Computer architecture is a fast evolving field, mostly because it is driven by rapidly changing technologies. We have all been accustomed to phenomenal improvements in the speed and reliability of computing systems since the mid 1990s, mostly due to technology improvements, faster clock rates, and deeper pipelines. These improvements have had a deep impact on society by bringing high-performance computing to the masses, by enabling the internet revolution and by fostering huge productivity gains in all human activities. We are in the midst of an information revolution of the same caliber as the industrial revolution of the eighteenth century, and few would deny that this revolution has been fueled by advances in technology and microprocessor architecture.
Unfortunately, these rapid improvements in computing systems may not be sustainable in future. Pipeline depths have reached their useful limit, and frequency cannot be cranked up for ever because of power constraints. As technology evolves and on-chip feature sizes shrink, reliability, complexity, and power/energy issues have become prime considerations in computer design, besides traditional measures such as cost, area, and performance. These trends have ushered a renaissance of parallel processing and parallel architectures, because they offer a clear path – some would say the only path – to solving all current and foreseeable problems in architecture. A widespread belief today is that, unless we can unleash and harness the power of parallel processing, the computing landscape will be very different very soon, and this dramatic change will have profound societal impacts.
Email your librarian or administrator to recommend adding this book to your organisation's collection.