The Metaflow architecture

The Metaflow architecture, a unified approach to maximizing the performance of superscalar microprocessors, is introduced. The Metaflow architecture exploits inherent instruction-level parallelism in conventional sequential programs by hardware means, without relying on optimizing compilers. It is based on a unified structure, the DRIS (deferred-scheduling, register-renaming instruction shelf), that manages out-of-order execution and most of the attendant problems. Coupling the DRIS with a speculative-execution mechanism that avoids conditional branch stalls results in performance limited only be inherent instruction-level parallelism and available execution resources. Although presented in the context of superscalar machines, the technique is equally applicable to a superpipelined implementation. Lightning, the first implementation of the Metaflow architecture, which executes the Sparc RISC instruction set is described     

Tracking and force control for a class of robotic manipulators

The increasing utilization of robotic manipulators in industrial tasks, such as assembly, forming or shaping of surfaces, and handling hazardous materials, depends greatly on available hybrid force and position control schemes. Since the robot and its environment are often subject to parameter uncertainties that cannot be neglected, it is necessary to design controllers that are robust with respect to these uncertainties. In addition, the dynamics of the robot are nonlinear, requiring consideration of nonlinear control concepts. Another aspect to be taken into account is the relative stiffness of the robot, the force sensor, and the manipulated surface. That is, the behavior of the system normal to the surface is relatively stiff, while that tangential to the surface is relatively free. Separation of the controller for these two directions is therefore indicated. We propose a controller design that accounts for this point of view and demonstrate its efficacy with respect to robustness and accuracy of position and force tracking by means of numerical simulations. The design is based on the control concept of Corless and Leitmann [l]. The example considered is a Manutecr3 robot with three degrees of freedom. In addition, we account for the dynamics of the actuator, which also possesses three degrees of freedom. The considered parameter uncertainties are friction moments in the links and friction between the end effector and the manipulated object, as well as nonlinear dynamics, which are difficult to characterize.Recommended by J. Skowronski     

Theory of residence-time control by output feedback

The problem of residence-time control by the observer-based output feedback is formulated and solved for the case of linear systems with small additive input noise. Both noiseless and noisy measurements are considered. In the noiseless measurements case, it is shown that the fundamental bounds on the achievable residence time depend on the nonminimum phase zeros of the system. In the noisy measurements case, the achievable residence time is shown to be always bounded, and an estimate of this bound is given. Controller design techniques are presented. The development is based on the asymptotic large deviations theory. Recommended by T K. Caughey     

A cost-effective approach to testing

A method for assessing the differential risk of failure among a system's modules is proposed. The procedure has three components: external-risk assessment, module exposure, and module-failure likelihood. External-risk assessment is a consideration of the system's environment, almost independent of the software's details. To estimate module exposure, the model relates individual modules and their potential faults to the external-failure modes and their economic consequences by reverse-engineering the specifications and analyzing each module's expected use. To estimate failure likelihood, the method uses a reliability model. The method constitutes theoretical foundation for the cost-effective development of software that attempts to reduce the risk of failure. Managers can use the failure-risk estimates to better determine how much testing effort can be economically justified