http://www.rssboard.org/rss-specification 720 https://escholarship.org/uc/ucsd_cse_technicalreports/rss Recent eScholarship items from Technical Reports Mon, 31 Aug 2026 21:29:16 +0000 https://escholarship.org/uc/item/5969f87m <p>This report presents findings from a comprehensive survey of commercial fleet professionals involved in fleet safety (specifically managers, coaches and decision-makers) regarding the perceived effectiveness, practical implementation, and strategic value of different driver coaching approaches. Grounded in the broader objective of improving safety outcomes through evidence-based coaching strategies, the survey sought to explore both individual preferences and organizational practices.</p><p>The survey respondents consisted of two distinct groups: individuals who identify coaching as a key part of their primary job, and commercial dashcam decision-makers who do not directly engage in driver coaching. This composition allowed for insights from both those actively involved in coaching practices and those responsible for implementing or overseeing dashcam technology without hands-on coaching responsibilities.</p> https://escholarship.org/uc/item/5969f87m Tue, 11 Nov 2025 00:00:00 +0000 Weibel, Nadir Yarmand, Matin Biehle, Cole Mac, Elysia Lu, Huimeng Xiang, Vivian Strudler, Gabriella Hathiram, Pari https://escholarship.org/uc/item/1z82z8pg <p>This report critically evaluates coaching methodologies for enhancing fleet driver safety, engagement, and overall organizational performance. Drawing on empirical research across education, behavioral science, and fleet management, this analysis identifies key dimensions of effective commercial driver coaching, highlights significant limitations of current practices, and outlines strategic recommendations focused explicitly on differentiation and optimization. The findings emphasize the superior effectiveness of personalized, manager-led coaching methods, yet also show how manager-led coaching alone is challenging to implement at scale given the scarce availability of expert coaches. On the other hand, this report shows that in-cabin automated warning systems and self-coaching offer strengths in scalability and flexible learning. However, without integration with more personalized and interactive feedback, these approaches alone do not achieve the same long-term safety outcomes... https://escholarship.org/uc/item/1z82z8pg Mon, 30 Jun 2025 00:00:00 +0000 Weibel, Nadir Yarmand, Matin Biehle, Cole Mac, Elysia Lu, Huimeng Xiang, Vivian https://escholarship.org/uc/item/29d3b140 <p>In this paper review, I will explain the fundamentals of a “laser soliton microcomb heterogeneously integrated on silicon”, presented by Xiang et. al of the Bowers Lab at UCSB [1]. I will try to relate concepts taught in class, which will be <strong>bolded</strong>, while presenting new material at a level that can be understood by someone who only has understanding of basic optics as well as the theory, components, and operation of a traditional Fabry-Perot cavity laser.</p><p>In this paper Xiang et. al present the design, the underlying photonics principles used, their fabrication process and considerations, and finally performance measurements and verification of the final device.</p> https://escholarship.org/uc/item/29d3b140 Tue, 7 Jan 2025 00:00:00 +0000 Ding, Jerry https://escholarship.org/uc/item/610345mv <p><strong>Background:&nbsp;</strong>Breastfeeding benefits both the mother and infant and is a topic of attention in public health. After childbirth, untreated medical conditions or lack of support lead many mothers to discontinue breastfeeding. For instance, nipple damage and mastitis affect 80% and 20% of US mothers, respectively. Lactation consultants (LCs) help mothers with breastfeeding, providing in-person, remote, and hybrid lactation support. LCs guide, encourage, and find ways for mothers to have a better experience breastfeeding. Current telehealth services help mothers seek LCs for breastfeeding support, where images help them identify and address many issues. Due to the disproportional ratio of LCs and mothers in need, these professionals are often overloaded and burned out.</p><p><strong>Objective:&nbsp;</strong>This study aims to investigate the effectiveness of 5 distinct convolutional neural networks in detecting healthy lactating breasts and 6 breastfeeding-related... https://escholarship.org/uc/item/610345mv Tue, 19 Nov 2024 00:00:00 +0000 de Souza, Jessica Kumar Viswanath, Varun Echterhoff, Jessica Maria Chamberlain, Kristina Wang, Edward Jay https://escholarship.org/uc/item/1bm0k1w0 <p>Prompts to large language models (LLMs) have evolved beyond simple user questions. For LLMs to solve complex problems, today's practices are to include domain-specific instructions, illustration of tool usages, and long context such as textbook chapters in prompts. As such, many parts of prompts are repetitive across requests, and their attention computation results can be reused. However, today's LLM serving systems treat every request in isolation, missing the opportunity of computation reuse.</p><p>This paper proposes Preble, the first distributed LLM serving platform that targets and optimizes for prompt sharing. We perform a study on five popular LLM workloads. Based on our study results, we designed a distributed scheduling system that co-optimizes computation reuse and load balancing. Our evaluation of Preble on two to 8 GPUs with real workloads and request arrival patterns on two open-source LLM models shows that Preble outperforms the state of the art avg latency by... https://escholarship.org/uc/item/1bm0k1w0 Fri, 17 May 2024 00:00:00 +0000 Srivatsa, Vikranth He, Zijian Abhyankar, Reyna Li, Dongming Zhang, Yiying https://escholarship.org/uc/item/8gc2k3kw Snapchat is one of the most popular social media apps in the world. It is no surprise, then, that many political ads are run on the service each year. Snap Inc.'s political ads library is part of an effort by the company to increase transparency in their advertising practices. The data analyzed in this project spans 2019-2020, and consists of information on every political ad that was run on the service in that timeframe, including who the ad buyer was, how much the ad cost, what areas it targeted, etc. Geographic and monetary distribution of ads is analyzed and possible explanations given for anomalies. Missingness of the data was evaluated and Vermont was identified as an area with unusual spending. With&nbsp;α&nbsp;= 0.05 the null hypothesis was rejected (p&nbsp;= 0.02); the distribution of ad dollars to Vermont is not wholly random. https://escholarship.org/uc/item/8gc2k3kw Thu, 16 May 2024 00:00:00 +0000 Gorlla, Cyril https://escholarship.org/uc/item/6413d1xx The Snapchat ads dataset contains political ad data for ads on Snapchat, oneof the largest social media networks in the world. A key feature of the datasetis how much money an organization spends on a particular ad, found in the`Spend` column. It is reasonable to assume that this amount varies based oncertain factors, but can we use those factors to figure out how much is spenton an ad? We can explore this by predicting ad spending through machinelearning. After feature analysis and engineering, we arrive at a linearregression model with $R^2$ = .85 and perform a fairness evaluation of thealgorithm. https://escholarship.org/uc/item/6413d1xx Thu, 16 May 2024 00:00:00 +0000 Gorlla, Cyril https://escholarship.org/uc/item/9xv8x6n6 This report surveys recent techniques for reconstructing surfaces from points. We describe four main ideas in the graphics literature: signed distance estimation, Voronoi-based reconstruction, implicit surface fitting, and moving least squares surfaces. The main challenges include reconstruction without surface normals, robustness to noise, accuracy to sharp features, and provable reconstruction guarantees. We compare various techniques and discuss possible avenues for future work.<p>Pre-2018 CSE ID: CS2008-0922</p> https://escholarship.org/uc/item/9xv8x6n6 Tue, 14 Mar 2023 00:00:00 +0000 Chang, William Y https://escholarship.org/uc/item/9x77f916 NAND flash-based solid-state drives (SSDs) are becoming a staple in consumer electronics and high-performance computing. Limited resources and a lack of flexible prototyping platforms constrain NAND flash researchers to use software simulation of the chips, where simulation parameters are based on conservative values from datasheets, leading to missed research opportunities. A tool to explore low-level NAND flash behavior and performance would give researchers an easy way to validate and guide new research ideas and designs. We present the Ming II platform which gives researchers complete control over flash chips. It includes a custom board that connects to a platform FPGA system and provides an automated interface for acquiring fine-grain power, latency, and bit error measurements of flash operations. The Ming II software stack includes an open-source userspace library, Linux driver, and development environment that make it easy to develop new software that targets flash. Possible... https://escholarship.org/uc/item/9x77f916 Tue, 14 Mar 2023 00:00:00 +0000 Bunker, Trevor Wei, Michael Swanson, Steven https://escholarship.org/uc/item/9wx7m5z6 In this work we present CitiSense, a new kind of "citizen infrastructure" for the monitoring of pollution and environmental conditions that users are exposed to. By utilizing mobile phones and affordable, small sensors placed in the environment and carried by users, data about pollutants such as ozone and carbon monoxide is collected and used to provide real-time feedback to users and enable them to make healthy changes in their behavior. Results can be reported to a back-end server for further processing and learning, allowing other stakeholders to better understand how diseases such as asthma develop and to help coordinate efforts within a user's community to improve conditions. What differentiates CitiSense from previous projects of this sort is the design of a complete system that addresses issues of mobile power management, data security, privacy, inference with commodity sensors, and integration into a highly extensible and adaptive infrastructure comprising of Open Rich... https://escholarship.org/uc/item/9wx7m5z6 Tue, 14 Mar 2023 00:00:00 +0000 Nikzad, Nima Ziftci, Celal Zappi, Piero Quick, Nichole Aghera, Priti Verma, Nakul Demchak, Barry Patrick, Kevin Shacham, Hovav Rosing, Tajana Simunic Krueger, Ingolf Griswold, William G Dasgupta, Sanjoy https://escholarship.org/uc/item/9tq826n1 Defending users against fraudulent Websites (i.e., phishing) is a task that is reactive in practice. Blacklists, spam filters, and takedowns all depend on first finding new sites and verifying that they are fraudulent. In this paper we explore an alternative approach that uses a combination of computer-vision techniques to proactively identify likely phishing pages as they are rendered, interactive queries to validate such pages with brand holders, and a single keyboard-entry filter to minimize false positives. We have developed a prototype version of this approach within the Firefox browser and we provide a preliminary evaluation of both the underlying technology (the accuracy and performance of logo recognition in Web pages) as well as its effectiveness in controlled small-scale user studies. While no such approach is perfect, our results demonstrate that this technique offers a significant new capability for minimizing response time in combating a wide range of phishing scams.<p>Pre-2018... https://escholarship.org/uc/item/9tq826n1 Tue, 14 Mar 2023 00:00:00 +0000 Wang, Ge Liu, He Becerra, Sebastian Wang, Kai Belongie, Serge Shacham, Hovav Savage, Stefan https://escholarship.org/uc/item/9nz3j133 We present a new method for efficiently simulating the scattering of light within participating media. Using a theoretical reformulation of volumetric photon mapping, we develop a novel photon gathering technique for participating media. Traditional volumetric photon mapping samples the in-scattered radiance at numerous points along the length of a single ray by performing costly range queries within the photon map. Our technique replaces these multiple point-queries with a single beam-query, which explicitly gathers all photons along the length of an entire ray. These photons are used to estimate the accumulated in-scattered radiance arriving from a particular direction and need to be gathered only once per ray. Our method handles both fixed and adaptive kernels, is faster than regular volumetric photon mapping, and produces images with less noise.