Skip to main content
eScholarship
Open Access Publications from the University of California

Scholarly Works (5 results)

  • Thesis
  • Peer Reviewed
UC San Diego Electronic Theses and Dissertations (2017)

System reliability is becoming a significant concern as technology continues to shrink. This is because of increasing variation in circuit characteristics seen in nanometer-scaled microsystems. These variations in semiconductor manufacturing manifest as increasing fault rates in devices, rising soft errors, and timing errors caused by accelerated aging in circuits. This dissertation seeks software-based techniques to detect, recover, and prevent such errors in compute and memory components.

Software-based error detection and recovery techniques suffer from high-performance penalty to the overall system. This thesis presents methods that minimize the performance overhead of software-based error mitigation. In particular, this work proposes two techniques for effective software-based error detection in compute units: fingerprinting and cross-laneinstruction. Fingerprinting combines multiple error detection events into one event by hashing, and cross-lane instruction enables error checking via low latency register-level communication. Furthermore, to reduce the performance overhead of software-based recovery, this thesis explores Application-Specific Approximate Recovery (ASAR). ASAR trades-off output quality to reduce the performance penalty of software-based recovery.

Variation affects not just compute but memory components, as well. For memory units, we focus on methods for proactive error prevention. Given the diversity of memory components in use, we focus on emerging Heterogeneous Memory Architectures (HMAs). An HMA consists of multiple memory modules with different performance and reliability characteristics. These differences can be caused by different types of memory modules, different error correcting codes, and effects due to aging. This thesis focuses on methods to place and move data items among memory modules in an HMA system with the goal of reducing the likelihood of encountering an error. Specifically, this work describes two novel data placement techniques: age-aware and vulnerability-aware data placement. The age-aware technique monitors the accumulation of faults in different memory modules as they age, while the vulnerability-aware technique estimates the vulnerability of data in memory to soft errors. The results presented in this work enable practical use of software-based error mitigation solutions for current and future hardware.

