Name
Abstract | ||
|---|---|---|
Performance is the raw material for computing. For more than 40 years, consistent and exponential improvement in transistor scaling coupled with continuous advances in general-purpose processor design has exponentially reduced its cost. However, as we enter the dark silicon era (as we projected in our study [1, 2] and others corroborate [3]), the benefits from transistor scaling are diminishing and the current paradigm of processor design significantly falls short of the traditional cadence of performance improvements due to power limitations. Performance has hit the power wall. These shortcomings can drastically curtail the industry's ability to continuously deliver new capabilities, breaking the backbone of its economic ecosystem. These challenges have coincided with the big data revolution. The rate of data generation is increasing at an overwhelming rate that is beyond the capabilities of current computing systems to match. Expert analyses show that the zettabyte barrier was cracked in 2010. In 2011, the amount of information generated surpassed 1.8 zettabytes (trillion gigabytes). By 2020, consumers will generate 50x this staggering figure [4]. While data generation is quadrupling each year, modern processors have seen a performance improvement of roughly 10--15% every two years. Worse yet, the long-standing memory sub-system bottlenecks--long access latency, bounded communication bandwidth, and limited capacity--leave little hope of managing this explosion of data through traditional incremental improvements. Moreover, many IoT objects (wearable devices/environmental sensors) must operate on hard-to-replace batteries or harvest energy from intermittent ambient sources. The application timeliness and the limited capacity of the wireless link compound the challenges arising from the power constraints. Given the scale of the problem, transformative research is essential across many domains, including computing, communication, and even control.
|
Year | DOI | Venue |
|---|---|---|
2015 | 10.1109/CASES.2015.7324540 | International Conference on Compilers, Architectures, and Synthesis for Embedded Systems |
Keywords | Field | DocType |
approximate acceleration,dark silicon,big data,cross-stack solutions,data communication,data storage | Dark silicon,Wireless,Computer science,Parallel computing,Real-time computing,Processor design,Side channel attack,Zettabyte,Big data,Test data generation,Performance improvement | Conference |
ISSN | ISBN | Citations |
2381-1560 | 978-1-4673-8320-2 | 0 |
PageRank | References | Authors |
0.34 | 7 | 1 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| H. Esmaeilzadeh | 1 | 1443 | 69.71 |