<p>Pre-2018 CSE ID: CS2008-0914</p> https://escholarship.org/uc/item/9nz3j133 Tue, 14 Mar 2023 00:00:00 +0000 Jarosz, Wojciech Zwicker, Matthias Jensen, Henrik Wann https://escholarship.org/uc/item/9jh7v86q We present a simple variant of the k-d tree which automatically adapts to intrinsic low dimensional structure in data without having to explicitly learn this structure.<p>Pre-2018 CSE ID: CS2007-0890</p> https://escholarship.org/uc/item/9jh7v86q Tue, 14 Mar 2023 00:00:00 +0000 Dasgupta, Sanjoy Freund, Yoav https://escholarship.org/uc/item/9dw8h2t7 In recent years, researchers have proposed systems for running trusted code on an untrusted operating system. Protection mechanisms deployed by such systems keep a malicious kernel from directly manipulating a trusted application’s state. Under such systems, the application and kernel are, conceptually, peers, and the system call API defines an RPC interface between them. We introduce Iago attacks, attacks that a malicious kernel can mount in this model. We show how a carefully chosen sequence of integer return values to Linux system calls can lead a supposedly protected process to act against its interests, and even to undertake arbitrary computation at the malicious kernel’s behest. Iago attacks are evidence that protecting applications from malicious kernels is more difficult than previously realized.<p>Pre-2018 CSE ID: CS2012-0984</p> https://escholarship.org/uc/item/9dw8h2t7 Tue, 14 Mar 2023 00:00:00 +0000 Checkoway, Steven Shacham, Hovav https://escholarship.org/uc/item/9dj789fj The growing sophistication of self-propagating worms and botnets presents a significant challenge for investigators to understand. While honeyfarms have emerged as a powerful tool for capturing and analyzing rapid malware, the size and complexity of large scale, high fidelity honeyfarms make them problematic to operate in a simultaneously safe and effective manner. This paper introduces a universe abstraction that guarantees isolation between multiple malware infestations in a single honeyfarm while maximizing the realism of the honeyfarm as observed by a propagating worm. We demonstrate that each malware strain can be completely isolated without distorting malware spreading behavior, and that this can in fact increase the scalability of honeyfarms.<p>Pre-2018 CSE ID: CS2007-0902</p> https://escholarship.org/uc/item/9dj789fj Tue, 14 Mar 2023 00:00:00 +0000 Chen, Jay McCullough, John Snoeren, Alex C https://escholarship.org/uc/item/9dh0r7hf We introduce an open labeling platform for Computer Vision researchers based on Captchas, creating as a byproduct labeled image data sets while supporting web security. For the two different tasks of annotation and detection, we provide a security analysis and explore usability issues. We present the interfaces used by researchers, website administrators and users and experimental results obtained using the platform. Finally, we discuss system sustainability issues in the context of a broader “ecosystem” for the platform.<p>Pre-2018 CSE ID: CS2009-0942</p> https://escholarship.org/uc/item/9dh0r7hf Tue, 14 Mar 2023 00:00:00 +0000 Faymonville, Peter Wang, Kai Miller, John Belongie, Serge https://escholarship.org/uc/item/9bx1c3pk Government, medical, financial, and web-based services increasingly depend on the ability to rapidly sift through huge, evolving data sets. These data-intensive applications perform complex multi-step computations over successive generations of data inflows (e.g., weekly web crawls, nightly telescope dumps, or hourly surveillance videos). Because of the data volumes involved, applications must avoid reprocessing old data when new data arrives and instead process incrementally. Unlike in stream-based systems, incoming data does not have to be processed immediately, permitting work to be amortized via bulk processing. Such bulk-incremental processing represents an emerging class of applications whose needs are not fully met by current systems. This paper presents a generalized architecture for bulk-incremental processing systems (BIPS), simplifying the creation of such programs. In contrast with incremental view maintenance in data warehousing, BIPS provides flexible low-level primitives... https://escholarship.org/uc/item/9bx1c3pk Tue, 14 Mar 2023 00:00:00 +0000 Logothetis, Dionysios Olston, Christofer Reed, Benjamin Webb, Kevin Yocum, Kenneth https://escholarship.org/uc/item/9bm5k56q Solid State Disks (SSDs) based on flash and other non-volatile memory technologies reduce storage latencies from 10s of milliseconds to 10s or 100s of microseconds, transforming previously inconsequential storage overheads into performance bottlenecks. This problem is especially acute in storage area network (SAN) environments where complex hardware and software layers (distributed file systems, block severs, network stacks, etc.) lie between applications and remote data. These layers can add hundreds of microseconds to requests, obscuring the performance of both flash memory and faster, emerging non-volatile memory technologies. This work examines the performance of QuickSAN, a SAN prototype that eliminates most software overheads and significantly reduces hardware overheads in SANs, as the number of nodes in the SAN increases. QuickSAN integrates a network adapter directly into SSDs, so the SSDs can communicate directly with one another to service storage accesses as quickly... https://escholarship.org/uc/item/9bm5k56q Tue, 14 Mar 2023 00:00:00 +0000 Caulfield, Adrian M Swanson, Steven https://escholarship.org/uc/item/9622h33g To evade blacklisting, the vast majority of spam email is sent from exploited MTAs (i.e., botnets) and with forged "From" addresses. In response, the anti-spam community has developed a number of domain-based authentication systems -- such as SPF and DKIM -- to validate the binding between individual domain names and legitimate mail sources for those domains. In this paper, we explore an alternative solution in which the mail recipient requests a real-time affirmation for each e-mail from the declared sender's MX of record. The "Occam" protocol is trivial to implement, offers authenticating power equivalent to SPF and DKIM and, most importantly, forces spammers to deploy and expose blacklistable servers for each domain they use during a campaign. We discuss the details of the protocol, compare its strengths and weaknesses with existing solutions and describe a prototype implementation in Sendmail.<p>Pre-2018 CSE ID: CS2007-0893</p> https://escholarship.org/uc/item/9622h33g Tue, 14 Mar 2023 00:00:00 +0000 Fleizach, Chris Voelker, Geoffrey M Savage, Stefan https://escholarship.org/uc/item/95x5m073 In datacenter applications, predictability in service time and controlled latency, especially tail latency, is essential for building performant applications. This is especially true for applications or services build by accessing data across thousands of servers to generate a user response. Current practice has been to run such services at low utilization to rein in latency outliers, which decreases efficiency and limits the number of service invocations developers can issue while still meeting tight latency budgets. In this paper, we analyze the three datacenter applications, Memcached, OpenFlow, and web search, to measure the effect on tail latency of 1) kernel socket handling, NIC interaction, and the network stack, 2) application locks contested in the kernel, and 3) application-layer queuing due to requests being stalled behind straggler threads. We propose a novel approach of reducing the above sources of latency by relying on support from the NIC hardware, and we find... https://escholarship.org/uc/item/95x5m073 Tue, 14 Mar 2023 00:00:00 +0000 Kapoor, Rishi Porter, George Tewari, Malveeka Voelker, Geoffrey M Vahdat, Amin https://escholarship.org/uc/item/9514v2jb We present a novel architecture for adaptive weight calculation (AWC) that uses the QR decomposition based recursive least squares (RLS) algorithm. Our AWC core achieves a throughput of 0.20M updates per second for a 4 x 4 matrix on a Xilinx Virtex4 SX FPGA. We show that our core is significantly faster than other published FPGA implementations and it requires fewer resources. This is largely a consequence of careful error analysis that allows us to take advantage of a fixed point data representation with little degradation in the final results. Finally, our proposed architecture scales well for larger size matrices.<p>Pre-2018 CSE ID: CS2009-0937</p> https://escholarship.org/uc/item/9514v2jb Tue, 14 Mar 2023 00:00:00 +0000 Irturk, Ali Mirzaei, Shahnam Kastner, Ryan https://escholarship.org/uc/item/9490c7h1 The processor industry is at an inflection point. In the past, performance was the driving force behind the processor industry. But in the coming manycore era, improving programmability and reliability of the system will be at least as important as improving raw performance. To meet this vision, this thesis presents a processor feature that assists programmers in understanding software failures. <p> Reproducing software failures is a significant challenge. The problem is severe especially for multi-threaded programs because the causes of failure can be non-deterministic in nature. The proposed processor feature continuously logs a program.s execution while sacrificing very little performance ( 1%). If the program crashes, the developer can use the log to debug the failure by deterministically replaying every single instruction executed as part of the failed program.s execution. Two key mechanisms enable this deterministic replay feature. One is BugNet, a checkpointing technique,... https://escholarship.org/uc/item/9490c7h1 Tue, 14 Mar 2023 00:00:00 +0000 Narayanasamy, Satish https://escholarship.org/uc/item/9357z8vg Requirements relationships express conceptual dependencies, constraints and associations among the requirements of a software system, such as dependencies and hint-relations. For stakeholders of a system, it is important and beneficial to identify requirements relationships for system design, maintenance and comprehension tasks. In this paper, we build on existing research and use features, realization of functional requirements in software, to automatically retrieve requirements relationships from existing test cases. We evaluate our approach on a chat system, Apache Pool, and Apache Commons CLI. We obtain precision/recall levels as good as or better than currently existing object-tracing and scenario-analysis based approaches when tested on the same case studies. Furthermore, our approach is resistant to scenario selection, and works for all types of systems with a profiler available, unlike existing techniques.<p>Pre-2018 CSE ID: CS2013-1001</p> https://escholarship.org/uc/item/9357z8vg Tue, 14 Mar 2023 00:00:00 +0000 Ziftci, Celal Krueger, Ingolf https://escholarship.org/uc/item/9278h2ww Emerging fast, persistent memories will enable systems that combine conventional DRAM with large amounts of non-volatile main memory (NVMM) and provide huge increases in storage performance. Fully realizing this potential requires fundamental changes in how system software manages, protects, and provides access to data that resides in NVMM. We address these needs by describing a NVMM-optimized file system called NOVA that is both fast and resilient in the face of corruption due to media errors and software bugs. We identify and propose solutions for the unique challenges in hardening an NVMM file system, adapt state-of-the-art reliability techniques to an NVMM file system, and quantify the performance and storage overheads of these techniques. We find that NOVA's reliability features increase file system size system size by 14.9% and reduce application-level performance by between 2% and 38%.