Due to their energy efficiency, heterogeneous Multi-Processor Systems-on-Chip (MPSoCs) are widely deployed as the engines for modern-day smart devices. As smart devices prevail, heterogeneous MPSoCs have become ubiquitous in our daily lives. These MPSoCs typically integrate a diverse collection of cores, including real-time agents such as the GPU, the DSP and video codec, as well as general-purpose cores such as the CPU. Different from traditional multi-core processors where memory sharing is mostly handled by the cache hierarchy, a heterogeneous system mainly uses DRAM as the medium for data sharing. That makes the memory subsystem, including networks-on-chip (NoC), last-level caches and memory controllers, a frequently shared resource among heterogeneous cores.

As traffic streams travel through the shared memory subsystem, memory interference is inevitable, which is even worsened by their disparate traffic patterns. As a result, the memory subsystem plays a crucial role in heterogeneous systems and has tremendous impact on system performance.

In particular, the design goal of the memory subsystem is two-fold: delivering end-to-end Quality-of-Service (QoS) to heterogeneous cores and improving memory efficiency in DRAM.

In this dissertation, we explore several design aspects of the memory subsystem for heterogeneous MPSoCs. First, we propose a self-aware resource allocation framework to enable distributed performance monitoring, priority-based adaptation and instant memory response. The proposed framework meets a diverse range of QoS demands from real-time cores. It can also configure itself in runtime to accommodate best-effort cores and real-time cores that are particularly prone to QoS failure. In the rest of the dissertation, we further look into each major memory fabric. We present the single-tier virtual queuing memory controller that maintains a single tier of request queues and employs an efficacious scheduler that considers both QoS requirements and DRAM bank states. Then we study the orchestration between last-level cache controller and memory scheduler to improve memory visibility and avoid unnecessary precharges and row activations in DRAM. At last, we introduce a locality-aware NoC router to prevent row-buffer locality dilution while providing QoS-aware service to heterogeneous cores.

The intense research interest in nanoscience and nanotechnology is largely fueled by the unique properties of nanoscale materials. In this dissertation, the research efforts are focused on surface functionalization and interfacial engineering of functional nanoparticles in the preparation of patchy nanoparticles (e.g., Janus nanoparticles and Neapolitan nanoparticles) such that the nanoparticle structures and properties may be manipulated to an unprecedented level of sophistication.

Experimentally, Janus nanoparticles were prepared by an interfacial engineering method where one hemisphere of the originally hydrophobic nanoparticles was replaced with hydrophilic ligands at the air|liquid or solid|liquid interface. The amphiphilic surface characters of the Janus nanoparticles were verified by contact angle measurements, as compared to those of the bulk-exchange counterparts where the two types of ligands were distributed rather homogeneously on the nanoparticle surface.

In a further study, a mercapto derivative of diacetylene was used as the hydrophilic ligands to prepare Janus nanoparticles by using hydrophobic hexanethiolate-protected gold nanoparticles as the starting materials. Exposure to UV irradiation led to effective covalent cross-linking between the diacetylene moieties of neighboring ligands and hence marked enhancement of the structural integrity of the Janus nanoparticles, which was attributable to the impeded surface diffusion of the thiol ligands on the nanoparticle surface, as manifested in fluorescence measurements of aged nanoparticles.

More complicated bimetallic AgAu Janus nanoparticles were prepared by interfacial galvanic exchange reactions of a Langmuir-Blodgett monolayer of 1-hexanethiolate-passivated silver nanoparticles on a glass slide with gold(I)-mercaptopropanediol complex in a water/ethanol solution. The resulting nanoparticles exhibited an asymmetrical distribution not only of the organic capping ligands on the nanoparticle surface but also of the metal elements in the nanoparticle cores, in contrast to the bulk-exchange counterparts where these distributions were homogeneous within the nanoparticles, as manifested in contact angle, UV-vis, XPS, and TEM measurements. More interestingly, the electrocatalytic performance of the Janus nanoparticles was markedly better than the bulk-exchange ones, suggesting that the segregated distribution of the polar ligands from the apolar ones might further facilitate charge transfer from Ag to Au in the nanoparticle cores, leading to additional improvement of the adsorption and reduction of oxygen.

