediction.”
  • FLAG: I did not find a primary page stating the exact ECMWF title/dates (“head of computing”) or his birth date — the “head of computing at ECMWF” attribution comes from the task brief and is consistent with the workshop-organiser role, but confirm against an ECMWF source before asserting. Living.

Suggested primary sources / citable papers (for the write phase)

  • Gustafson, “Reevaluating Amdahl’s Law,” CACM 31(5), May 1988 — johngustafson.net/pubs/pub13/pub13.htm
  • Dongarra, Hempel, Hey, Walker, “A Proposal for a User-Level, Message-Passing Interface in a Distributed Memory Environment,” ORNL/TM-12231, Feb 1993 (the seed document).
  • Dongarra, Otto, Snir, Walker, “A Message-Passing Standard for MPP and Workstations,” CACM, 1993 (ResearchGate copy exists).
  • MPI Forum, “MPI: A Message-Passing Interface Standard” — MPI-1.0, May 1994 (mpi-forum.org).
  • Gropp & Lusk, “Sowing MPICH…,” Int. J. Supercomputer Applications, 1997.
  • Snir, Otto, Huss-Lederman, Walker, Dongarra, MPI: The Complete Reference (MIT Press).
  • Wedi, Hamrud, Mozdzynski, “A Fast Spherical Harmonics Transform…,” MWR, 2013.
  • Hoffmann (ed.), The Dawn of Massively Parallel Processing in Meteorology, Springer (ECMWF workshop proceedings).

KEY UNCERTAINTIES / FLAGS (re-verify before publishing)

  1. Eugene Brooks — the “III” in “Eugene D. Brooks III” not confirmed against a primary LLNL source; talk year SC90 vs SC89 ambiguous (use “1989–1990, crystallised at SC90”). No birth date.
  2. Seymour Cray accident day: 22 vs 23 September 1996 (sources disagree); death 5 Oct 1996 is firm.
  3. Gustafson “Reevaluating Amdahl’s Law” issue month — the CACM issue is May 1988 (one secondary source says March); verify if stating a month. No birth date.
  4. Birth dates NOT found for: Brooks, Gustafson, Gropp, Snir, Walker, Hempel, Otto, Mozdzynski, Hoffmann. Do not invent.
  5. Geerd-R. Hoffmann’s exact ECMWF title/tenure not confirmed against an ECMWF primary page.
  6. Vector-industry collapse dates (TMC 1994, Cray Research→SGI 1996, TOP500 June 1993) taken from the brief; confirm each in the write phase.
  7. Tony Hey honours (CBE/knighthood) not verified here.

Sources (URLs actually fetched or returned in search this session)

  • https://en.wikipedia.org/wiki/Killer_micro (fetched)
  • https://en.wikipedia.org/wiki/Seymour_Cray (fetched)
  • https://en.wikipedia.org/wiki/Jack_Dongarra (search snippet)
  • https://amturing.acm.org/award_winners/dongarra_3406337.cfm (search result)
  • http://www.johngustafson.net/pubs/pub13/pub13.htm (search result — “Reevaluating Amdahl’s Law”)
  • https://dl.acm.org/doi/10.1145/42411.42415 (CACM “Reevaluating Amdahl’s law”, search result)
  • https://www.washingtonpost.com/archive/local/1996/10/06/computer-pioneer-seymour-cray-dies/ (search result)
  • https://www.postbulletin.com/computer-pioneer-cray-dies-from-traffic-accident-injuries (search result)
  • https://www.computerhistory.org/tdih/october/5/ (search result — Cray death)
  • https://www.mcs.anl.gov/~lusk/ (search result — Rusty Lusk homepage)
  • https://www.newswise.com/articles/lusk-named-director-of-mathematics-and-computer-science-division-at-argonne (search result)
  • https://journals.sagepub.com/doi/10.1177/109434209701100204 (Gropp & Lusk, “Sowing MPICH”, search result)
  • https://siebelschool.illinois.edu/about/people/all-faculty/snir (search result — Marc Snir)
  • http://sc13.supercomputing.org/content/parallel-computing-pioneer-marc-snir-receive-2013-ieee-seymour-cray-award-sc13.html (search result)
  • https://impact.ornl.gov/en/publications/mpi-a-standard-message-passing-interface/ (search result — Walker/Dongarra; 1993 ORNL proposal)
  • https://en.wikipedia.org/wiki/Tony_Hey (search result)
  • https://computerhistory.org/events/bio/Tony,Hey (search result)
  • https://dblp.org/pid/76/235.html (Rolf Hempel publications, search result)
  • https://www.anl.gov/article/pioneers-of-highperformance-computing-library-reunite (search result — PARMACS pioneers)
  • https://www.researchgate.net/publication/220421514_A_Message_Passing_Standard_for_MPP_and_Workstations (Otto/Snir/Dongarra/Walker, search result)
  • https://en.wikipedia.org/wiki/Oregon_Graduate_Institute (search result)
  • https://journals.ametsoc.org/view/journals/mwre/141/10/mwr-d-13-00016.1.xml (Wedi/Hamrud/Mozdzynski fast SH transform, search result)
  • https://www.sciencedirect.com/science/article/abs/pii/0167819196800020 (“The IFS model: A parallel production weather code”, search result)
  • https://dblp.org/pid/67/1818.html (Geerd-R. Hoffmann publications, search result)
  • https://www.goodreads.com/author/show/2822679.Geerd_R_Hoffmann (ed. “The Dawn of Massively Parallel Processing in Meteorology”, search result)
  • Jargon File “killer micro” entry (catb.org/~esr/jargon/html/K/killer-micro.html — TLS/cert error on direct fetch; content confirmed via search snippets and outpost9.com / foldoc.org mirrors)