<p>Pre-2018 CSE ID: CS2017-1018</p> https://escholarship.org/uc/item/9278h2ww Tue, 14 Mar 2023 00:00:00 +0000 Xu, Jian Zhang, Lu Memaripour, Amirsaman Gangadharaiah, Akshatha Borase, Amit Brito Da Silva, Tamires Rudoff, Andy Swanson, Steven https://escholarship.org/uc/item/9130p3t5 Accuracy is one of the most important performance metrics in clock synchronization. While state-of-the-art synchronization protocols achieve μsec-order average accuracy, they usually do not focus on the worst case accuracy and do not have any deterministic guarantees. This lack of accuracy guarantee makes it hard for sensor networks to be incorporated into larger systems that require more reliability than e.g., typical environmental monitoring applications do. In this paper, we present a clock synchronization algorithm with deterministic accuracy guarantee. A key observation is that the variability of oscillation frequency is much smaller in a single crystal than between different crystals. Our algorithm leverages this to achieve much tighter accuracy guarantee compared to the interval-based synchronization methods mostly proposed in the literature of distributed systems. We designed an algorithm to solve a geometric problem involving tangents to convex polygons, and implemented... https://escholarship.org/uc/item/9130p3t5 Tue, 14 Mar 2023 00:00:00 +0000 Sugihara, Ryo Gupta, Rajesh https://escholarship.org/uc/item/8z62z06f Service Level Objectives (SLOs) describe the desired behavior of a service. They present agreed-upon target behaviors (e.g. capacity, performance, availability, cost) that the clients of a service can expect, and that the service provider can manage against. Both clients and service providers need a way to monitor how a service is doing against its SLOs, and display the result of that monitoring in an intuitive fashion. This paper describes a tool that provides interactive visual exploration for services; makes it easy to navigate around the multi-dimensional space of services, their instances, users, and metrics; and is driven by the machine-readable description of the SLO itself, so that changes can be propagated automatically. The contributions of the paper include a description of the requirements and implementation of our system; the techniques used to display aspects of --- and navigate around --- the multi-dimensional space; and how the SLOs are used to automate the display... https://escholarship.org/uc/item/8z62z06f Tue, 14 Mar 2023 00:00:00 +0000 Kim, Han Suk Verma, Sunita Wilkes, John https://escholarship.org/uc/item/8x74g752 Solid State Disks (SSDs) based on flash and other non-volatile memory technologies reduce storage latencies from 10s of milliseconds to 10s or 100s of microseconds, transforming previously inconsequential storage overheads into performance bottlenecks. This problem is especially acute in storage area network (SAN) environments where complex hardware and software layers (distributed file systems, block severs, network stacks, etc.) lie between applications and remote data. These layers can add hundreds of microseconds to requests, obscuring the performance of both flash memory and faster, emerging non-volatile memory technologies. We describe QuickSAN, a SAN prototype that eliminates most software overheads and significantly reduces hardware overheads in SANs. QuickSAN integrates a network adapter directly into SSDs, so the SSDs can communicate directly with one another to service storage accesses as quickly as possible. QuickSAN can also give applications direct access to both local... https://escholarship.org/uc/item/8x74g752 Tue, 14 Mar 2023 00:00:00 +0000 Caulfield, Adrian M Swanson, Steven https://escholarship.org/uc/item/8ws91483 Traditionally, capacity on the Internet has been allocated between competing flows through a distributed fairness calculation implemented by end-to-end congestion control protocols like TCP. Increasingly, however, network operators are deploying fair queuing and other forms of router-based enforcement mechanisms to prevent greedy or misbehaving end points from consuming more than their fair share of the network’s capacity. In environments where fairness is enforced by the network itself, it seems worthwhile to reconsider the role of the congestion control protocol. In particular, we ask if it might be both safe and sensible in the long term for self-interested senders to send at rates that exceed the capacity of the network. Through simulation, we identify and quantify the source of inefficiency in this regime, which we term zombie packets. Surprisingly, we show that such aggressive mechanisms are not only tenable in a wide variety of network structures, but, combined with effective... https://escholarship.org/uc/item/8ws91483 Tue, 14 Mar 2023 00:00:00 +0000 Raghavan, Barath McCollough, John Snoeren, Alex C https://escholarship.org/uc/item/8wn4m97p In this paper we introduce Echidna, a hybrid schematic/ text-based language for describing PCB circuit schematics. Echidna allows designers to use high-level programming con- structs to describe schematics, supports modular, reusable design components with well-defined interfaces, and provides for complex parameterization of those modules. Echidna deeply integrates a high-level programming language into a schematic-based design flow. The designer can describe schematics in code, as a schematic, or as a seamless combination of the two. We demonstrate its usefulness with several case studies.<p>Pre-2018 CSE ID: CS2016-1017</p> https://escholarship.org/uc/item/8wn4m97p Tue, 14 Mar 2023 00:00:00 +0000 Wang, Shengye Merrill, Devon Taylor, Chris Swanson, Steven https://escholarship.org/uc/item/8t4853vm We observe that Monte Carlo (SPICE) simulation provides the most accurate and trustable statistical timing analysis, while the existing SSTA method has completely ignored the effect of input statistics on chip timing performance, and provides either accurate estimate nor pessimistic bound of the actual chip timing performance statistics. We propose signal probability (i.e., the logic one occurrence probability for a signal) based statistical timing analysis for improved accuracy and reduced pessimism over the existing SSTA methods, and improved efficiency over Monte Carlo (SPICE) simulation. Our experimental results show that our proposed SPSTA computes mean (standard deviation) of signal arrival times within 6.2% (18.6%), while SSTA computes mean (standard deviation) of signal arrival times within 13.40% (64.3%) of Monte Carlo simulation results; SPSTA also provides signal probaiblity estimation within 14.28% of Monte Carlo simulation results for the ISCAS'89 benchmark circuits. <p>Pre-2018... https://escholarship.org/uc/item/8t4853vm Tue, 14 Mar 2023 00:00:00 +0000 Liu, Bao https://escholarship.org/uc/item/8m7737dh Variation in performance and power across manufactured parts and their operating conditions is an accepted reality in aggressive CMOS processes. This paper considers challenges and opportunities in identifying this variation and methods to combat or even use these variations for improved computing systems. We introduce the notion of instruction-level vulnerability (ILV) and concurrent instruction reuse (CIR) to expose variation and its effects to the software stack for use in architectural and runtime optimizations. Going further up on the hardware-software stack, we also introduce the notion of tasklevel vulnerability (TLV) as metadata to characterize dynamic variations. In fact, TLV is a vertical abstraction that reflects manifestation of circuit-level hardware variability in specific software context for parallel execution model.<p>Pre-2018 CSE ID: CS2014-1004</p> https://escholarship.org/uc/item/8m7737dh Tue, 14 Mar 2023 00:00:00 +0000 Rahimi, Abbas Benini, Luca Gupta, Rajesh https://escholarship.org/uc/item/8j85g8jn Recent proposals have employed optical circuit switching (OCS) to reduce the cost of data center networks. However, the relatively slow switching times (10-100 ms) assumed by these approaches, and the accompanying latencies of their control planes, has limited its use to only the largest data center networks with highly aggregated and constrained workloads. As faster switch technologies become available, designing a control plane capable of supporting them becomes a key challenge. In this paper, we design and implement an OCS prototype capable of switching in 11.5 microseconds, and we use this prototype to expose a set of challenges that arise when supporting switching at microsecond time scales. In response, we propose a microsecond-latency control plane based on a circuit scheduling approach we call Traffic Matrix Scheduling (TMS) that proactively communicates circuit assignments to communicating entities so that circuit bandwidth can be used efficiently.<p>Pre-2018 CSE ID:... https://escholarship.org/uc/item/8j85g8jn Tue, 14 Mar 2023 00:00:00 +0000 Porter, George Strong, Richard Farrington, Nathan Forencich, Alex Sun, Pang-Chen Rosing, Tajana Simunic Fainman, Yeshaiahu Papen, George Vahdat, Amin https://escholarship.org/uc/item/8h1713t5 Data centers lower costs by sharing the physical infrastructure among multiple services. However, the data center network should also ideally provide bandwidth guarantees to each service in a tunable manner while maintaining high utilization. We describe {\em NetShare}, a new statistical multiplexing mechanism for Data Center networks that does this without requiring changes to existing routers. NetShare allows the bisection bandwidth of the network to be allocated across services based on simple weights specified by a manager. Bandwidth unused by a service is shared proportionately by other services. More precisely, NetShare provides weighted hierarchical max-min fair sharing, a generalization of hierarchical fair queuing of individual {\em links}. We present three mechanisms to implement NetShare including one that leverages TCP flows and requires no changes to routers or servers. We show experiments using multiple Hadoop instances and a network of Fulcrum switches and... https://escholarship.org/uc/item/8h1713t5 Tue, 14 Mar 2023 00:00:00 +0000 Lam, Terry Radhakrishnan, Sivasankar Vahdat, Amin Varghese, George https://escholarship.org/uc/item/8fh2z68w Sensor networks have been used in various types of applications and still have a huge potential. As an application logic becomes more complex, growing number of people realize that programming sensor networks is difficult. Part of the difficulty comes from their distributed nature, but the large fraction is from other harsh situations they face such as unreliable communication, faulty nodes, and extremely constrained resources. Researchers have proposed different programming models to alleviate the difficulty, sharing an ultimate goal of making programming easy while making full use of given resources. In this paper, we first explore the requirements for programming models for sensor networks. Then we present a taxonomy of the programming models, classified according to the level of abstractions they provide. We evaluate each work in terms of the requirements and discuss the strategies how the programming models strive to satisfy these requirements. Further, we discuss some... https://escholarship.org/uc/item/8fh2z68w Tue, 14 Mar 2023 00:00:00 +0000 Sugihara, Ryo Gupta, Rajesh https://escholarship.org/uc/item/8dp743qb As paper-based communication is being supplanted by digital channels like E-mail, Facebook, and Flickr, so is handwritten and sketch-based expression. We posit that digital pen-and-paper technologies can bring the unique affordances of paper-based communication to social media, providing new channels of expression to those who favor handwriting and sketching. We present UbiSketch, a ubiquitous real-time sketch-based communication tool that supports social communication using natural paper-based interaction. UbiSketch allows people to create drawings, doodles, and text notes with digital pens and paper, and publish them via their mobile phones to online social networks or e-mail. In a 4-week user study, this novel modality successfully brought sketching into the public realm, exhibiting characteristics of digital social media such as rapid authoring and dissemination, while retaining sketching's unique qualities of visual communication. Friend responses to sketches were more... https://escholarship.org/uc/item/8dp743qb Tue, 14 Mar 2023 00:00:00 +0000 Cowan, Lisa Weibel, Nadir Pina, Laura R Hollan, James D Griswold, William G https://escholarship.org/uc/item/8bv2j2hp Recent increase in energy prices has led researchers to find better ways for capacity provisioning in data centers to reduce energy wastage due to the variation in workload. This paper explores the opportunity for cost saving and proposes a novel approach for capacity provisioning