V-11 (1986)

The V-11 (code name Scorpio) was DEC's first VAX microprocessor design, although the second to ship. Bill Johnson was the project manager; Ed Burdick was lead engineer on the IE chip, Bill Grundmann on the M chip; Dick Sites led the architecture and microcode teams.

The Scorpio program aimed at launching DEC into the VLSI design era. In addition V-11 itself, Scorpio encompassed a new CAD suite (CHAS) and DEC's first in-house process development effort (and the industry's first double-metal process), ZMOS. V-11 implemented a full VAX, with a robust memory and console system, suitable for mid-range systems as well as embedded systems.

V-11 consisted of four chips, one of which could be replicated: the ROM/RAM chip, the IE (instruction execution) chip, the M (memory management) chip, and the F (floating point chip). Five copies of the ROM/RAM were required to implement the full VAV instruction set. V-11 initially ran at 5Mhz, with a high-speed bin point at 6.25Mhz.

Name Number Size Transistors Comments
ROM/RAM DC327 344x285 208,000 sites The ROM/RAM implements a slice of the V-11's patchable control store. Each ROM/RAM provides 16k x 8b of control store ROM, 1k x 8b of patch store RAM, and 32 patching CAMs. Five ROM/RAMs comprise the complete control store. Key features include:
  • 16K x 8b ROM
  • 1K x 8b RAM
  • 32 x 14b CAM (cascadable; 160 CAMs in complete control store)
  • 100ns access time

Power: 1W.

IE DC328 354x358 60,000 The IE chip implements the instruction execution kernel of the V-11 chip set. It operates under control of microwords fetched from the ROM/RAM chips. Its key features are:
  • Instruction prefetch
  • Instruction decode for all 304 VAX instructions
  • Optimized VAX specifier parsing
  • VAX architectural registers (32b)
  • Microcode temporary registers (32b)
  • Full function 32b arithmetic/logic unit
  • Shifter
  • First level (mini) translation buffers for I- and D-stream
  • External transaction sequencer
  • Microsequencer for control store
  • Microtrap generation and control

Power: 5W.

M DC329 339x332 54,000 The M chip provides the cache tag store and control, the backup translation buffer and control, the address translation logic, auxiliary storage for microcode temporaries, architected VAX I/O (interrupts, clock, and console), and the master clock generator for the entire chip set. The chip's features include:
  • Cache tag store and control for 8KB direct map cache
  • Backup translation buffer tag store and control for 512 entry backup translation buffer
  • Address translation logic
  • 24 auxiliary scratch registers
  • Architected I/O
    • Interrupt controller
    • Four serial ports
    • Interval timer
    • Time of year clock
  • Chip set clock generator

Power: 3W.

F DC330 341x288 29,600 The F chip is a high performance, single chip, floating point unit for the V-11 chip set. Its key features are:
  • Acceleration for floating point instructions (5X) and integer multiply/divide (2X)
  • Support for f_floating, d_floating, and g_floating data types
  • Support for most VAX floating point instructions, including POLYf
  • Arithmetic checking and reporting

Power: 2.5W

V-11 shipped in early 1986 in the VAX 8200 and 8300 mid-range systems. By that time, MicroVAX II had been out for more than six months and had captured the attention of the company and the customer base. A mid-life kicker, the VAX 8250 and 8350, based on binned parts, shipped in 1987.

Personal Narrative

V-11 attempted to do "six impossible things before breakfast" -- new methodology, new CAD tools, new architecture, new process, new design team, new organization -- and it paid the price for being a pioneer: a long development schedule filled with mishaps. MicroVAX, by starting later and dropping much of the complexity of the VAX architecture, reaped the benefits of V-11's (bitter) experiences, pillaged its microarchitecture and design objects, and shipped first. Yet V-11 fulfilled the intentions of its creators in revolutionizing DEC's approach to chip design. Before V-11, DEC used partners for process design, did its chip design work on paper, and had a small design group capable of one design at a time. After V-11, DEC did its own process development, did its chip design work entirely via CAD tools, and had a design group capable of building multiple chips in parallel.

Once MicroVAX was on the drawing boards, the V-11 management team made a determined effort to simplify the design and get it out the door. V-11's unique F-chip was dropped in favor of a derivative of the J-11 FPA. Features were stripped from the IE chip to get the size and power down. Even though MicroVAX could be considered an internal competitor, the entire V-11 team was extraordinarily gracious in providing support to the MicroVAX team.

I became the manager of microprocessor development in October, 1984, and thus supervised the last 15 months of V-11. My only real contribution was to utilize spare space in the control store patch RAM to recode CALLS and CALLG for higher performance, based on studies done for CVAX.

V11 was presented at the 1984 International Solid State Circuits Conference.


Updated 30-Jan-2007 by Bob Supnik (simh AT trailing-edge DOT com - anti-spam encoded)