RTsim: A register transfer simulator

Lam, 1983

Category: VLSI CAD

Overall Rating

1.3/5 (9/35 pts)

While the specific technical implementation of RTsim is obsolete, its structured architecture for formally defining and passing signal states (beyond simple logic levels) between high-level functional blocks and lower-level physics-aware simulation kernels represents a less explored conceptual approach to mixed-level simulation.
However, this structural idea, while potentially inspiring for highly specialized simulation frameworks..., is overshadowed by the paper's outdated models and custom, impractical implementation.

While register transfer simulation and mixed-level simulation are not new concepts, RTsim's specific implementation details, such as the explicit handling of seven signal states (including driven vs. charged distinction and charge-hold modeling)...present a less explored paradigm compared to modern HDL-based flows.
Repurposing this structured approach to handle complex, non-binary, or analog-influenced signals in specific parts of a mixed-domain system, alongside abstract functional models, holds significant latent potential.
Modern computational power (GPUs, cloud computing) makes large-scale simulation far more feasible than in 1983, addressing one of the inherent limitations of detailed simulation mentioned in the paper.
RTsim's mixed-level approach and explicit handling of capacitance/charge-hold and distinct signal states (driven/charged) align remarkably well with the type of modeling needed to simulate these systems efficiently – using high-level abstraction for digital control while dropping down to a physics-aware...level for critical or novel components, accurately propagating non-standard signal information across the boundary.

The simulator's explicit focus on "MOS circuitry" and its simplified "MOS capacitance" model ("charge-hold-period") are insufficient for modern CMOS processes with vastly more complex transistor behaviors, leakage currents, and intricate parasitic effects.
This paper likely faded into obscurity because it was quickly superseded by industry-standard hardware description languages (HDLs) and their associated commercial simulators.
RTsim's reliance on a custom "register transfer description (RTD) language," an "embedded functional modeling language" within MAINSAIL, and a niche implementation language (MAINSAIL) created a closed ecosystem that lacked the interoperability and broad tool support of the emerging standards.
The dependence on an interface to MOSSIM II, a switch-level simulator from the same era, ties RTsim to another likely obsolete tool, creating a dependency on outdated technology at multiple levels of abstraction.

Ignore