under bounded latency requirements for the workload. We investigate how many servers to be kept active and how much workload to be delayed for energy saving while meeting every deadline. We present an offline LP formulation for capacity provisioning by dynamic deferral and give two online algorithms to determine the capacity of the data center and the assignment of workload to servers dynamically. We prove the feasibility of the online algorithms and show that their worst case performance are bounded by a constant factor with respect to the offline formulation. We validate our algorithms on synthetic workload generated from two real HTTP traces and show that they actually perform much better in practice... https://escholarship.org/uc/item/8bv2j2hp Tue, 14 Mar 2023 00:00:00 +0000 Adnan, Muhammad Abdullah Ma, Yan Sugihara, Ryo Gupta, Rajesh https://escholarship.org/uc/item/899803r2 In this paper we consider the problem of detecting whether a compromised router is maliciously manipulating its stream of packets. In particular, we are concerned with a simple yet effective attack in which a router selectively drops packets destined for some victim. Unfortunately, it is quite challenging to attribute a missing packet to a malicious action because normal network congestion can produce the same effect. Modern networks routinely drop packets when the load temporarily exceeds a router's buffering capacity. Previous detection protocols have tried to address this problem using a user-defined threshold: too many dropped packets implies malicious intent. However this heuristic is fundamentally unsound; setting this threshold is, at best, an art and will necessarily create unnecessary false positives or mask highly-focused attacks. We have designed, developed and implemented a compromised router detection protocol that dynamically infers, based on measured traffic... https://escholarship.org/uc/item/899803r2 Tue, 14 Mar 2023 00:00:00 +0000 Mizrak, Alper T Marzullo, Keith Savage, Stefan https://escholarship.org/uc/item/8969r8tc Manifold learning is a popular recent approach to nonlinear dimensionality reduction. Algorithms for this task are based on the idea that the dimensionality of many data sets is only artificially high; though each data point consists of perhaps thousands of features, it may be described as a function of only a few underlying parameters. That is, the data points are actually samples from a low-dimensional manifold that is embedded in a high-dimensional space. Manifold learning algorithms attempt to uncover these parameters in order to find a low-dimensional representation of the data. In this paper, we discuss the motivation, background, and algorithms proposed for manifold learning. Isomap, Locally Linear Embedding, Laplacian Eigenmaps, Semidefinite Embedding, and a host of variants of these algorithms are examined.<p>Pre-2018 CSE ID: CS2008-0923</p> https://escholarship.org/uc/item/8969r8tc Tue, 14 Mar 2023 00:00:00 +0000 Cayton, Lawrence https://escholarship.org/uc/item/86c1m73c As the web evolves, it is becoming easier to form communities based on shared interests, and to create, publish, and query data on a wide variety of topics. In order to fully deliver on the promise of free data exchange, any community-supporting infrastructure needs to enforce the key requirement to preserve privacy of the association of content providers with potential sensitive published information. This privacy preserving publishing requirement prevents censorship, harassment, or discrimination of users by third parties. We propose a novel privacy-preserving distributed infrastructure in which data resides only with the publishers owning it. The infrastructure disseminates user queries to publishers, who answer them at their own discretion. The infrastructure enforces a publisher k-anonymity guarantee, which prevents leakage of information about which publishers are capable of answering a certain query. Given the virtual nature of the global data collection, we study the... https://escholarship.org/uc/item/86c1m73c Tue, 14 Mar 2023 00:00:00 +0000 Curtmola, Emiran Deutsch, Alin Ramakrishnan, K. K. Srivastava, Divesh https://escholarship.org/uc/item/863626c7 Wireless sensor nodes are increasingly being tasked with computation and communication intensive functions while still subject to constraints related to energy availability. On these embedded platforms, once all low power design techniques have been explored, duty-cycling the various subsystems remains the primary option to meet the energy and power constraints. This requires the ability to provide spurts of high MIPS and high bandwidth connections. However, due to the large overheads associated with duty-cycling the computation and communication subsystems, existing high performance sensor platforms are not efficient in supporting such an option. In this paper, we present the design and optimizations taken in a gateway node that provides access to a Wi-Fi radio in an on-demand basis. We discuss our strategies to reduce duty-cycling related costs by partitioning the system and by reducing the amount of time required to activate or deactivate the high-powered components. We compare... https://escholarship.org/uc/item/863626c7 Tue, 14 Mar 2023 00:00:00 +0000 Jin, Zhong-Yi Schurgers, Curt Gupta, Rajesh https://escholarship.org/uc/item/82z8j1w8 Double patterning lithography (DPL) is in current production for memory products, and is widely viewed as inevitable for logic products at the 32nm node. DPL decomposes and prints the shapes of a critical-layer layout in two exposures. In traditional single-exposure lithography, adjacent identical layout features will have identical mean critical dimension (CD), and spatially correlated CD variations. However, with DPL, adjacent features can have distinct mean CDs, and uncorrelated CD variations. This introduces a new set of 'bimodal' challenges for timing analysis and optimization. We assess the potential impact of DPL on timing analysis error and guardbanding, and find that the traditional `unimodal' characterization and analysis framework may not be viable for DPL. For example, using 45nm models, we find that different DPL mask layout solutions can cause 50ps skew in clock distribution that is unseen by traditional analyses. Different mask layouts can also result in 20\%... https://escholarship.org/uc/item/82z8j1w8 Tue, 14 Mar 2023 00:00:00 +0000 Jeong, Kwangok Kahng, Andrew B https://escholarship.org/uc/item/82f168vg Wikipedia infoboxes is an example of a seemingly structured, yet extraordinarily heterogenous dataset, where any given record has only a tiny fraction of all possible fields. Such data cannot be queried using traditional means without a massive a priori integration effort, since even for a simple request the result values span many record types and fields. On the other hand, the solutions based on keyword search are too imprecise to exactly capture the user's intent. To address these limitations, we propose WikiAnalytics system that utilizes a novel search paradigm in order to derive tables of precise and complete results from Wikipedia infobox records. The user starts with a keyword search query that finds a superset of the result records, and then browses clusters of records deciding which are and are not relevant. WikiAnalytics uses three categories of clustering features based on record types, fields, and values that matched the query keywords, respectively. Since the system... https://escholarship.org/uc/item/82f168vg Tue, 14 Mar 2023 00:00:00 +0000 Balmin, Andrey Curtmola, Emiran https://escholarship.org/uc/item/7zd7j8hp "Big Data" computing increasingly utilizes the MapReduce programming model for scalable processing of large data collections. Many MapReduce jobs are I/O-bound, and so minimizing the number of I/O operations is critical to improving their performance. In this work, we present ThemisMR, a MapReduce implementation that reads and writes data records to disk exactly twice, which is the minimum amount possible for data sets that cannot fit in memory. In order to minimize I/O, ThemisMR makes fundamentally different design decisions from previous MapReduce implementations. ThemisMR performs a wide variety of MapReduce jobs – including click log analysis, DNA read sequence alignment, and PageRank – at nearly the speed of TritonSort’s record-setting sort performance.<p>Pre-2018 CSE ID: CS2012-0983</p> https://escholarship.org/uc/item/7zd7j8hp Tue, 14 Mar 2023 00:00:00 +0000 Rasmussen, Alexander Conley, Michael Kapoor, Rishi Lam, Vinh The Porter, George Vahdat, Amin https://escholarship.org/uc/item/7w80w06s The problem of using pictures of objects captured under ideal imaging conditions (here referred to as in vitro) to recognize objects in natural environments (in situ) is an emerging area of interest in computer vision and pattern recognition. Examples of tasks in this vein include assistive vision systems for the blind and object recognition for mobile robots; the proliferation of image databases on the web is bound to lead to more examples in the near future. Despite its importance, there is still a need for a freely available database to facilitate study of this kind of training/testing dichotomy. In this work one of our contributions is a new multimedia database of 120 grocery products, GroZi-120. For every product, two different recordings are available: in vitro images extracted from online grocery websites, and in situ images extracted from camcorder video collected inside a grocery store. As an additional contribution, we present the results of applying three commonly used... https://escholarship.org/uc/item/7w80w06s Tue, 14 Mar 2023 00:00:00 +0000 Merler, Michele Galleguillos, Carolina Belongie, Serge https://escholarship.org/uc/item/7s6184dq We propose a hybrid approach to sanitizing SSDs that combines speed of cryptographic key destruction with the verifiability of explicitly erasing the storage media. The technique, called Scramble and Finally Erase (SAFE), stores encrypted data in the drive and uses a two step process for sanitization. First, it destroys the key. Then, SAFE erases every physical page in the SSD. After this step, verification is a simple matter of dismantling the drive and verifying that the flash chips are actually erased.<p>Pre-2018 CSE ID: CS2011-0963</p> https://escholarship.org/uc/item/7s6184dq Tue, 14 Mar 2023 00:00:00 +0000 Swanson, Steven Wei, Michael https://escholarship.org/uc/item/7nz0178r The transistor density continues to increase exponentially, but the power dissipation per transistor improves only slightly with each generation of Moore’s law. Given the constant chip-level power budgets, this exponentially decreases the fraction of the transistors that can be active simultaneously with each technology generation. Hence, while the area budget continues to increase exponentially, the power budget has become a first-order design constraint in current processors. In this regime, utilizing transistors to design specialized cores that optimize energy-per-computation becomes an effective approach to improve the system performance. To trade transistors for energy efficiency in a scalable manner, we propose quasi application-specific integrated circuits, or QASICs, specialized processors capable of executing multiple general- purpose applications while providing an order-of-magnitude more energy efficiency than a general-purpose processor. The QASIC design flow is based... https://escholarship.org/uc/item/7nz0178r Tue, 14 Mar 2023 00:00:00 +0000 Venkatesh, Ganesh Sampson, Jack Goulding, Nathan Swanson, Steven Taylor, Michael https://escholarship.org/uc/item/7jn4f274 We present a simple, agnostic active learning algorithm that works for any hypothesis class of bounded VC dimension, and any data distribution. Our algorithm extends a scheme of Cohn, Atlas, and Ladner to the agnostic setting, by (1) reformulating it using a reduction to supervised learning and (2) showing how to apply generalization bounds even for the non-i.i.d. samples that result from selective sampling. We provide a general characterization of the label complexity of our algorithm. This quantity is never more than the usual PAC sample complexity of supervised learning, and is exponentially smaller for some hypothesis classes and distributions. We also demonstrate improvements experimentally.