Special issue on specialized computer architecture simulators that see the present and may hold the future

Abstract

Welcome to this special issue of the Journal of Educational Resources in Computing (JERiC), which is focused on the topic of specialized computer architecture simulators, and is part of back-to-back JERiC issues describing the state-of-the-art in computer architecture simulators for educational purposes. JERiC is a unique forum, in that the simulation software corresponding to each of the articles in this issue is available for download and execution. The guest editorial in the previous issue of JERiC described how these back-to-back special issues came to be dedicated to computer architecture simulators. In this issue I would like to focus on promoting creativity as an important part of being an educator. While the general von Neumann architecture still predominates as it has for the last halfcentury, there are clearly new, advanced technologies on the horizon. While part of our role as educators is to help students master existing technology, we must also nurture the seeds of creativity that may someday break the current paradigms. While it is unclear what the future may hold -- perhaps biological, molecular, or quantum computing -- some clues can be found in current advanced, specialized research. Just as it is recursive to teach how a computer works by simulating a virtual computer on a computer, we can also learn from past and present specialized areas to begin developing the future. In this issue we present interactive simulation tools for teaching the specialized areas of computer architecture. The articles span from digital logic simulators (Logisim) to historic machine simulators (Atari-6502, PDP-8), to a real machine simulator (DEC Alpha/KScalar), to an embedded system (LegoSim), to parallel computer architectures (High Performance Computer Architecture and Algorithm Simulator). Although these simulators are clearly useful for teaching general computer architecture, they are perhaps most valuable for specialized topics in advanced or elective courses. This issue begins with Logisim, a visual simulator for building and testing digital logic circuits. While general computer architecture survey courses have abstracted away from lower-level digital logic circuits, this teaching tool is invaluable for computer engineering or electrical engineering courses that focus specifically on logic design. Logisim can be used hierarchically, and is powerful; the author notes that he has successfully created a simple 8-bit CPU in the system. The use of computer simulators provides a number of secondary benefits -- the most pertinent is highlighted in the next article on the PDP-8 Emulator. Simulators make it feasible to interact with computers that no longer exist physically (except possibly in a museum). Thus, obsolescence is no longer the dominant obstacle in teaching computer architecture. An extinct machine is oftentimes the best example of an architectural concept or may have the best educational documentation. A perfect example of an historic architecture to study is the PDP-8. The PDP-8 was a breakthrough 12-bit machine in 1965, to which many of our modern architectural features can be traced. Through this emulator, students can now interact with a virtual PDP-8, instead of simply reading about it in a textbook. The next two articles are related because they both supplement seminal textbooks in the field by Hennessy and Patterson [1996] and Patterson and Hennessy [1997]. While the DLX, MIPS, and SPIM series of simulators are well documented, SLOOP-SMOK and KScalar are new simulators that complement these established tools. The SLOOP-SMOK toolkit emulates an Atari 6502 (a good choice for student motivation) integrated with a machine organization simulator. The KScalar simulator simulates the DEC Alpha with a wide range of possible scalar performance enhancements such as pipelining, caching, and speculative execution. LegoSim is an important contribution because it focuses on the simulation of an embedded system. Embedded systems, defined as a component in a larger system that relies on its own processor, have been around for decades. However, it was not until the early 1980s, when 16-bit processors performing sophisticated specialized applications began to appear in automobiles, that designers realized the embedded system's almost limitless possibilities for ubiquitous everyday products [Wolf 2002]. However, at present, almost all of our educational focus is still on general-purpose computers, even though the number of specialized embedded systems currently dwarfs traditional computing machines by orders of magnitude (a fact that we have ignored in part), and this gap is increasing. A recent study by the National Academy of Science documents this development [Computer Science and Telecommunications Board 2001]. It is my hope that LegoSim will be the first in a long line of teaching tools for this rapidly expanding area of study. The final article in this issue simulates the tradeoffs of efficiency and performance inherent in parallel architectures. The High-Performance Computer Architecture and Algorithm Simulator allows students to design and interact with different forms of parallelism such as multiple pipelines, multiple processors, clustered computer systems, and distributed computing systems. While the promise of parallel computing has yet to be realized on a commercial scale (to be fair, parallel programming is at least equally to blame), research continues, and students who learn by using this tool may make the difference. The underlying thread weaving through all these articles is the interaction with the details of computer architecture: how a computer processes bits at the lowest level (LogiSim), why computers developed the way they did (KScalar, PDP-8, SLOOP-SMOK), the future capabilities of embedded processors (LegoSim), and how to make computers work together (High-Performance Computer Architecture and Algorithm Simulator). In this issue it is our intent that the specialized computer architecture simulators described here provide powerful teaching tools to facilitate and enhance student creativity for developing the next paradigms of computer architecture. Finally, I have very much enjoyed serving JERiC and ACM in my capacity as guest editor, and I would like to repeat some acknowledgements from my previous guest editorial, lest anyone think I assembled this issue myself. I would like to thank JERiC Co-Editors-in- Chief Lillian (Boots) Cassel and Ed Fox, Deborah Knox of the Computer Science Teaching Center (http://www.cstc.org/), and Jono Hardjowirogo of the ACM Press. I would also like to especially thank the many authors and peer-reviewers with whom I have traded email (about 60 people); they have been responsive, patient, and accommodating. This issue would not have been possible without them. A special word to the authors of the articles and simulators we declined: acceptance decisions were particularly difficult (60% of submissions were declined). But I encourage you to persevere - this is truly a growth area and your contributions are needed. Again, I W. Yurcik hope you enjoy what we have assembled and that it inspires many ideas for teaching and research!