This interfacial protocol was then adopted to prepare trimetallic Ag@AuPt Neapolitan nanoparticles by two sequential galvanic exchange reactions of 1-hexanethiolate-capped silver nanoparticles with gold(I)-thiomalic acid and platinum(II)-hexanethiolate complexes. As both reactions were confined to an interface, the Au and Pt elements were situated on two opposite poles of the original Ag nanoparticles, which was clearly manifested in elemental mapping of the nanoparticles, and consistent with the damping and red-shift of the nanoparticle surface plasmon resonance.

As nanoscale analogs to conventional amphiphilic molecules, the resulting Janus nanoparticles were found to form oil-in-water micelle-like or water-in-oil reverse micelle-like superparticulate structures depending on the solvent media. These unique characteristics were exploited for the effective transfer of diverse guest nanoparticles between organic and water phase. The transfer of hydrophobic nanoparticles from organic to water media or water-soluble nanoparticles to the organic phase was evidenced by TEM, DLS, UV-Vis, and PL measurements. In particular, line scans based on EDS analysis showed that the vesicle-like structures consisted of multiple layers of the Janus nanoparticles, which encapsulated the guest nanoparticles in the cores. The results highlight the unique effectiveness of using Janus nanoparticles in the formation of functional nanocomposites.

Part of the dissertation research was also devoted to graphene quantum dots (GQDs)-supported platinum (Pt/G) nanoparticles and their electrocatalytic activity in oxygen reduction reaction. These Pt/G nanocomposites were prepared by a hydrothermal procedure at controlled temperatures. Spectroscopic measurements based on FTIR, Raman and XPS confirmed the formation of various oxygenated structural defects on GQDs and the variation of their concentrations with the hydrothermal conditions. Interestingly, electrocatalytic activity of GQD/Pt composites exhibited a volcano-shaped variation with the GQD structural defects, with the best identified as the samples prepared at 160 °C for 6 h where the mass activity was found to meet the DOE target for 2015. This remarkable performance was accounted for by the deliberate manipulation of the adsorption of oxygen and reaction intermediates on platinum by the GQD structural defects through partial charge transfer. The strategy presented herein may offer a new paradigm in the design and engineering of nanoparticle catalysts for fuel cell electrochemistry.

In addition, studies were also carried out to study intervalence charge transfer between ferrocenyl moieties bonded on carbon nanoparticle surfaces by diazonium reaction. Electrochemical studies exhibited two pairs of voltammetric waves with a difference of their formal potentials at about 78 mV, suggesting nanoparticle-mediated intraparticle charge delocalization at mixed valence as a result of the strong core-ligand covalent bonds and the conductive sp2 carbon matrix of the graphitic cores. Consistent behaviors were observed in near-infrared measurements, indicating that the particles behaved analogously to a Class I/II mixed-valence compound.

SUMOylation is an important post-translational protein modification mechanism which plays important roles in a variety of biological processes. Conjugation and deconjugation of SUMO is a cascade event requiring multiple protein-protein interactions between SUMO, SUMO E1-E3 enzymes and SUMO substrates. The nature of the SUMOylation network indicates a great potential for small chemical inhibitors to be used in the investigation and manipulation of this important pathway. FRET is an energy transfer phenomenon which occurs between two spectrum-overlapping fluorophores in close proximity. The efficiency of FRET is highly dependent on the distance between the donor and acceptor fluorophores, which makes FRET very powerful in the detection of molecular interactions. The goal of my research is to develop FRET-based techniques to study protein-protein interactions in the SUMOylation network and to identify bioactive chemical compounds which can inhibit the activity of this pathway. Using engineered fluorescent proteins to modify protein components involved in the SUMOylation pathway, I set up a FRET-based assay and developed mathematical algorithms to determine the binding affinity of non-covalent protein-protein interactions. I then developed this FRET-based assay into high-throughput screening platforms to look for small chemical inhibitors which can disrupt the interaction of SUMO1 and its E2 enzyme, Ubc9. I further used FRET-based strategy to set up a whole-pathway screening assay to identify SUMOylation/deSUMOylation inhibitors. Over 200,000 compounds were screened and the positive hits were characterized for their activities in vitro and in cells. Finally, I was able to use FRET to detect the formation of intermediate protein complex during SUMOylation reactions, and used the assay to map the targets of the SUMOylation inhibitors identified from the high-throughput screening. These FRET-based applications provide powerful tools for the study of the SUMOylation pathway and can be used in the investigation of protein-protein interactions in other biological pathways.

Background

Methods and results

Conclusion