Detail enhances the visual richness and realism of computer-generated images. Our stochastic modelling approach, called particle systems, builds complex pictures from sets of simple, volume-filling primitives. For example, structured particle systems have been used to generate trees and a grass-covered forest floor. Particle systems can produce so much irregular, three-dimensional detail that exact shading and visible surface calculations become infeasible. We describe approximate and probabilistic algorithms for shading and the visible surface problem. Because particle systems algorithms generate richly-detailed images, it is hard to detect any deviation from an exact rendering. Recent work in stochastic modelling also enables us to model complex motions with random variation, such as a field of grass blowing in the breeze. We analyze the performance of our current algorithms to understand the costs of our stochastic modelling approach.

Smalltalk on a RISC (SOAR) is a simple, Von Neumann computer that is designed to execute the Smalltalk-80 system much faster than existing VLSI microcomputers. The Smalltalk-80 system is a highly productive programming environment but poses tough challenges for implementors: dynamic data typing, a high level instruction set, frequent and expensive procedure calls, and object-oriented storage management. SOAR compiles programs to a low level, efficient instruction set. Parallel tag checks permit high performance for the simple common cases and cause traps to software routines for the complex cases. Parallel register initialization and multiple on-chip register windows speed procedure calls. Sophisticated software techniques relieve the hardware of the burden of managing objects. We have initial evaluations of the effectiveness of the SOAR architecture by compiling and simulating benchmarks, and will prove SOAR's feasibility by fabricating a 35,000-transistor SOAR chip. These early results suggest that a Reduced Instruction Set Computer can provide high performance in an exploratory programming environment.

@; programming environment are both to decrease the average page fault rate and to minimize the pauses caused by clusters of page faults. We have applied program restructuring techniques to the Smalltalk-80 object memory in order to improve the locality of reference. The analysis in this paper considers the clustering of page faults over time and distinguishes between steady-state behavior and phase transitions. We compare the effectiveness of different restructuring strategies in reducing the amount of main memory needed to obtain desired levels of performance.