PDP-10

This interoffice memorandum, dated April 7, 1983, from Mike Uhler, evaluates the development path for the next PDP-10 CPU. The memo compares two primary options: a conversion of the existing KL10 design versus a variant of the new 'Jupiter' architectural design.

Key arguments presented include: * Jupiter Design Advantages: The Jupiter architecture incorporates eight years of learning from the KL10 and KS10 designs, offering full 30-bit virtual address space support, correct implementation of extended addressing, improved hardware monitor interfaces, enhanced timesharing efficiency, and native corporate bus interfaces. * KL10 Conversion Challenges: The existing KL10 design suffers from bugs (primarily in extended addressing), is limited by bus width to 4 Mwords of physical memory, and lacks optimal I/O interfaces. Converting it is described as a high-risk, non-trivial effort that essentially requires a complete redesign. * Performance and Time-to-Market Tradeoff: The memo concludes that a Jupiter-based machine with a nominal performance of 4x that of a KL10 provides the best balance between performance goals and time-to-market. It suggests that this target does not significantly impact the schedule compared to a KL10 conversion.