<p>Pre-2018 CSE ID: CS2007-0898</p> https://escholarship.org/uc/item/7jn4f274 Tue, 14 Mar 2023 00:00:00 +0000 Dasgupta, Sanjoy Hsu, Daniel Monteleoni, Claire https://escholarship.org/uc/item/7j213325 Performance comparisons are ubiquitous in computer science. The proceedings of most conferences are filled with bar charts comparing the performance of some computer system to another. For example, computer architects compare the performance of processors, and compiler writers compare the performance of generated code. It is difficult to prove that one computer system is always faster than another for all possible workloads, so these performance comparisons are used as predictors: performance is compared on several representative workloads, and the results are used to argue that one computer system is generally faster than another. Unfortunately, there are many scenarios where it is difficult to make a fair performance comparison. This dissertation focuses on two such scenarios. The first scenario involves simulations in computer architecture. Computer architects typically evaluate new processor designs through slow cycle-level simulation. Because of the poor performance of cycle-level... https://escholarship.org/uc/item/7j213325 Tue, 14 Mar 2023 00:00:00 +0000 Lau, Jeremy https://escholarship.org/uc/item/7hf8346g Despite flash memory’s promise, it suffers from many idiosyncrasies such as limited durability, data integrity problems, and asymmetry in operation granularity. As architects, we aim to find ways to overcome these idiosyncrasies while exploiting flash memory’s useful characteristics. To be successful, we must understand the trade-offs between the performance, cost (in both power and dollars), and reliability of flash memory. In addition, we must understand how different usage patterns affect these characteristics. Flash manufacturers provide only vague guidelines about these metrics, and this lack of detail makes it difficult to design systems that fully exploit flash memory’s capabilities. We have empirically characterized flash memory technology from five manufacturers by directly measuring the performance, power, and reliability. We demonstrate that performance varies significantly between vendors, devices, and from the data-sheet. We also demonstrate and quantify some unexpected device characteristics... https://escholarship.org/uc/item/7hf8346g Tue, 14 Mar 2023 00:00:00 +0000 Grupp, Laura Caulfield, Adrian Coburn, Joel Swanson, Steven Yaakobi, Eitan Siegel, Paul https://escholarship.org/uc/item/7ct9w971 Large-scale network services can consist of tens of thousands of machines running thousands of unique software configurations spread across hundreds of physical networks. Testing such services for complex performance problems, configuration errors, and fault tolerance remains a difficult problem. Existing testing techniques, for example through simulation or running smaller instances of a service, have limitations in predicting overall service behavior. Although technically and economically infeasible at this time, testing should ideally be performed at the same scale and with the same configuration as the deployed service. We present DieCast, an approach to scaling network services; we multiplex all of the nodes in a given service configuration as virtual machines (VM) spread across a much smaller number of physical machines in a test harness. CPU, network, and disk are then accurately scaled to provide the illusion that each VM matches a machine from the original service in... https://escholarship.org/uc/item/7ct9w971 Tue, 14 Mar 2023 00:00:00 +0000 Gupta, Diwaker Vishwanath, Kashi Vahdat, Amin https://escholarship.org/uc/item/7c87654g Modern organizations face increasingly complex information management requirements. A combination of commercial needs, legal liability and regulatory imperatives has created a patch-work of mandated policies. Among these, personally identifying customer records must be carefully access-controlled, sensitive files must be encrypted on mobile computers to guard against physical theft and intellectual property must be protected from both exposure and ``poisoning.'' However, enforcing such policies can be quite difficult in practice since users routinely share data over networks and derive new files from these inputs -- incidentally laundering any policy restrictions. In this paper, we describe a VMM system called Neon that transparently labels derived data using byte-level ``tints'' and tracks these labels end-to-end across commodity applications, operating systems and networks. We demonstrate that this mechanism allows the enforcement of a variety of data management policies,... https://escholarship.org/uc/item/7c87654g Tue, 14 Mar 2023 00:00:00 +0000 Zhang, Qing McCullough, John Ma, Justin Schear, Nabil Vrable, Michael Vahdat, Amin Snoeren, Alex C Voelker, Geoffrey M Savage, Stefan https://escholarship.org/uc/item/77b8c38f Energy consumption of IT equipment is significant in enterprises. While low power sleep modes have existed for some time, their use in wall-powered desktop PCs is not prevalent due to the requirement of maintaining network presence and availability for network facing applications. In this paper, we describe the architecture and implementation of SleepServer, a system that enables hosts to transition to low-power sleep states while keeping them reachable at the highest layers of the network and software stack. SleepServer works across heterogeneous computing environments and hierarchical multi-domain networks. SleepServer requires no changes to the underlying networking infrastructure or any additional hardware on the end hosts, requiring only simple software agents to be installed. We detail results from our experience in deploying SleepServer in a medium scale enterprise with a sample set of thirty machines that were instrumented to provide accurate real-time measurements of... https://escholarship.org/uc/item/77b8c38f Tue, 14 Mar 2023 00:00:00 +0000 Agarwal, Yuvraj Savage, Stefan Gupta, Rajesh https://escholarship.org/uc/item/7682n2h8 Digital currencies have flourished in recent years, buoyed by the tremendous success of Bitcoin. These blockchain-based currencies, called altcoins, have attracted enthusiasts who enter the market by mining or buying them. To mine or to buy, however, can be a difficult decision; each altcoin is different from another, and the market tends to be volatile. In this work, we analyze the profitability of mining and speculation for 36 altcoins using real-world blockchain and trade data. Using opportunity cost as a metric, we estimate the mining cost for a coin with respect to a more popular coin. For every dollar invested in mining or buying a coin, we also estimate the revenue under various conditions, such as time of market entry and hold positions. While some coins offer the potential for spectacular returns, many follow a simple bubble-and-crash scenario, which highlights the extreme risks---and potential gains---in altcoin markets.<p>Pre-2018 CSE ID: CS2017-1019</p> https://escholarship.org/uc/item/7682n2h8 Tue, 14 Mar 2023 00:00:00 +0000 Huang, Danny Yuxing Levchenko, Kirill Snoeren, Alex C https://escholarship.org/uc/item/73f2501s Sensor network simulators are important tools for the design, implementation and evaluation of wireless sensor networks. Due to the large computational requirements necessary for simulating wireless sensor networks with high fidelity, many wireless sensor network simulators, especially the cycle accurate ones, employ distributed simulation techniques to leverage the combined resources of multiple processors or computers. However, the large overheads in synchronizing sensor nodes during distributed simulations of sensor networks result in a significant increase in simulation time. In this paper, we present a novel technique that could significantly reduce such overheads by minimizing the number of sensor node synchronizations during simulations. We implement this technique in Avrora, a widely used parallel sensor network simulator, and achieve a speedup of up to 11 times in terms of average simulation speed in our test cases. For applications that have lower duty cycles, the speedups... https://escholarship.org/uc/item/73f2501s Tue, 14 Mar 2023 00:00:00 +0000 Jin, Zhong-Yi Gupta, Rajesh https://escholarship.org/uc/item/72t332tp We develop an efficient optimization procedure for learning linear transformations of data to produce structured outputs, e.g., nearest-neighbor rankings or connectivity graphs induced by distance. The proposed method solves a semi-definite programming problem by reducing to a sequence of small quadratic programs, resulting in significant reductions in training complexity. Experiments demonstrate that the proposed method is robust, efficient, and outperforms alternative methods in high-noise settings.<p>Pre-2018 CSE ID: CS2012-0979</p> https://escholarship.org/uc/item/72t332tp Tue, 14 Mar 2023 00:00:00 +0000 Mcfee, Brian Lanckriet, Gert https://escholarship.org/uc/item/72p4n0b4 Manufacturing and environmental variability lead to timing errors in computing systems that are typically corrected by error detection and correction mechanisms at the circuit level. The cost and speed of recovery can be improved by memoization-based optimization methods that exploit spatial or temporal parallelisms in suitable computing fabrics such as general-purpose graphics processing units (GPGPUs). We propose here a temporal memoization technique for use in floating-point units (FPUs) in GPGPUs that uses value locality inside data-parallel programs. The technique recalls (memorizes) the context of error-free execution of an instruction on a FPU. Therefore, it avoids redundant execution and saves energy for FPU. To enable scalable and independent recovery, a single-cycle lookup table (LUT) is tightly coupled to every FPU to maintain contexts of recent error-free executions. The LUT reuses these memorized contexts to exactly, or approximately, correct errant FP instructions... https://escholarship.org/uc/item/72p4n0b4 Tue, 14 Mar 2023 00:00:00 +0000 Rahimi, Abbas Benini, Luca Gupta, Rajesh https://escholarship.org/uc/item/70s1n7n9 Image segmentation and object recognition are among the most fundamental problems in computer vision, and the potential interaction between these tasks has been discussed for many years. The usefulness of recognition for segmentation has been demonstrated with various top-down segmentation algorithms, however, the impact of bottom-up image segmentation as pre-processing for object recognition is not well understood. One factor impeding the utility of segmentation for recognition is the unsatisfactory quality of image segmentation algorithms. In this work we take advantage of a recently proposed method for computing multiple stable segmentations and illustrate the application of bottom-up image segmentation as a preprocessing step for object recognition and categorization. We extend a popular bag-of-features recognition model to provide multiple class categorization and localization of objects in images. We compare our categorization results to that of a conventional bag-of-features... https://escholarship.org/uc/item/70s1n7n9 Tue, 14 Mar 2023 00:00:00 +0000 Rabinovich, Andrew Vedaldi, Andrea Belongie, Serge https://escholarship.org/uc/item/6v11p5b0 This paper introduces a simple shading model for cloth. The model can simulate both the anisotropic highlights on cloth as well as the complex color shifts seen in cloth made of different colored threads. Our model is based on extensive Bidirectional Reflectance Distribution Function (BRDF) measurements of several cloth samples. We have also measured the scattering profile of several different individual cloth threads. Based on these measurements we have derived a simple model capable of predicting the light scattered by different threads. We model cloth as a collection of threads in a weave pattern which provides information about the coverage of the different thread types as well as their tangent directions. Our model also accounts for shadowing and masking by the threads. We validate our model by comparing predicted and measured values and we show how we can use the model to recover parameters for different cloth samples including silk, velvet, linen, and polyester with varying... https://escholarship.org/uc/item/6v11p5b0 Tue, 14 Mar 2023 00:00:00 +0000 Sadhegi, Iman Bisker, Oleg De Deken, Joachim Wann Jensen, Henrik https://escholarship.org/uc/item/6tt9z3n5 Power over-subscription can reduce costs for modern data centers. However, designing the power infrastructure for a lower operating power point than the aggregated peak power of all servers requires dynamic techniques to avoid high peak power costs and, even worse, tripping circuit breakers. This work presents an architecture for distributed per-server UPSs that stores energy during low activity periods and uses this energy during power spikes. This work advocates explicit sizing of the distributed UPS batteries for power capping and provides a methodology to properly select the properties of the battery in order to reduce total datacenter cost of ownership per server. Depending on workload diurnal patterns and oversubscription assumption this approach can achieve up 5-10% reduction in overall costs per server.