The performance of a processor is limited by the specific bottlenecks that a benchmark exposes while running on that processor. Since the quantification of these bottlenecks can be extremely time-consuming, our prior work proposed using... more
The performance of a processor is limited by the specific bottlenecks that a benchmark exposes while running on that processor. Since the quantification of these bottlenecks can be extremely time-consuming, our prior work proposed using the Plackett and Burman design as a statistically-rigorous, but time- efficient method of determining the processor's most significant performance bottlenecks. In this paper, we use
Research Interests:
Research Interests:
Research Interests:
Abstract Multi-level cell (MLC) NAND flash memories are very popular storage media because of their power efficiency and big storage density. This paper proposes to use nonlinear t-error-correcting codes to replace linear BCH codes for... more
Abstract Multi-level cell (MLC) NAND flash memories are very popular storage media because of their power efficiency and big storage density. This paper proposes to use nonlinear t-error-correcting codes to replace linear BCH codes for error detection and correction in MLC NAND flash memories. Compared to linear BCH codes with the same bit-error correcting capability t, the proposed codes have less errors miscorrected by all codewords and nearly no undetectable errors. For example, the proposed (8281, 8201, 11 ...
Research Interests:
Research Interests:
Research Interests:
Research Interests:
We report a B3LYP density-functional theory (DFT) analysis of O(2) adsorption on 27 Au(n)M(m) (m, n = 0-3 and m + n = 2 or 3; M = Cu, Ag, Pd, Pt, and Na) clusters. The LANL2DZ pseudopotential and corresponding double-zeta basis set was... more
We report a B3LYP density-functional theory (DFT) analysis of O(2) adsorption on 27 Au(n)M(m) (m, n = 0-3 and m + n = 2 or 3; M = Cu, Ag, Pd, Pt, and Na) clusters. The LANL2DZ pseudopotential and corresponding double-zeta basis set was used for heavy atoms, while a 6-311+G(3df) basis set was used for Na and O. We employed basis-set superposition error (BSSE) corrections in the electronic adsorption energies at 0 K (deltaE(ads)) and also calculated adsorption thermodynamics at standard conditions (298.15 K and 1 atm), i.e., internal energy of adsorption (deltaU(ads)) and Gibbs free energy of adsorption (deltaG(ads)). Natural Bond Orbital (NBO) analysis showed that all the clusters donated electron density to adsorbed O(2) and we successfully predicted intuitive linear correlations between the NBO charge on adsorbed O(2), O-O bond length, and O-O stretching frequency. Although there was no clear trend in the O(2) binding energy (BE = -deltaE(ads)) on pure and alloy dimers, we found the following interesting trend for trimers: BE (MAu(2)) < BE (M(3)) < or = BE (M(2)Au). The alloy trimers containing only one Au atom are most reactive toward O(2) while those with two Au atoms are least reactive. These trends are discussed in the context of the ensemble effect and coulomb interactions. We found an approximate linear correlation between the O(2) BE and charge transfer to O(2) for all 27 clusters. The clusters having strongly electropositive Na atoms (e.g., Na(3) and Na(2)Au) donated almost one full electron to adsorbed O(2), and the BE is maximum on these clusters. Although O(2) dissociation is likely in such cases, we have restricted this study to trends in the adsorption of molecular O(2) only. We also found an approximate linear correlation between the charge transfer and BE versus energy difference between the bare-cluster HOMO and O(2) LUMOs, which we speculate to be a fundamental descriptor of the reactivity of small clusters toward O(2). Part of the scatter in these correlations is attributed to the differences in the O(2) binding orientations on different clusters (geometric effect). Relatively higher bare-cluster HOMO energy eases the charge transfer to adsorbed O(2) and enhances the reactivity toward O(2). The Frontier Orbital Picture (FOP) is not always useful in predicting the most favorable O(2) binding site on clusters. It successfully predicted the cluster-O(2) ground-state configurations for 10 clusters, but failed for the others. Finally, the energetics of fragmentation suggest that the bare and O(2)-covered clusters reported here are stable.
Research Interests:
This paper presents the results of a study in which the case study method was used to sort out the internal and external conditions that might affect the success of a work site trip reduction program. Investigators attempted to disprove a... more
This paper presents the results of a study in which the case study method was used to sort out the internal and external conditions that might affect the success of a work site trip reduction program. Investigators attempted to disprove a null hypothesis, stated as "The effectiveness of work site trip reduction programs does not depend on organizational culture." The
Research Interests:
Research Interests:
Page 1. Moving Shadow Detection with Low-and Mid-Level Reasoning Ajay J. Joshi, Stefan Atev, Osama Masoud, and Nikolaos Papanikolopoulos Dept. of Computer Science and Engineering, University of Minnesota Twin Cities... more
Page 1. Moving Shadow Detection with Low-and Mid-Level Reasoning Ajay J. Joshi, Stefan Atev, Osama Masoud, and Nikolaos Papanikolopoulos Dept. of Computer Science and Engineering, University of Minnesota Twin Cities {ajay,atev,masoud,npapas}@cs.umn.edu ...
