Peter Rounce

High latitude thermospheric circulation during the Energy Budget Campaign

Abstract The thermospheric wind data obtained during 16 nights of observation from Esrange during the Energy Budget Campaign (EBC), using a Fabry-Perot interferometer (FPI) to observe the OI 630 emission line, are assessed with the aid of a global theoretical dynamical model of the thermosphere. The systematic behaviour of thermospheric winds, as seen from Kiruna, northern Sweden, is of a poleward day-time wind, which evolves into a nearly ubiquitous equatorward wind between about 1600 UT and 0300 UT. The times of the changes between poleward and equatorward winds are influenced by the nature and intensity of the geomagnetic activity on any given night. The magnitude of the night-time equatorward wind is related to auroral and magnetic activity, but not directly to indices such as K p . Equatorward winds of 400 m s −1 or more occurred at some time on about 50% of the nights reported here, with a highest wind of above 600 m s −1 (11 Nov. 1981). Zonal winds respond more quickly and to a larger extent to magnetic activity than do meridional winds. Under very quiet magnetic conditions, the zonal wind may flow eastward (anti-sunward) until 0200 UT and then reverse to westward until the end of observations at dawn (0700 UT), following the general pattern of winds at middle latitudes due to the solar-driven diurnal pressure gradients. With increasing activity, a period of evening westward winds occurs, usually with only a small meridional component. These westward winds are due to ion drag acceleration in the afternoon/evening sector of the auroral oval, with entrainment of the gas in the region of rapid westward ion motion for periods of many hours due to an approximate balance between Coriolis and inertial acceleration terms. The amplitude of the westward winds increases and their time of onset becomes earlier as the level of geomagnetic activity increases. After about 2000–2100 UT (magnetic midnight at Kiruna) on geomagnetically disturbed nights, there is a sudden onset of strong eastward and equatorward winds, with the onset occurring earlier with increasing activity. For the same ground-level magnetic disturbance at Kiruna, westward (evening) winds exceed eastward (morning) winds by a factor of 2. Westward windsof 500 ms −1 and eastward winds of 300 ms −1 are equally frequent. However, since the eastward winds are usually associated with an approximately equal southward meridional component, the total wind magnitude and energy does not differ too much between the two regimes. Even under very disturbed conditions, the wind usually returns poleward or anti-sunward before 0600 UT. This behaviour has to be interpreted in the context of global circulation, due to an ubiquitous low-latitude solar EUV energy source and a complex high latitude energy and momentum source dependent on geomagnetic activity, but it is also modulated by UT and season, due to the variable solar illumination of the geomagnetic polar cap. At 1200 UT the auroral oval is far poleward of northern Scandinavia (1000 km) except during extremely disturbed conditions, whereas between 1800 and 2200 UT the auroral oval is usually overhead and may be well equatorward under very disturbed conditions. With increasing magnetic activity there is a general equatorward expansion of the auroral oval, so that the time at which the locality comes under the influence of the auroral oval shifts successively to earlier UT and local time. During disturbed periods the cross-polar cap electric potential usually increases, driving ions to higher velocities. The enhancement of ionospheric electron density which occurs virtually throughout the entire auroral oval at the time of geomagnetic disturbances increases both the conductivity, and thus the ionospheric current (auroral electrojet), and also the effectiveness of ion drag in the regions of enhanced convective ion velocities. At magnetic midnight the auroral oval is usually close to Kiruna, however, due to continuity, the thermospheric circulation regime which is observed is dominated by fast anti-sunward winds flowing over the polar cap. This is the reason for the earlier change to eastward and equatorward winds with increasing activity. Intense local heating and ion drag acceleration occurs in the region about magnetic midnight during discrete magnetospheric substorms, which causes marked modulation of the thermospheric wind. There were three occasions during the 16 nights reported here when the local input during a substorm was sufficient to halt or temporarily reverse a pre-existing equatorward wind of 300 ms −1 .

Dynamically Scheduling VLIW Instructions

Very long instruction word (VLIW) machines potentially provide the most direct way to exploit instruction-level parallelism; however, they cannot be used to emulate current general-purpose instruction set architectures. In addition, programs scheduled for a particular implementation of a VLIW model cannot be guaranteed to be binary compatible with other implementations of the same machine model with a different number of functional units or functional units with different latencies. This paper describes an architecture, named dynamically trace scheduled VLIW (DTSVLIW), that can be used to implement machines that execute code of current RISC or CISC instruction set architectures in a VLIW fashion, with backward code compatibility. Preliminary measurements of the DTSVLIW performance, obtained with an execution-driven simulator running the SPECint95 benchmark suite, are also presented.