<p>Pre-2018 CSE ID: CS2012-0985</p> https://escholarship.org/uc/item/6tt9z3n5 Tue, 14 Mar 2023 00:00:00 +0000 Kontorinis, Vasileios Sampson, Jack Zhang, Liuyi Eric Akslani, Baris Homayoun, Houman Rosing, Tajana Simunic Tullsen, Dean https://escholarship.org/uc/item/6ps421k5 Solving "Big Data" problems requires bridging massive quantities of compute, memory, and storage, which requires significant amounts of bisection bandwidth. Topologies like Fat-tree, VL2, and HyperX achieve a scale-out design by leveraging multiple paths from source to destination. However, traditional routing protocols are not able to effectively utilize these links while also harnessing spare network capacity to better statistically multiplex network resources. In this work we present Dahu, a switch mechanism for efficiently load balancing traffic in multipath networks. Dahu avoids congestion hot-spots by dynamically spreading traffic uniformly across links, and forwarding traffic over non-minimal paths where possible. By performing load balancing primarily using local information, Dahu can act more quickly than centralized approaches, and responds to failure gracefully. Our evaluation shows that Dahu delivers up to 50% higher throughput relative to ECMP in an 8,192 server Fat-tree... https://escholarship.org/uc/item/6ps421k5 Tue, 14 Mar 2023 00:00:00 +0000 Radhakrishnan, Sivasankar Kapoor, Rishi Tewari, Malveeka Porter, George Vahdat, Amin https://escholarship.org/uc/item/6pg8f6vh As the web evolves, it is becoming easier to form communities based on shared interests, and to create and publish data on a wide variety of topics. With this "democratization of information creation" comes the natural desire to make one's data accessible for querying within the community and also be able to query the global collection that is the union of all local data collections of others within the community. In order to fully deliver on the promise of free data exchange, any community-supporting infrastructure needs to enforce the key requirement of being resistant to censorship by third parties, be they of governmental, corporate, of other special interest nature. Censorship resistance precludes some obvious approaches that reuse and build on existing centralized technologies, e.g., search engines, hosted online communities, etc. We propose a distributed infrastructure in which data resides only with the publishers owning it. The infrastructure disseminates user queries... https://escholarship.org/uc/item/6pg8f6vh Tue, 14 Mar 2023 00:00:00 +0000 Curtmola, Emiran Deutsch, Alin Ramakrishnan, K. K. Srivastava, Divesh https://escholarship.org/uc/item/6mj094vk We present a novel scalable architecture for matrix inversion that uses the modified Gram-Schmidt algorithm based on QR decomposition. Our core achieves a throughput of 0.18M updates per second for a 4 x 4 matrix using 19 bits of precision on a Xilinx Virtex4 SX FPGA. We also present two different designs which use longer data lines, 26 and 32 bits, and compare our results with another matrix inversion architecture which is the only scalable approach so far. We show that our core is significantly faster than the other published FPGA implementation as it requires fewer resources due to the usage of fixed point arithmetic and an effective resource utilization. We show that our proposed architecture is scalable by presenting the results for 6 x 6 and 8 x 8 matrices.<p>Pre-2018 CSE ID: CS2009-0938</p> https://escholarship.org/uc/item/6mj094vk Tue, 14 Mar 2023 00:00:00 +0000 Irturk, Ali Mirzaei, Shahnam Kastner, Ryan https://escholarship.org/uc/item/6mc0v607 Simulators are important tools for the design, implementation and evaluation of sensor networks and have been the subject of intense research in the past decade. The resource constrained nature of sensor platforms as well as the unique working environments and deployment scenarios of sensor networks give special requirements and challenges to simulator design. More than 20 simulators have been developed for sensor networks by various research groups to address different problems and needs. In this paper, we first explore the requirements for a modern sensor network simulator. Then we study simulation techniques and classify existing sensor network simulators according to their architectures, designs and levels of abstractions. Finally, we discuss the future research directions of wireless sensor network simulators.<p>Pre-2018 CSE ID: CS2008-0925</p> https://escholarship.org/uc/item/6mc0v607 Tue, 14 Mar 2023 00:00:00 +0000 Jin, Zhong-Yi Gupta, Rajesh https://escholarship.org/uc/item/6km5s82h Migrate is a system service that allows network applications to suspend upon disconnection and properly resume when connectivity is restored. Migrate uses session continuations, which allow applications to explicitly record all the state and resources required to correctly resume. Migrate virtualizes network connections to function across network address changes, allows portions of an application to be suspended and securely resumed, and optimizes scarce system resources on servers handling many suspended clients. We demonstrate that the continuation abstraction is both useful and efficient by showing how two important applications, SSH and Icecast, can use Migrate to implement a suspend/resume feature. We find that the required source code changes are between 0.5% and 1.75% of the total, that the generated continuations are between 1% and 5% of the memory footprint of the entire process, and that session resumption times are small (a few hundred milliseconds) compared to typical... https://escholarship.org/uc/item/6km5s82h Tue, 14 Mar 2023 00:00:00 +0000 Snoeren, Alex C Panigrahi, Debashis Mukhopadhyay, Shoubhik Balakrishnan, Hari Kaashoek, M. Frans https://escholarship.org/uc/item/6gs6z92r We study the problem of querying data sources that accept only a limited set of queries, such as sources accessible by Web services which can implement very large (potentially infinite) families of queries. We revisit a classical setting in which the application queries are conjunctive queries and the source accepts families of (possibly parameterized) conjunctive queries specified as the expansions of a (potentially recursive) Datalog program with parameters. We say that query Q is expressible by the program P if it is equivalent to some expansion of P. Q is supported by P if it has an equivalent rewriting using some finite set of P's expansions. We present the first study of expressibility and support for sources that satisfy integrity constraints, which is generally the case in practice. We start by closing a problem left open by prior work even while ignoring constraints, namely the precise relationship between expressibility and support: surprisingly, we show them to be... https://escholarship.org/uc/item/6gs6z92r Tue, 14 Mar 2023 00:00:00 +0000 Cautis, Bogdan Deutsch, Alin Onose, Nicola https://escholarship.org/uc/item/6f6904sc We describe CACTI-IO, an extension to CACTI that includes power, area and timing models for the IO and PHY of the off-chip memory interface for various server and mobile configurations. CACTI-IO enables quick design space exploration of the off-chip IO along with the DRAM and cache parameters. We describe the models added to CACTI-IO that help include the off-chip impact to the tradeoffs between memory capacity, bandwidth and power. This technical report also provides three standard configurations for the input parameters (DDR3, LPDDR2 and Wide-IO) and illustrates how the models can be modified for a custom configuration. The models are validated against SPICE simulations and show that we are within 0-15% error for different configurations. We also compare with measured results.<p>Pre-2018 CSE ID: CS2012-0986</p> https://escholarship.org/uc/item/6f6904sc Tue, 14 Mar 2023 00:00:00 +0000 Jouppi, Norman Kahng, Andrew Muralimanohar, Naveen Srinivas, Vaishnav https://escholarship.org/uc/item/6cq720mg As single TCP flows approach 10 Gbps, static hash ECMP load balancing — used by routers today– does a poor job of balancing load in data centers. We describe a new load balancing algorithm, Flame, that is implementable at 480 Gbps with small memory and uses two novel mechanisms. First, Flame uses a Discounting Rate Estimator (DRE); unlike exponential averaging, DRE quickly measures bursts and yet retains memory of recent bursts. Second, Flame binds flows to hash functions and not to paths. We show Flame is more resilient and efficient than the earlier Flare scheme, and provides better load balancing and is more deployable than Hedera. Flame also allows rebalancing of flows in hardware at rapid rates. This is interesting because we show TCP experiments at 1 and 10 Gbps that demonstrate that recent Linux stacks after 2.6.14 can tolerate rebalancing once every 10 packets with negligible loss of throughput. On the other hand, Windows 2008 stacks have degraded TCP throughput if rebalancing... https://escholarship.org/uc/item/6cq720mg Tue, 14 Mar 2023 00:00:00 +0000 Edsall, Tom Fingerhut, Andy Lam, Terry Pan, Rong Varghese, George https://escholarship.org/uc/item/6c4778j8 In many machine learning applications, precisely labeled data is either burdensome or impossible to collect. Multiple Instance Learning (MIL), in which training data is provided in the form of labeled bags rather than labeled instances, is one approach for dealing with ambiguously labeled data. In this paper we argue that in many applications of MIL (e.g. image, audio, text, bioinformatics) a single bag actually consists of a large or infinite number of instances, such as all points on a low dimensional manifold. For practical reasons, these bags get subsampled before training. Instead, we introduce a MIL formulation which directly models the underlying structure of these bags. We propose and analyze the query bag model, in which instances are obtained by repeatedly querying an oracle in a way that can capture relationships between instances. We show that sampling more instances results in better classification performance, which motivates us to develop algorithmic strategies... https://escholarship.org/uc/item/6c4778j8 Tue, 14 Mar 2023 00:00:00 +0000 Babenko, Boris Dollar, Piotr Belongie, Serge https://escholarship.org/uc/item/69g3t3hg Unlike multi-hop forwarding, an alternative and relatively new approach for efficient data collection in sensor networks is to exploit the mobility. Particularly, we consider collecting data from static sensor nodes using a “data mule” via wireless communication. A data mule is a mobile node that has wireless communication capability and also a sufficient amount of storage to store the data from the sensors in the field. A data mule travels across the sensing field and collects data from each sensor node when the distance is short, and later deposits all the data to the base station. In this way, each sensor node can conserve a significant amount of energy, since it only needs to send the data over a shorter distance and has no need to forward other sensors’ data all the way to the base station. In this paper, we are interested in the following problem: “how to control a data mule such that it collects data from all the nodes in the minimal amount of time”. We call it the data... https://escholarship.org/uc/item/69g3t3hg Tue, 14 Mar 2023 00:00:00 +0000 Sugihara, Ryo Gupta, Rajesh https://escholarship.org/uc/item/67p7k1pr We consider the problem of planning path and speed of a “data mule” in a sensor network. This problem is encountered in various situations, such as modeling the motion of a