This document outlines the discussions held on March 3, 1983, concerning the replanning of the Jupiter II CPU. Key topics included:
- Multiprocessing Configurations: Challenges in implementing SMP, including the need for multi-port memory, cache consistency, and inter-processor communication.
- Array Processors: Difficulties in connecting external array processors due to the tightly coupled MBOX and memory architecture.
- COBOL Performance: Concerns regarding the performance of existing EXTEND instructions and potential microcode or hardware optimizations.
- Performance Measurement: Suggestions for better performance instrumentation, such as dedicated backplane signals, opcode histograms, and PC trace buffers.
- Translation Buffer (TB) Organization: Proposals for improved TB configurations and the addition of user process context numbers to avoid unnecessary flushes.
- I/O Strategy: Concerns regarding I/O bandwidth and reliance on corporate front-ends, with proposals for a UNIBUS adapter or a dedicated I/O processor.