Research Interests:
We report a detailed density functional theory (B3LYP) analysis of the gas-phase H2O2 formation from H2 and O2 on Au3, Au4+, Au5, and Au5-. We find that H2, which interacts only weakly with the Au clusters, is dissociatively added across... more
We report a detailed density functional theory (B3LYP) analysis of the gas-phase H2O2 formation from H2 and O2 on Au3, Au4+, Au5, and Au5-. We find that H2, which interacts only weakly with the Au clusters, is dissociatively added across the Au-O bond, upon interaction with AunO2. One H atom is captured by the adsorbed O2 to form the hydroperoxy intermediate (OOH), while the other H atom is captured by the Au atom. Once formed, the hydroperoxy intermediate acts as a precursor for the closed-loop catalytic cycle. An important common feature of all the pathways is that the rate-determining step of the catalytic cycle is the second H2 addition to form H2O2. The H2O2 desorption is followed by O2 addition to AunH2 to form the hydroperoxy intermediate, thus leading to the closure of the cycle. On the basis of the Gibbs free energy of activation, out of these four clusters, Au4+ is most active for the formation of the H2O2. The 0 K electronic energy of activation and the DeltaGact at the standard conditions are 12.6 and 16.6 kcal/mol respectively. The natural bond orbital charge analysis suggests that the Au clusters remain positively charged (oxidic) in almost all the stages of the cycle. This is interesting in the context of the recent experimental evidence for the activity of cationic Au in CO oxidation and water-gas shift catalysts. We have also found preliminary evidence for a correlation between the activation barrier for the first H2 addition and the O2 binding energy on the Au cluster. It suggests that the minimum activation barrier for the first H2 addition is expected for the Au clusters with 7.0-9.0 kcal/mol O2 binding energy, i.e., in the midrange of the expected interaction energy. This represents a balance between more favorable H2 dissociation when the Aun-O2 interaction is weaker and high O2 coverage when the interaction is stronger. On the basis of this work, we predict that the hydroperoxy intermediate formation can be both thermodynamically and kinetically viable only in a narrow range of the O2 binding energy (10.0-12.0 kcal/mol)-a useful estimate for computationally screening Au-cluster-based catalysts. We also show that a competitive channel for the OOH desorption exists. Thus, in propylene epoxidation both OOH radicals and H2O2 can attack the active Ti in/on the Au/TS-1 and generate the Ti-OOH sites, which can convert propylene to propylene oxide.
Research Interests:
Research Interests:
We performed density-functional theory analysis of nondissociative CO adsorption on 22 binary Au-alloy (Au(n)M(m)) clusters: n=0-3, m=0-3, and m+n=2 (dimers) or 3 (trimers), M=Cu/Ag/Pd/Pt. We report basis-set superposition error... more
We performed density-functional theory analysis of nondissociative CO adsorption on 22 binary Au-alloy (Au(n)M(m)) clusters: n=0-3, m=0-3, and m+n=2 (dimers) or 3 (trimers), M=Cu/Ag/Pd/Pt. We report basis-set superposition error corrections to adsorption energies and include both internal energy of adsorption (DeltaU(ads)) and Gibbs free energy of adsorption (DeltaG(ads)) at standard conditions (298.15 K and 1 atm). We found onefold (atop) CO binding on all the clusters except Pd2 (twofold/bridged), Pt2 (twofold/bridged), and Pd3 (threefold). In agreement with the experimental results, we found that CO adsorption is thermodynamically favorable on pure Au/Cu clusters but not on pure Ag clusters and also observed the following adsorption affinity trend: Pd>Pt>Au>Cu>Ag. For alloy dimers we found the following patterns: Au2>M Au>M2 (M=Ag/Cu) and M2>M Au>Au2 (M=Pd/Pt). Alloying Ag/Cu dimers with (more reactive) Au enhanced adsorption and the opposite effect was observed for PdPt dimers. The Ag-Au, Cu-Au, and Pd-Au trimers followed the trends observed on dimers: Au3>M Au2>M2Au>M3 (M=Ag/Cu) and Pd3>Pd2Au>PdAu2>Au3. Interestingly, Pt-Au trimers reacted differently and alloying with Au systematically increased the adsorption affinity: PtAu2>Pt2Au>Pt3>Au3. A strikingly different behavior of Pt is also manifested by the triplet spin state and onefold (atop) binding in Pt3-CO which is in contradiction with the singlet spin state and threefold binding in Pd3-CO. We found a linear correlation between CO binding energy (BE) and elongation of the CO bond. For Ag-Au and Cu-Au clusters, the increase in CO BE (and elongation of the C-O bond which is probably due to the back donation) is accompanied by the decrease in the cluster-CO distance suggesting that the donation (from 5sigma highest occupied molecular orbital in CO to cluster lowest unoccupied molecular orbital) mechanism also contributes to the BE. For Pd-Au clusters, the cluster-CO distance (and CO bond length) increases with increase in the BE, suggesting that the donation mechanism may not be important for those clusters. No clear trend was observed for Pt-Au clusters.