    Cover page: Software Techniques to Enhance Reliability of Emerging Compute and Memory Units
    • Article
    • Peer Reviewed
    Proceedings of the Annual Meeting of the Cognitive Science Society, Volume 44 (2022)
    How does the brain represent different modes of information? Can we design a system that can automatically understand what the user is thinking? We can make progress towards answering such questions by studying brain recordings from devices such as functional magnetic resonance imaging (fMRI). The brain encoding problem aims to automatically generate fMRI brain representations given a stimulus. The brain decoding problem is the inverse problem of reconstructing the stimuli given the fMRI brain representation. Both the brain encoding and decoding problems have been studied in detail in the past two decades and the foremost attraction of studying these solutions is that they serve as additional tools for basic research in cognitive science and cognitive neuroscience. Recently, inspired by the effectiveness of deep learning models for natural language processing and computer vision, such models have been applied for neuroscience as well. In this tutorial, we plan to discuss different kinds of stimulus representations, and popular encoding and decoding architectures in detail. The tutorial will provide a working knowledge of the state of the art methods for encoding and decoding, a thorough understanding of the literature, and a better understanding of the benefits and limitations of encoding/decoding with deep learning.
      Cover page: Deep Learning for Brain Encoding and DecodingCreative Commons 'BY' version 4.0 license
      • Article
      • Peer Reviewed
      UC San Francisco Previously Published Works (2021)
      Since the arrival of DNA nanotechnology nearly 40 years ago, the field has progressed from its beginnings of envisioning rather simple DNA structures having a branched, multi-strand architecture into creating beautifully complex structures comprising hundreds or even thousands of unique strands, with the possibility to exactly control the positions down to the molecular level. While the earliest construction methodologies, such as simple Holliday junctions or tiles, could reasonably be designed on pen and paper in a short amount of time, the advent of complex techniques, such as DNA origami or DNA bricks, require software to reduce the time required and propensity for human error within the design process. Where available, readily accessible design software catalyzes our ability to bring techniques to researchers in diverse fields and it has helped to speed the penetration of methods, such as DNA origami, into a wide range of applications from biomedicine to photonics. Here, we review the historical and current state of CAD software to enable a variety of methods that are fundamental to using structural DNA technology. Beginning with the first tools for predicting sequence-based secondary structure of nucleotides, we trace the development and significance of different software packages to the current state-of-the-art, with a particular focus on programs that are open source.
        Cover page: The Art of Designing DNA Nanostructures with CAD Software
        • Article
        • Peer Reviewed
        UC San Francisco Previously Published Works (2024)
        STUDY DESIGN: Retrospective cohort study. OBJECTIVES: Identify optimal lumbar lordosis in adult deformity correction to achieve age-adjusted targets and sustained alignment. METHODS: Surgical adult spinal deformity patients reaching an age-adjusted ideal alignment at one year were identified. Multilinear regression analysis was used to identify the relationship between regional curvatures (LL and TK) that enabled achievement of a given global alignment (T1 pelvic angle, TPA) based on pelvic incidence (PI). RESULTS: 347 patients out of 1048 available reached their age-adjusted TPA within 5° (60-year-old, 72% women, body mass index 29 ± 6.2). They had a significant improvement in all sagittal parameters (except PI) from pre-operative baseline to 1 year following surgery (P < .001). Multilinear regression predicting L1-S1 based on TK, TPA, and PI demonstrated excellent results (R2 = .85). Simplification of the coefficients of prediction combined with a conversion to an age-based formula led to the following: LL = PI - 0.3TK - 0.5Age + 10. Internal validation of the formula led to a mean error of -.4°, and an absolute error of 5.0°. Internal validation on patients with an age-adjusted alignment revealed similar accuracy across the entire age-adjusted TPA spectrum (ranges of LL errors: ME = .2° to 1.7°, AE = 4.0° to 5.3°). CONCLUSION: This study provides a simple guideline to identify the amount of LL needed to reach a given alignment (i.e., age-adjusted target) based on PI and associated TK. Implementation of this predictive formula during pre-operative surgical planning may help to reduce unexpected sub-optimal post-operative alignment outcomes.
          Cover page: How Much Lumbar Lordosis does a Patient Need to Reach their Age-Adjusted Alignment Target? A Formulated Approach Predicting Successful Surgical Outcomes.
          • Article
          • Peer Reviewed
          UC Berkeley Previously Published Works (2015)
          Field-scale biostimulation and desorption tracer experiments conducted in a uranium (U) contaminated, shallow alluvial aquifer have provided insight into the coupling of microbiology, biogeochemistry, and hydrogeology that control U mobility in the subsurface. Initial experiments successfully tested the concept that Fe-reducing bacteria such as Geobacter sp. could enzymatically reduce soluble U(VI) to insoluble U(IV) during in situ electron donor amendment (Anderson et al., 2003; Williams et al., 2011). In parallel, in situ desorption tracer tests using bicarbonate amendment demonstrated rate-limited U(VI) desorption (Fox et al., 2012). These results and prior laboratory studies underscored the importance of enzymatic U(VI)-reduction and suggested the ability to combine desorption and bioreduction of U(VI). Here we report the results of a new field experiment in which bicarbonate-promoted uranium desorption and acetate amendment were combined and compared to an acetate amendment-only experiment in the same experimental plot. Results confirm that bicarbonate amendment to alluvial aquifer sediments desorbs U(VI) and increases the abundance of Ca-uranyl-carbonato complexes. At the same time, the rate of acetate-promoted enzymatic U(VI) reduction was greater in the presence of added bicarbonate in spite of the increased dominance of Ca-uranyl-carbonato aqueous complexes. A model-simulated peak rate of U(VI) reduction was ~3.8 times higher during acetate-bicarbonate treatment than under acetate-only conditions. Lack of consistent differences in microbial community structure between acetate-bicarbonate and acetate-only treatments suggest that a significantly higher rate of U(VI) reduction in the bicarbonate-impacted sediment may be due to a higher intrinsic rate of microbial reduction induced by elevated concentrations of the bicarbonate oxyanion. The findings indicate that bicarbonate amendment may be useful in improving the engineered bioremediation of uranium in aquifers.
            Cover page: Bicarbonate impact on U(VI) bioreduction in a shallow alluvial aquifer
            Top