This is Beagle-Ears -> Lars -> Professional -> Computer History -> VAX


During the 1970's and 1980's, Digital Equipment Corporation designed and manufactured the immensely successful PDP-11 family of minicomputers. Around 1978, it was clear that as prices were falling and people could afford more powerful computers, larger models would be needed to serve the same markets, but the PDP-11 with its 16-bit address registers could not be easily extended, and a 32-bit architecture was designed. The goals for this project were clear:

The design team had close co-operation between hardware engineers led by the legendary Gordon Bell and a software team led by Dave Cutler. The resulting system was the VAX-11/780 computer and the VMS operating system, and was available in 1978.

Around 1982, the computer science research group at University of California at Berkeley was contracted by the US Department of Defense Advanced Research Projects Agency (ARPA) to port the Unix operating system to the VAX-11 architecture and extend it with networking protocol software so that it could be used on the Internet in a similar manner as the TENEX operating system that Bolt Beranek and Newman had written for the PDP-10.

The VAX was very successful with both operating systems.

Hardware Evolution

(Corrections to the list below are actively solicited)

See also:

VAX-11/78019781.0 The first VAX. SBI backplane, with Massbus and UNIBUS adapters. An LSI-11 with dual floppy drives was built in for use as a diagnostic control unit, and also used for microcode loading. Most important novelty was the Massbus, a mainframe-like data channel for disk and tape drives.
VAX-11/75019820.6 A less expensive machine for those who wanted a VAX but could not afford a 780. Most remarkable peripheral: TU58 serial cassette tape drive used as diagnostic bootstrap medium.
VAX-11/78219821.8 Two 11/780s sharing an MA-780 memory bank.
VAX-11/78419833.5 Four 780s sharing a memory bank. Very rare.
VAX-11/78519821.5 A 780 whose performance was boosted with a faster cache memory. A dual-processor version was called the 11/787. T878 Massbus tape drives
VAX-11/730 (11/725)19840.3 A low-end, slow and fairly cheap machine with only a UNIBUS as the system bus. This machine was so slow that its main market was as a software development system for small software houses that needed a "toy" machine for their software engineering lab. But it was different enough from the larger models that you still needed test time on a "real" VAX. Amazingly, the russians cloned this machine as well, TU58 console tape and all!
MicroVAX I198?0.5 First single-chip VAX CPU. Qbus with 22-bit addressing as main bus. Did not run teribly well.
MicroVAX II198?0.9 A wonderful machine. Cheaper, faster and more robust than the 730.
MicroVAX 2000????0.9 Workgroup server
MicroVAX 3000198?1.0 SynopsisNew gear
VAXstation198?1.0 SynopsisNew gear
VAX-86001984? Upgrade path from 780. Still uses SBI as the main system bus, and thus shares all the 780 peripherals. TD>
VAX-8650198?5.0 Small mainframe. Last production VAX with PDP-11 compatibility mode.
VAX-8200?0.9 Smallest machines with VAXBI bus. 8250 was dual-processor version. VAXBI (BI bus). RX50 floppy (5.25 inch) as console load media.
VAX-83001.4 Slightly faster version of 8200
VAX-85003.0 Much faster version of 8200
VAX-8800?? Upgrade path from 780. VAXBI (BI bus)
VAX-90001990??? May have been intended as an attempt to build a supercomputer, but by the time it came out, it did not make the grade and was presented as a mainframe instead. This was right after the Jupiter project (large PDP-10) was cancelled, and the 36-bit hardware team was taked with building a VAX instead. The process is said to have been painful, and the resulting machine had reliability problems and lots of field rework (one ECo replaced most of the boards with new revisions!). Only a few dozen were built.
MicroVAX 3100199416 or 38 Desktop workstation
VAX 4000199438 Desktop or Rackmount Server. Qbus.

Cluster Hardware

The VAX family pioneered the "Storage Area Network" with the CI (Cluster Interconnect) bus. At the center of the cluster was the Star Coupler, a mostly passive termination device. The cluster consisted of CPUs and HSC50 storage controllers. The CPUs had Cluster Interface Adapters that plugged into the SBI of the 11/780 or the BI bus of the later large models, and coax cables went from the CIAs and HSCs to the Star Coupler.

The HSC50 had a 13 MB/sec internal bus, and had several disk channels each capable of 3.125MB/sec to tape and disk drives (RA81). The data rate from CIA to HSC could reach 70 Mbps.

  $Log: vax.htm,v $
  Revision 1.3  2001/10/26 13:28:03  lars
  Replaced CMC -> Beagle-Ears

  Revision 1.2  2001/01/02 06:44:25  lars
  *** empty log message ***

  Revision 1.1  2000/08/15 01:25:08  lars
  Rearranged files, moving these from computer/ to comphist/