Research Interests:
Page 1. Learning of Moving Cast Shadows for Dynamic Environments Ajay J. Joshi and Nikolaos Papanikolopoulos Department of Computer Science and Engineering University of Minnesota-Twin Cities {ajay,npapas}@cs.umn.edu ...
Research Interests:
Research Interests:
Research Interests:
Research Interests:
This paper will present the initial findings of a research into distributed computer rendering. The goal of the research is to create a distributed computer system capable of rendering a 3D model into an MPEG-4 stream. This paper outlines... more
This paper will present the initial findings of a research into distributed computer rendering. The goal of the research is to create a distributed computer system capable of rendering a 3D model into an MPEG-4 stream. This paper outlines the initial design, software architecture and hardware setup for the system. Distributed computing means designing and implementing programs that run on
Research Interests:
Research Interests:
Abstract As we approach serious scaling roadblocks in the next few process nodes, it is imperative to identify new emerging technologies that can complement or supplant CMOS in the future. We present an integrated cyclic approach to... more
Abstract As we approach serious scaling roadblocks in the next few process nodes, it is imperative to identify new emerging technologies that can complement or supplant CMOS in the future. We present an integrated cyclic approach to explore new interconnect technologies in the nanometer era for many core systems, where on-chip interconnects are jointly optimized at all the levels in the design hierarchy to develop a complete interconnect solution-from interconnect technology to network topology.
Abstract The aim of this work is to propose such a framework, called BenchMaker, for constructing synthetic benchmarks whose code properties can be easily altered. BenchMaker enables altering inherent workload characteristics of a program... more
Abstract The aim of this work is to propose such a framework, called BenchMaker, for constructing synthetic benchmarks whose code properties can be easily altered. BenchMaker enables altering inherent workload characteristics of a program by varying a ...
Research Interests:
Research Interests:
It is essential that a subset of benchmark programs used to evaluate an architectural enhancement, is well distributed within the target workload space rather than clustered in specific areas. Past efforts for identifying subsets have... more
It is essential that a subset of benchmark programs used to evaluate an architectural enhancement, is well distributed within the target workload space rather than clustered in specific areas. Past efforts for identifying subsets have primarily relied on using microarchitecture- ...
Research Interests:
On August 24, 2006, the Standard Performance Evalua-tion Corporation (SPEC) announced CPU2006 the next gen-eration of industry-standardized CPU-intensive benchmark suite. The SPEC CPU benchmark suite has become the most frequently used... more
On August 24, 2006, the Standard Performance Evalua-tion Corporation (SPEC) announced CPU2006 the next gen-eration of industry-standardized CPU-intensive benchmark suite. The SPEC CPU benchmark suite has become the most frequently used suite for simulation-...
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests: Distributed Computing, VLSI, Computer Hardware, System Design, Low Power, and 15 moreLow Power Electronics, Energy Dissipation, High performance, Chip, Network Routing, Crosstalk, Network on chip, Optimal Design, Interconnection, Time Division Multiplexing, Electrical And Electronic Engineering, Power Dissipation, Constrained Optimization, Repeater, and Integrated Circuit Design
Research Interests:
Abstract In modern-day VLSI systems, performance and manufacturing costs are being driven by the on-chip wiring needs due to the continuous increase in the number of transistors. This paper proposes a low overhead wave-pipelined... more
Abstract In modern-day VLSI systems, performance and manufacturing costs are being driven by the on-chip wiring needs due to the continuous increase in the number of transistors. This paper proposes a low overhead wave-pipelined multiplexed (WPM) routing technique that harnesses the inherent intraclock period interconnect idleness to implement wire sharing throughout the various hierarchical levels of design. It is illustrated in this paper that the WPM network can be readily incorporated into future gigascale integration (GSI) ...
Research Interests:
Research Interests:
Research Interests: Information Systems, Algorithms, Artificial Intelligence, Computer Vision, Support Vector Machines, and 14 moreSemi-supervised Learning, Image Classification, Probability Distribution & Applications, Dynamic Software Adaptation, Support vector machine, Motion, Image Enhancement, Training Algorithm, Moving Object Recognition, Dynamic Environment, Electrical And Electronic Engineering, Feature Space, Statistical Distributions, and Shadow Detection
Research Interests:
Research Interests:
Research Interests:
Research Interests:
A synthetic workload whose performance correlates well with long-running application programs is of great benefit to the computer architecture community because it reduces simulation time, fosters benchmark sharing by abstracting... more
A synthetic workload whose performance correlates well with long-running application programs is of great benefit to the computer architecture community because it reduces simulation time, fosters benchmark sharing by abstracting proprietary codes, and ...
Research Interests:
AbstractWe present a model for estimating the power consumption of SDRAM at an architectural level. The approach is based on identifying the various operating states for a typical SDRAM, and using the knowledge of current drawn by the... more
AbstractWe present a model for estimating the power consumption of SDRAM at an architectural level. The approach is based on identifying the various operating states for a typical SDRAM, and using the knowledge of current drawn by the memory chip, and fraction of ...