A Brief History of Computing at Cornell's Laboratory of Nuclear Studies

• 1967: Newman Lab
  • CDC 160A
  • Ray Helmke hired
• 1968: Wilson Lab
  • IBM 1800
    • ?KB core memory
    • card reader, punch, line printer, removable disk cartridges
    • accelerator control, daq analysis
  • Wang shared-logic calculator
    • 4function, nixie tubes, keyboards in offices
  • Mike Kolesar hired
  • Erich Knobil hired
• 1969
• 1970
• 1971
  • DEC PDP-8A for DAQ
• 1972
  • 3x DEC PDP-11/20 for DAQ
    • 4 or 8K 16bit Words (8-16KB) core memory
    • assembly language daq code
    • 9track tape drive
  • raised floor installed in W206-210
    • Clay Ball's crew carefully raised the 1800
  • DEC PDP-10 (KA model 1050)
    • ~0.25Mips
    • 64K 36bit words
    • 2x 20MB RP02 removable disk pack drives
    • papertape diagnostics & boot loader
    • DECtape
    • 4x(?) 800bpi 9track tape drives
    • 6 ft tall black cabinets
    • lotsa blinken lites and toggle switches
    • rented 1000Cps card reader (IBM 1800 compat)
    • competition was XDS & IBM
    • cross assembler for PDP-11s
    • home built programmed I/O --> DMA to PDP 11/20s
    • 4x KSR33 Teletypes
• 1973
  • Graphics displays for PDP-10 (Tektronix 610; analog phosphor memory)
• 1974
  • Mike K. leaves, going to HP
  • June 1974: Selden Ball hired
  • Upgrade to DEC PDP-10 (KI model 1070)
    • upgrade of CPU and memory only: ~0.5Mips
    • 128K(?) 36bit words core memory,
    • another 2x 20MB removable disk pack drives
    • 6 ft tall black cabinets (painted over blue)
    • lotsa blinken lites and toggle switches
• 1975
  • UofR PDP-11/40
  • Hazeltine 2000 video terminals
    • core memory; later semiconductor
    • no lower case alphabetics, but dual intensity
    • ~$2,000 each
• 1976
  • KI
    • upgraded with 128KW Ampex core memory
    • 16 interactive users plus online datataking
• 1977
  • Joining Arpanet vaguely considered but rejected
  • raised floor installed in W221

  • DEC PDP-10 (KL model 1099)
    • ~$1.2M with tradein (KI traded in, donated back, left in place) $100K/year maintenance
    • two large ECL processors, ~1.5Mips each (2bays 3' wide, 1bay 19" wide)
    • Bert and Ernie
    • PDP-11/40 console computers
    • 768K 36bit words multiport core memory, later upgraded to 1MWord
    • 2x RP06 removable disk pack drives (~170MB each)
    • PDP-11/40 terminal server (48port?)
    • 6x TU72: STC 100IPS, 6250BPI, 9Track Tape drives (150MB), shared w/KI, PDP-8 based memory channel
    • 6 ft tall blue cabinets
    • fewer blinken lites: just PDP-11 and memory boxes.
    • Competition was CDC, Honeywell & IBM
    • DEC renamed PDP-10 to DECsystem10

• 1978
• 1979
• 1980
  • Mike Roman hired

• 1981
  • 1099: +2x RP06, +256KW core memory
  • 3rd shift full time batch operator
  • 168E development (SEB & Chris Bebek went to SLAC for training)

  • 1x DEC PDP-11/34 for CLEO
  • 3x DEC PDP-11/34 for CESR
    • 64KB semiconductor memory
    • RT-11 operating system, quickly changed to RSX-11M
    • 56Kbit serial network interfaces never used (used same home built PDP-10/PDP-11 interfaces as before)
  • Xbus designed
• 1982
  • DEC VAX-11/780 (~1Mips) (Primarily for CESR)
  • +12 Z19 (modified ROMs)
  • KL-10
    • +16 serial ports
    • 7th TU72
  • broadband cables to Newman
  • KI still in use
  • New STC tape drives for CUSB

• 1983
  • Thickwire Ethernet (DECnet)
  • 370/E project started (1Mip, 2MB)
  • BITnet: Because It's Time
    • network to campus computer for 370/E support
    • 9600 baud leased lines between Universities
    • e'mail and file transfer, no interactive logins

  • 2x DEC VAX-11/750 (~0.5Mips) (KI replacement)
  • Diablo daisywheel printers
  • +36 Z19 terminals replacing Hazeltines

• 1984
  • KI retired in January
  • 6x 370/E commissioning
  • +2x VAX-11/750
  • PDP-11/34 LAT terminal server
  • MASS-11 word processing
  • ?x DEC VT220 (downloadable fonts for Mass-11 math symbols)

• 1985
  • DEC VAX 8600 (clustered w/780) [lns61]
  • +2 RA81 (456MB) (7 total)
  • 6x 370/E unreliable
  • MicroVAX II for CESR
  • Talaris laser printer