Dynamically Trace Scheduled VLIW Architectures

This paper presents a new architecture organisation, the dynamically !race scheduled VLIW (DTSVLIW), that can be used to implement machines that execute the code or current RISC or CISC instruction set architectures in a VLIW fashion, with backward code compatibility.

'Slick Systems' and 'Happy Hackers': experience with group projects at UCL

The authors describe their experience of group projects in the practical teaching of software engineering over a period of eight years. Their initial projects tended to be too challenging, and few groups managed to produce complete pieces of work. They have deliberately simplified tasks slightly, resulting in less frustration and better projects, so that students reap more benefits. Students learn about division of work, co-operation with others and scheduling of time. As students are required to provide assessments of other projects and of the contributions of members of their own project group, they are also encouraged to develop critical faculties. The staff effort involved in this method of teaching compares quite reasonably with traditional lectures.

Architectures within the ESPRIT SPAN project

An overview is given of the SPAN project, which pooled the resources of numerous researchers in several countries to integrate symbolic and numeric computing on parallel systems. The resulting Kernel System architectures provided a central model for which two programming languages and two parallel-system architectures were developed. The Kernel System architecture, Parle high-level procedural language, Virtual Machine Code, Sprint processor architecture, and DICE distributed memory architecture are examined.<<ETX>>

Effect of Multicycle Intructions on the Integer Performance of the Dynamixcally Trace Scheduled VLIW Architecture

Dynamically trace scheduled VLIW (DTSVLIW) architectures can be used to implement machines that execute code of current RISC or CISC instruction set architectures in a VLIW fashion, delivering instruction level parallelism (ILP) with backward code compatibility. This paper presents the effect of multicycle instructions on the performance of a DTSVLIW architecture running the SPECint95 benchmarks.

Dynamic instruction scheduling in a trace-based multi-threaded architecture

Simulation results are presented using the hardware-implemented, trace-based dynamic instruction scheduler of our single process DTSVLIW architecture to schedule instructions from several processes into multiple streams of VLIW instructions for execution by a wide-issue, simultaneous multi-threading (SMT) execution engine. The scheduling process involves single instruction execution of each process, dynamically scheduling executed instructions into blocks of VLIW instructions cached for subsequent SMT execution: SMT provides a mechanism to reduce the impact of horizontal and vertical waste, and variable memory latencies, seen in the DTSVLIW. Preliminary experiments explore this extended model. Results achieve PE utilization of up to 87% on a 4-thread, 1-scalar, 8 PE design, with speed-ups of up to 6.3 that of a single processor. Noticeably it only needs a single scalar process to be scheduled at any time, with main memory fetches being 1–4% that of a single processor.

The mDTSVLIW: a Multi-Threaded Trace-based VLIW Architecture

A multi-threaded, dynamically trace-based architecture is presented, designed to make fuller usage of the PEs of a VLIW. The mDTSVLIW extends our single process DTSVLIW design to reduce the effect of both horizontal and vertical waste, and of variable latencies. The DTSVLIW performs single instruction execution of a process, dynamically scheduling these in hardware into blocks of VLIW instructions for VLIW execution. The mDTSVLIW maintains these features and adds simultaneous multithreading by issuing and executing instructions from VLIW blocks from several threads. Preliminary experiments to explore the design varied the number of threads, scalar processors and cache sizes. Results achieve PE utilization of up to 87% on a 4-thread, 1-scalar, 8 PE design, with speed-ups of up to 6.3 that of a single processor. Noticeably it only needs a single scalar process to be scheduled at any time, with main memory fetches being 1-4% that of a single processor

Using databases to support the development of microcode

Abstract The design and verification of microcode for a processor with a complex instruction set can be a complicated task. In this paper it is shown how a standard relational database can be used at various stages to simplify such a design process, as automatic microcode generation is not always suitable for this process. Using the example of a coprocessor for list manipulation, it is demonstrated how the database can be used to enter the microprogram, check various aspects of consistency, generate simulation patterns for the datapath and finally to generate the optimized microcode.

An Architecture for Implementing Control and Signal Processing Neural Networks

This paper presents an architecture that has been developed to implement neural networks for control and signal processing applications. This architecture offers a single chip solution that can be used standalone in small and medium sized systems, or operate as a preprocessor in larger applications.