Increasing demands on computing power have spurred the development of faster, higher-density Integrated Circuits (ICs), compounding power and complexity concerns in design budgets. The clock distribution network is a significant contributor to such power and complexity concerns. Resonant rotary clocking is a relatively new technology that realizes several benefits over current clocking methods, including power, frequency, and variation tolerance, yet lacks the automation tools to promote increased use. Towards this end, an automated rotary clock routing methodology is presented that generates custom topology rotary ring routes with tree subnetworks. In addition to the benefits of adiabatic clocking, the presented custom topology router permits 38.6% shorter wirelengths on average for register tapping, compared to traditional prescribed skew, binary tree routing. ACM Reference Format:Taskin, B., DeMaio, J., Farell, O., Hazeltine, M., and Ketner, R. 2009. Custom topology rotary clock router with tree subnetworks.

Experiments were conducted in which Ss received to-be-remembered sequences of two, three, or four simultaneous pairs of digits. Both digits of each pair were recorded by the same male speaker and both were presented binaurally, thus eliminating cues of spatial location and voice by which Ss could "channel" their reports. Even in the absence of these stimulus channels, Ss reported the digits sequentially. High bias ratings in the first experiment suggested the possibility that sequential reports were induced by uncontrolled stimulus characteristics (e.g., temporal synchrony, intensity, and pitch). Pulse-coded speech stimuli, which provided greater control over nonlinguistic stimulus features, were used in the terminal experiment. Bias ratings were reduced, but the majority of Ss continued to report sequentially. These results suggest that the presence of stimulus channels is not a necessary condition for the occurrence of sequential reporting.In the most commonly employed variant of the dichotic memory task, Ss receive a sequence of simultaneous pairs of digits, one member of each pair delivered to the right ear and one to the left ear, with instructions to reproduce, in the order of their choice, as many of the digits as possible. Given rapid stimulus presentation, highly consistent recall patterns are obtained with this experimental procedure (e.g., Broadbent, 1954;Bryden, 1962). There are two important aspects of these recall patterns: first, the reports are sequential, i.e., Ss report one digit from the first pair received, then one digit from the second pair, and one digit from the third pair. This set of digits (sometimes called the first half-set) is followed in recall by the remaining members of the first, second, and third pairs, respectively. Second, recall is "channeled" or grouped according to spatial location. The Ss recall all of the digits received by the right ear followed by those presented to the left ear, or vice versa. If both members of each digit pair are presented bin aurally (thus removing the cue of spatial location) but are made to differ along dimensions of pitch or intensity, Ss still report sequentially and their reports are channeled according to those dimensions (Broadbent, 1956).While "channel-by-channel" reporting has formed the basis for models of memory and attention (e.g., Broadbent, 1958Broadbent, , 1971Yntema & Trask, 1963), the relation between sequential reporting and channeling has received relatively little interest. The purpose of the present research is to determine whether the presence of channels is a necessary condition for the occurrence of sequential reporting. In previous research directed toward this question, Savin (1967) changed the usual dichotic paradigm by presenting both recorded messages over a single loudspeaker. Stimulus lists consisted of two pairs of digits recorded by the same speaker, thus channels, in the sense in which they have been defined (intensity, pitch, and spatial location), were effectively eliminated in this experiment. If channels were n...

No abstract

Detection and latency performance of instructor pilots and students pilots was compared in a task which required searching an aircraft instrument display for target deviations from a desired course. Performance of instructors was superior to that of students on both detection and latency measures. The use of stable fixation queues was reflected in the latency performance of students. The latency performance of instructors was much less affected by the use of such queues. Results were interpreted to suggest that experienced pilots are able to place a greater reliance on peripheral vision in scanning instrument displays.

No abstract