• 1986
  • DEC VAX 8600 [lns62] to replace KL1099 (DECsystem-10)
    • ~12MB main memory on each 8600
    • "bathtub curve" of failure rate in 1099
  • TU72 tape drives replace TU/TA78 tape drives (special channel interface)
  • VAX-11/780 sold for $60K
  • PHYSNET (HEPnet) via CUSB 9600 baud leased line to Brookhaven
  • 2x reliable 370/E, 2x intermittant, 1x ng
    Apparently 370/E circuitry designs had used "typical" values for chips instead of "worst case". As a result, chips often had to be hand-selected to find one that would work reliably in the CPU.
  • uVAXII + VME 68000 +XBus in development
  • use of Theory Center Production Supercomputer Facility
• 1987
  • use of CNSF
  • rigorous 9track tape cleaning/certification
  • free uVAX replacement for CLEO VAX-11/750 (intermittant problems caused by design error)
  • 8600s upgraded to 20MB
  • 2x Apple Laserwriters +LN03R upgrade to PS
  • SASD training for CLEO (big flop)
  • 4x diskless VAXstation 2000 (clustered w/VAX-11/750)

• 1988
  • 2x 533MByte System Industries disks for 8600 cluster
  • TheoryNet (Pronet-10 campus optical network; Internet)
  • Accelerator Physics cluster [cesr10]
    • VAXserver 6600
    • 5x VAXstation 3200
    • 4x VAXstation 2000
  • Loaner DEC VAX 6360
  • Major proposed upgrade to distributed systems
    • Apollo, Convex, DEC, IBM, Sun
    • Win: DEC VAX 9000-420 + VAXstation3100 + DECstation3100

• 1989
  • Major upgrade not quite as proposed
    • DEC VAX 6000-460 (4x 10Mips CPUs)
    • 40GB disk space (RA90 disks, ~1.2GB)
    • 69x VAXstation 3100 (VMS; 16x color)
    • 7x DECstation 3100 (Ultrix)
  • Anvil drafting software
  • 8mm tape drives
  • thinwire Ethernet

• 1990
  • RISC mainframe search (DEC, HP/Apollo, SGI, Solbourne)
  • 10x VAXstation 3100
  • 7x DECstation 3100
  • 33x 660MB disks
  • 2x DECstation 5000
  • evaluating 4mm tape

• 1991
  • C. Ray Ng
  • RISC mainframe
    • SGI 4D/310
    • SGI 4D/380S (8x 10Mips)
    • 20GB SCSI: multiple Seagate disks
    • 6x 8mm tape
  • upgrade DEC 6000-460 [lns62] to 6000-540 (4x 12Mips CPUs)
  • upgrade 38GB of 1.2GB RA90 to 48GB of 1.5GB RA92 disks
  • replace 8600 [lns61] by VAX 3500
  • 6x X terminals
  • 5x DECstation 5000
  • DECstation 5000 compute farm for offline data reduction (includes Florida systems)
  • 4mm tape in use
  • DECstation 5000 for Mafia
  • FDDI
  • Network Systems FDDI/Ethernet bridge/router

  • ~150 serial terminals in use

• 1992
  • sold DEC RA92 disks to buy SCSI disks
  • VAXstation 4000-60 cluster server for Accelerator computing
  • VAXstation 4000-60 cluster server for Accelerator control
  • 4x Tek Xterminals for drafting (replacing VAXstation 3100)
  • DECstation replacement for SUN for Machine Shop
  • Beta tested DEC "Cobra" DEC 4000 AXP (Alpha based)
  • 6000-540 --> 4000-500 (32Mips) + 4x 4000-60 (12Mips) (VAX based)
  • 30x used DECstations from CS dept.
  • 13x VAXstations sold (18x still on desks; 72x DECstations)

• 1993
  • FDDI couldn't work in -60, free upgrades to -90
  • December: WWW.LNS.CORNELL.EDU

• 1994
• 1995
• 1996
• 1997

• 1998
• 1999
• 2000
• 2001
• 2002
• 2003

• 2006
  • March 22: WWW.LEPP.CORNELL.EDU
• linux clusters
• At some time before 2006, LNS61 became a DigitalPersonalWorkStation 600au (based on DEC's proprietary Alpha CPU chip) connected to a half-dozen 9-70GB SCSI disks. LNS62 became a VAXstation 4000-90, running diskless, clustered with LNS61.
  
  ...
• 2013
  • January: Ray Helmke retired
  • David Kreinick as temporary director of computer group
  • Werner Sun hired as permanent director of computer group

• 2016
  • July 26:
    • central VMS systems LNS61 and LNS62 retired
    • CESR VMS systems shut down as a test, then restored; permanent shutdown in December?
  • November: Selden to retire

• Computer staff no longer here
  • Chris Hopper
  • David Jones (to video consulting)
  • Erich Knobil (to ROLM; then teaching)
  • Gary Percy (to west coast with HEP wife, then Wind River Systems)
  • Zach Pesold (to computer consulting)
  • Turhan Rahman (to robotics research)
  • Suzanne Richichi (nee Banks) (to CUAdmin)
  • Mark Rondinaro (to India and Watkins Glen)
  • Scott Smyers
  • Alex Sunguroff (to Impediment)

• Notable Undergrads
  • Karl Smolenski (then CHESS, now CESR)
  • Eric Sachse

• Misc
  • Macs
  • PCs
  • Alphas: OSF/1
  • Internet
  • twisted pair ethernet
  • NT
  • Linux
  • Sun