data-collecting UAV in a field of sensors for structural health monitoring. Our specific context here is use of a data mule as an alternative or supplement to multihop forwarding in a sensor network. While a data mule can reduce the energy consumption at each sensor node, it increases the latency from the time the data is generated at a node to the time the base station receives it. In this paper, we introduce the “data mule scheduling” or DMS framework that enables data mule motion planning to minimize the data delivery latency. The DMS framework is general; it can express many previously proposed problem formulations and problem settings related to data mules. We design algorithms for DMS and extend to the more general case of combined data mule and multihop forwarding to enable a flexible trade-off between... https://escholarship.org/uc/item/67p7k1pr Tue, 14 Mar 2023 00:00:00 +0000 Sugihara, Ryo Gupta, Rajesh https://escholarship.org/uc/item/6325p8bz Virtual Network Computing, or VNC, is a popular thin client application used to access files and applications on remote computers. It is especially relevant as infrastructure to support ubiquitous computing applications, as it offers a way to run data-and-computation-intensive applications and allow users to access them through lightweight devices. However, VNC can suffer from significant losses in throughput when there is high latency between the client and server. In this work, we present a Message Accelerator proxy for VNC. This Message Accelerator mitigates high latency network effects while maintaining the advantages of a client-pull system. By operating near/on the server, it can send updates to the client at a rate corresponding to proxy-server interactions which are faster than client-server interactions. When testing using video, our Message Accelerator design results in frame rates an order of magnitude higher than plain VNC when running under high latency conditions.<p>Pre-2018... https://escholarship.org/uc/item/6325p8bz Tue, 14 Mar 2023 00:00:00 +0000 Taylor, Cynthia Pasquale, Joseph https://escholarship.org/uc/item/62f8k8cq We present a new algorithm, Locally Smooth Manifold Learning (LSML), that learns a warping function from a point on an manifold to its neighbors. Important characteristics of LSML include the ability to recover the structure of the manifold in sparsely populated regions and beyond the support of the provided data. Applications of our proposed technique include embedding with a natural out-of-sample extension and tasks such as tangent distance estimation, frame rate up-conversion, video compression and motion transfer.<p>Pre-2018 CSE ID: CS2007-0876</p> https://escholarship.org/uc/item/62f8k8cq Tue, 14 Mar 2023 00:00:00 +0000 Dollar, Piotr Rabaud, Vincent Belongie, Serge https://escholarship.org/uc/item/5w9600dd The prolific spread of mobile phones through all corners of the globe has only been matched by their rapid increase in computing power. As cellular phones become further integrated into the fabric of everyday life, their value to attackers will rise accordingly. As a result, the widespread debilitating outbreak of self-propagating malware in the cell phone environment is a matter of "when", rather than "if." Although self-propagating malware is well understood in the Internet, mobile phone networks have very different characteristics in terms of topologies, services, provisioning and capacity, devices, and communication patterns. To understand the propagation of malware in this new environment, we have developed an event-driver simulator that captures the characteristics and constraints of mobile phone networks. Key elements of the simulator are a network topology generator (RACoON), which creates realistic topologies and provisioned capacities of the network infrastructure,... https://escholarship.org/uc/item/5w9600dd Tue, 14 Mar 2023 00:00:00 +0000 Fleizach, Chris https://escholarship.org/uc/item/5kp9f7cq We consider Location-based Service (LBS) settings, where a LBS provider logs the requests sent by mobile device users over a period of time and later wants to publish/share these logs. Log sharing can be extremely valuable for advertising, data mining research and network management, but it poses a serious threat to the privacy of LBS users. Sender anonymity solutions prevent a malicious attacker from inferring the interests of LBS users by associating them with their service requests after gaining access to the anonymized logs. With the fast-increasing adoption of smartphones and the concern that historic user trajectories are becoming more accessible, it becomes necessary for any sender anonymity solution to protect against attackers that are trajectory-aware (i.e. have access to historic user trajectories) as well as policy-aware (i.e they know the log anonymization policy). We call such attackers TP-aware. This paper introduces a first privacy guarantee against TP-aware attackers,... https://escholarship.org/uc/item/5kp9f7cq Tue, 14 Mar 2023 00:00:00 +0000 Deutsch, Alin Hull, Richard Vyas, Avinash Keliang Zhao, Kevin https://escholarship.org/uc/item/5jd3c1gv As industry moves towards many-core chips, networks-on-chip (NoCs) are emerging as the scalable fabric for interconnecting the cores. With power now the first-order design constraint, early-stage estimation of NoC power has become crucially important. ORION was amongst the first NoC power models released, and has since been fairly widely used for early-stage power estimation of NoCs. However, when validated against recent NoC prototypes – the Intel 80-core Teraflops chip and the Intel Scalable Communications Core (SCC) chip – we saw significant deviation that can lead to erroneous NoC design choices. This prompted our development of ORION 2.0, an extensive enhancement of the original ORION models which includes completely new subcomponent power models, area models, as well as improved and updated technology models. Validating against the two Intel chips saw ORION 2.0 bringing a substantial improvement in accuracy over the original ORION. A case study with these power models plugged... https://escholarship.org/uc/item/5jd3c1gv Tue, 14 Mar 2023 00:00:00 +0000 Kahng, Andrew Li, Bin Peh, Li-Shiuan Samadi, Kambiz https://escholarship.org/uc/item/5gk6v3fx In this paper, we propose a novel set of techniques that allows multicore programmers and architects alike to rapidly estimate the availability of parallelism in their target programs. We present p-ray, which, given an application and the input, shows the nested relationship and availability of parallelism across different regions in complex programs, using a parallelism chart, or p-chart. P-ray also prioritizes the regions, suggesting the ordering that the programmer should attack them in, based on the estimated parallelism and the code complexity. P-ray can also estimate the kind of parallelism (TLP, DLP or ILP) that is found in each region, which allows the user to determine the available techniques are likely to work. In this paper, we show three case studies, including SpecInt's gzip, and Mat2C's capacitor, and Nasa7's Vpenta, and follow up with results which compare actual parallelization results of a benchmark suite against the results predicted by p-ray.<p>Pre-2018 CSE... https://escholarship.org/uc/item/5gk6v3fx Tue, 14 Mar 2023 00:00:00 +0000 Garcia, Saturnino Louie, Christopher Jeon, Donghwan Kota Venkata, Sravanthi Gupta, Anshuman Taylor, Michael https://escholarship.org/uc/item/5gh377sq Performance testing is a standard practice for evolving systems to detect performance issues proactively. It samples various performance metrics that will be compared with a stable baseline to judge whether the measurement data is abnormal. This type of comparative analysis requires domain expertise, which can take experienced performance analysts days to conduct. In an effort to build an automatic solution for a leading data warehousing company to improve the comparative performance analysis efficiency, we implemented machine learning approaches proposed by existing research. But the initial result has a 86% false negative rate on average, which means the majority of performance defects would be missed. To investigate causes for this unsatisfying result, we take a step back to revisit the performance data itself and find several important data related issues that are overlooked by existing work. In this paper, we discuss in detail these issues and share our hindsights to address... https://escholarship.org/uc/item/5gh377sq Tue, 14 Mar 2023 00:00:00 +0000 Huang, Peng Schechter, Craig Chen, Vincent Hill, Steven Shen, Dongcai Zhou, Yuanyuan Saul, Lawrence https://escholarship.org/uc/item/5b87j38b The Smart Grid is drawing attention from various research areas. Distributed control algorithms at different scales within the grid are being developed and deployed; yet their effects on each other and the grid's health and stability has not been sufficiently studied due to the lack of a capable simulator. Simulators in the literature can solve the power flow by modeling the physical system, but fail to address the cyber physical aspect of the smart grid with multiple agents. To answer these questions, we have developed S2Sim: Smart Grid Swarm Simulator. S2Sim allows any object within the grid to have its own independent control, transforming physical elements into cyber-physical representations. Objects can have any size ranging from a light bulb to a whole microgrid and their representative data can be supplied from a real device, simulation, distributed control algorithm or a database. S2Sim shields the complexity of the power flow solution from the control algorithms and directly... https://escholarship.org/uc/item/5b87j38b Tue, 14 Mar 2023 00:00:00 +0000 Akyurek, Alper Sinan Akslani, Baris Rosing, Tajana Simunic https://escholarship.org/uc/item/5754j51b We report on the development and deployment of a pilot of the new Advanced Placement CS Principles course in the United States. The course is designed to introduce core computational concepts and instill computational thinking practices. We report on an initial offering with 571 university students, mostly non-CS majors taking the course as a general education requirement. We discuss the instructional design supporting the course, describe how the various components were implemented, and review student work and valuation of the course. Though the course appears to “teach programming” in Alice, students reported gaining significant analysis and communication skills they could use in their daily life. We reflect on how instructional design decisions are likely to have supported this experience and consider the implications for other K-12 computing/IT education efforts as well as for regular CS1 courses.<p>Pre-2018 CSE ID: CS2011-0965</p> https://escholarship.org/uc/item/5754j51b Tue, 14 Mar 2023 00:00:00 +0000 Simon, Beth Esper, Sarah Quintin, Cutts https://escholarship.org/uc/item/54g6w2g5 Fast non-volatile memories will soon make their appearance on the processor memory bus. They offer the potential for extremely low-latency, high-bandwidth access to persistent files. However, existing interfaces will impose large system call and file system overheads on those accesses, squandering the memories' performance. At the same time, we would like to leverage existing file systems as much as possible, to reduce the cost of adopting these new storage technologies. We present a user space library, called Quill, that avoids system call overheads on most accesses by interposing on file operations and converting them into memory operations. Quill enforces the protections that the file system provides by transferring protection information into the process's page table and accessing files directly via load and store instructions. We describe the Quill system and compare it to a conventional software stack accessing the same storage device.