27 December 2018 in Proceedings of the IEEE
The world's appetite for analyzing massive amounts of structured and unstructured data has grown dramatically. The computational demands of these abundant-data applications, such as deep learning, far exceed the capabilities of today's computing systems and are unlikely to be met with isolated improvements in transistor or memory technologies, or integrated circuit architectures alone. To achieve unprecedented functionality, speed, and energy efficiency, one must create transformative nanosystems whose architectures are based on the salient properties of the underlying nanotechnologies. Our Nano-Engineered Computing Systems Technology (N3XT) approach makes such nanosystems possible through new computing system architectures leveraging emerging device (logic and memory) nanotechnologies and their dense 3-D integration with fine-grained connectivity to immerse computing in memory and new logic devices (such as carbon nanotube field-effect transistors for implementing high-speed and low-energy logic circuits) as well as high-density nonvolatile memory (such as resistive memory), and amenable to ultradense (monolithic) 3-D integration of thin layers of logic and memory devices that are fabricated at low temperature. In addition, we explore the use of several device and integration technologies in the N3XT beyond the specific ones mentioned earlier that are also used in our main nanosystem prototypes. We also present an efficient resiliency technique to overcome endurance challenges in certain resistive memory technologies. N3XT hardware prototypes demonstrate the practicality of our architectures. We evaluate the benefits of the N3XT using a simulation framework calibrated using experimental measurements. System-level energy-delay product of common implementations of abundant-data workloads improves by three orders of magnitude in the N3XT compared with conventional architectures. These improvements impact a broad range of application workloads and architecture configurations, from embedded systems to the cloud.
1 October 2003 in Quantum Information Processing
20 May 2008 in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
In this paper, we describe significant improvements to core routing technologies and outperform the best results from the International Symposium on Physical Design 2007 Global Routing Contest and the International Conference on Computer-Aided Design 2007 in terms of route completion and total wirelength.
19 March 2007 in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
We demonstrate that Steiner-tree wirelength (StWL) correlates with routed wirelength (rWL) much better than the more common half-perimeter wirelength (HPWL) objective. Therefore, we develop a technique to optimize StWL in global and detail placement without a significant runtime penalty. This new optimization, along with congestion-driven whitespace distribution, improves overall Place-and-Route results, making the use of HPWL unnecessary. Additionally, our empirical results provide ample evidence that the fidelity of net-length estimates is more important than their accuracy in Place-and-Route. The new data structures that make our min-cut algorithms fast can also be useful in multilevel analytical placement. Our placement algorithm Rigorous Optimization Of Steiner-Trees Eases Routing (ROOSTER) outperforms the best published results for Dragon, Capo, FengShui, mPL-R/WSA, and APlace in terms of rWL by 10.7%, 5.6%, 9.3%, 5.5%, and 4.2%, respectively. Via counts, which are especially important at 90 nm and below, are improved by 15.6% over mPL-R/WSA and 11.9% over APlace
14 January 2013 in Physical Review A
A major obstacle to implementing Shor's quantum number-factoring algorithm is the large size of modular-exponentiation circuits. We reduce this bottleneck by customizing reversible circuits for modular multiplication to individual runs of Shor's algorithm. Our circuit-synthesis procedure exploits spectral properties of multiplication operators and constructs optimized circuits from the traces of the execution of an appropriate greatest-common-divisor algorithm. Empirically, gate counts are reduced by four to five times, and circuit latency is reduced by larger factors. DOI: http://dx.doi.org/10.1103/PhysRevA.87.012310 ©2013 American Physical Society
19 January 2012 in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
Despite numerous advantages of 3-D integrated circuits (ICs), their commercial success remains limited. In part, this is due to the wide availability of trustworthy intellectual property (IP) blocks developed for 2-D ICs and proven through repeated use. Block-based design reuse is imperative for heterogeneous 3-D ICs where memory, logic, analog, and microelectromechanical systems dies are manufactured at different technology nodes and circuit modules cannot be partitioned among several dies. In this paper, we show how to integrate 2-D IP blocks into 3-D chips without altering their layout. Experiments indicate that the overhead of proposed integration is small, which can help accelerate industry adoption of 3-D integration.