Dolphin Diagnostic Plan Feb78

This document outlines the "Diagnostic Engineering Project Plan for DOLPHIN," dated February 1979, detailing the strategy for developing comprehensive diagnostic capabilities for the new Dolphin computer system.

Key Points:

• Dolphin System: An advanced CPU system utilizing Macro Cell Array (MCA) technology, intended as a successor to PDP-10/20 and a potential high-end VAX. The diagnostic plan is crucial for achieving high system reliability, availability, and maintainability (RAMP performance) to penetrate new commercial markets.

• Diagnostic Product: A comprehensive set of tests and support software designed to diagnose faults across the entire Dolphin system. It includes:

  • Console Subsystem Software: Runs on a PDP-11 based console processor for system startup, configuration, error handling, and operator interface.
  • CPU Cluster Diagnostics: Detects and isolates faults in the CPU, console processor, and bus interfaces, utilizing IIL logic and a "known good" machine state database.
  • Peripheral Subsystem Diagnostics: For bus adapters, controllers, and devices, operating in "reliability" (thorough) and "device" (high-resolution) modes.

• Goals:

  • Functionality: Support MCA design verification, provide console processor bootstraps and comprehensive console software, and offer both console-based and Dolphin-based diagnostics for CPU, memory, and various peripherals (e.g., RP06/07, TM03/77/78, HSC50, Mercury communications). Also includes a system exerciser and quick-verify support for manufacturing.
  • Performance: Detect over 98% of solid hardware faults in core logic and the highest percentage permitted by vendor hardware for peripherals. Resolve faults to the Field Replaceable Unit (FRU) over 90% of the time, and to a single MCA for over 95% of CPU faults. Minimize diagnosis time.

• Non-Goals: The project does not include funding for 32-bit Dolphin support, additional console peripherals (beyond video, TU-58, hard copy, APT/KLINIK ports), MCA incoming inspection tests (beyond design verification patterns), CPU board tests (beyond quick-verify), or support for expanded systems with bus repeaters lacking a complete console subsystem.

• Audiences: Large System Hardware Engineering (design verification), Large System Manufacturing (system bring-up, quick verify, volume production), and Field Service (on-site and remote maintenance).
• Key Dependencies & Risks: The project is critically dependent on the timely availability and design of Dolphin hardware features (IIL diagnostic logic, F-11 console, MAD-30 memory), the Monitor group's implementation of the "DIAG JSYS" functionality, and the availability of the VOTE fault simulation system for quality assurance. Delays in these areas pose significant risks to the project schedule.
• Implementation & Resources: The plan covers the period from February 1979 to mid-1981, aiming for a preliminary diagnostic package by June 1981 to support the first customer ship. It outlines specific tasks, manpower loading (totaling 104.5 person-quarters), and funding (estimated at $1.254 million).

In essence, the document details a critical project to develop a robust diagnostic suite for the new Dolphin computer, aiming for high fault detection and resolution, with a clear scope, timeline, and acknowledgement of significant interdependencies and risks.