<p>Pre-2018 CSE ID: CS2013-0991</p> https://escholarship.org/uc/item/54g6w2g5 Tue, 14 Mar 2023 00:00:00 +0000 Eisner, Louis Alex Mollov, Todor Swanson, Steven https://escholarship.org/uc/item/52m1h315 Poor air quality is a growing global health concern that impacts millions of people worldwide. Although we are beginning to understand the health impacts of air pollution, it remains a challenge to provide people with the information they need to be able to make health-conscious choices. The CitiSense system gives individuals the real-time tools they need to be able to identify when and where they are exposed to poor air. We present the results of a qualitative study regarding a 4-week “in the wild” deployment of the CitiSense air-quality sensor and system. We focus on how the 16 participants responded to their new-found information about their environment, how they shared information, and what kinds of actions were enabled by having access to real-time air-quality data. Quantitative data gathered through the course of the study frames participant responses by showing what levels of pollution were experienced and what activities heightened exposure. We found that CitiSense’s real-time... https://escholarship.org/uc/item/52m1h315 Tue, 14 Mar 2023 00:00:00 +0000 Bales, Elizabeth Nikzad, Nima Ziftci, Celal Quick, Nichole Griswold, William Patrick, Kevin https://escholarship.org/uc/item/51v4s0p7 Distributed simulation techniques are commonly used to improve the speed and scalability of simulators for wireless sensor networks. However, accurate simulations of dynamic interactions of sensor network applications incur large synchronization overheads and severely limit the performance of existing distributed simulators. In this paper, we present two novel techniques that significantly reduce such overheads by minimizing the number of sensor node synchronizations during simulations. These techniques work by exploiting radio and MAC specific characteristics without reducing simulation accuracy. In addition, we present a new mechanism that makes it possible to exploit any potential application specific characteristics for synchronization reductions. We implement and evaluate these techniques in a cycle accurate distributed simulation framework that we developed based on Avrora, a popular parallel sensor network simulator. In our experiments, the techniques achieve a speedup from... https://escholarship.org/uc/item/51v4s0p7 Tue, 14 Mar 2023 00:00:00 +0000 Jin, Zhong-Yi Gupta, Rajesh https://escholarship.org/uc/item/51t2h7xv Log data from DNS resolvers contain rich information that is quite useful for various research use cases such as estimating the popularity of websites. Log data from approximately 39k resolvers has been collected and stored on HDFS. The data is so huge and not optimally structured that it takes a lot of time and resources to search records of interest from the the log. In this project, we investigate techniques to port the log data to a new format so that it speeds up the query time and takes less resources both to store the data and to query the data. We investigated bzip compression, reformatting/pruning unessential records and partitioning the records into separate buckets and from our experiments, we found that using a combination of reformatting/pruning records with partitioning and efficiently sorting the records based on multiple fields speeds up the domain query by 6 times and takes approximately 8 times less resources to query in comparison with unstructured data. Also,... https://escholarship.org/uc/item/51t2h7xv Tue, 14 Mar 2023 00:00:00 +0000 Dhakshinamurthy, Prabu https://escholarship.org/uc/item/4wm358ht Advanced non-volatile, solid-state storage technologies such as phase change memory promise enormous gains in performance relative to both conventional disks and flash memory-based storage devices. However, existing abstractions for accessing non-volatile data (i.e., file systems and the associated system calls) cannot exploit the performance and flexibility these memories offer. We describe a new, transaction-based interface that allows programmers to implement fast, scalable, durable data structures that are robust in the face of unexpected system failures. The interface maps the non-volatile storage directly into the application's address space and allows volatile and non-volatile data structures to interact seamlessly in programs. Since the interface removes the operating system from the critical path of basic read and write operations, it can realize up to 2350X speedups relative to transactional storage systems that target the conventional file-based interface. We describe... https://escholarship.org/uc/item/4wm358ht Tue, 14 Mar 2023 00:00:00 +0000 Coburn, Joel Caulfield, Adrian Grupp, Laura Adel, Ameen Swanson, Steven https://escholarship.org/uc/item/4w79f9rb Systems that provide powerful transaction mechanisms often rely on write-ahead logging (WAL) implementations that were designed with slow, disk-based systems in mind. The emerging class of fast, byte-addressable, non-volatile memory (NVM) technologies (e.g., phase change memories, spin-torque MRAMs, and the memristor), however, present performance characteristics very different from both disks and flash-based SSDs. This paper addresses the problem of designing a WAL scheme optimized for these fast NVM-based storage systems. We examine the features that a system like ARIES, a WAL algorithm popular for databases, must provide and separate them from the implementation decisions ARIES makes to optimize for disk-based systems. We design a new NVM-optimized WAL scheme (called MARS) in tandem with a novel SSD multi-part atomic write primitive that combine to provide the same features as ARIES does without any of the disk-centric limitations. The new atomic write primitive makes... https://escholarship.org/uc/item/4w79f9rb Tue, 14 Mar 2023 00:00:00 +0000 Coburn, Joel Bunker, Trevor Gupta, Rajesh K Swanson, Steven https://escholarship.org/uc/item/4w71405x Matrix computations such as matrix decomposition and inversion are essential for various algorithms which are employed in wireless communication. FPGAs are ideal platforms for such applications; however, the need for vast amounts of customization throughout the design process of a matrix computation core can overwhelm the designer. This paper presents an automatic generation and optimization methodology for different matrix computation architectures using a generator tool, GUSTO, that we developed to enable easy design space exploration with different parameterization options. We especially concentrate on wireless communication MIMO-OFDM applications which often use small matrix dimensions. We present automatic generation of a variety of general purpose matrix computation architectures and optimized application specific architectures. GUSTO’s application specific architectures have comparable results to published architectural implementations, but offer the advantage of providing... https://escholarship.org/uc/item/4w71405x Tue, 14 Mar 2023 00:00:00 +0000 Irturk, Ali Benson, Bridgit Laptev, Nikolay Kastner, Ryan https://escholarship.org/uc/item/4w0246js Engineering large-scale data center applications built from thousands of commodity nodes requires both an underlying network that supports a wide variety of traffic demands, and low latency at microsecond timescales. Many ideas for adding innovative functionality to networks, especially active queue management strategies, require either modifying packets or performing alternative queuing to packets in-flight on the data plane. However, configuring packet queuing, marking, and dropping is challenging, since buffering in commercial switches and routers is not programmable. In this work, we present NetBump, a platform for experimenting with, evaluating, and deploying a wide variety of active queue management strategies to network data planes with minimal intrusiveness and at low latency. NetBump leaves existing switches and endhosts unmodified by acting as a ``bump on the wire,'' examining, marking, and forwarding packets at line rate in tens of microseconds to implement a variety... https://escholarship.org/uc/item/4w0246js Tue, 14 Mar 2023 00:00:00 +0000 Porter, George Kapoor, Rishi Das, Ambit Al-Fares, Mohammad Weatherspoon, Hakim Prabhakar, Balaji Vahdat, Amin https://escholarship.org/uc/item/4qj0p4w7 Enterprise Chat has emerged as one of the key tools for rapid communication, decision making, and situational awareness across a wide spectrum of application domains, ranging from massively multi-player online games (MMOs) to multi-national corporations to public safety and defense. This report is motivated by the increasing need to flexibly bridge multiple existing and emerging standards and technology platforms for Enterprise Chat. A promising approach to this effect is the use of Service-Oriented Architecture methodology and technology, as this combination has been used successfully across the industry in other enterprise integration projects. To determine the viability of this approach, we have performed a case study on the use of an SOA-based approach to unifying disparate IM/chat systems into a system-of-systems framework. This approach leverages enterprise-wide services, but without risking the IM/chat systems’ existing functionality. In collaboration with the Space and... https://escholarship.org/uc/item/4qj0p4w7 Tue, 14 Mar 2023 00:00:00 +0000 Demchak, Barry Krueger, Ingolf https://escholarship.org/uc/item/4p74g215 Accurately estimating all link-level properties of a large network has proven to be very difficult. The measurements used for these estimates require significant collaboration from all endpoints on the network, significantly reducing their applicability for large scale Internet measurements. We present a scalable approach using a small number of hosts without collaboration from existing routers and minimal collaboration between the hosts. Our approach is based on adaptive sampling. Initially, each host probes a set of receivers at a low frequency. When packet losses are detected, the sampling rate increases. By detecting correlations between time series and combining them with information about network connectivity, the host identifies a set of suspected lossy routers. Hosts then communicate with each other, combining evidence to identify routers with high packet loss. Our experiments show that using a relatively small set of hosts and receivers, we can gather sufficient evidence... https://escholarship.org/uc/item/4p74g215 Tue, 14 Mar 2023 00:00:00 +0000 Ettinger, Evan McFee, Brian Freund, Yoav https://escholarship.org/uc/item/4nr7f343 Microarchitectural simulation of multithreaded architectures with shared resources, such as simultaneous multithreading (SMT) cores and multi-core processors with shared caches, is time-consuming and the results of simulation may be dicult to interpret. It is time-consuming because modern benchmarks run for hundreds of billions (or even trillions) of instructions, and accurate multi-core and SMT simulation requires higher-detail models than single-threaded simulation. The statistics collected when two programs execute together can be dicult to interpret because the programs both exhibit independent phase behavior and affect each other's execution. Starting one program slightly later than during the original execution will change the phases that execute together and thus change the eects that the programs have on each other. <p> Accurate sampled simulation requires accurate sample collection. We evaluate techniques to improve sampling accuracy and performance, both for single-threaded... https://escholarship.org/uc/item/4nr7f343 Tue, 14 Mar 2023 00:00:00 +0000 VanBiesbrouck, Michael https://escholarship.org/uc/item/4hr514qs We investigate a recently proposed family of positive-definite kernels that mimic the computation in large neural networks. We examine the properties of these kernels using tools from differential geometry; specifically, we analyze the geometry of surfaces in Hilbert space that are induced by these kernels. When this geometry is described by a Riemannian manifold, we derive results for the metric, curvature, and volume element. Interestingly, though, we find that the simplest kernel in this family does not admit such an interpretation. We explore two variations of these kernels that mimic computation in neural networks with different activation functions. We experiment with these new kernels on several data sets and highlight their general trends in performance for classification.<p>Pre-2018 CSE ID: CS2011-0972</p> https://escholarship.org/uc/item/4hr514qs Tue, 14 Mar 2023 00:00:00 +0000 Cho, Youngmin Saul, Lawrence https://escholarship.org/uc/item/4gp4r437 Today’s large-scale services generally exploit loosely-coupled architectures that restrict functionality requiring tight cooperation (e.g., leader election, synchronization, and reconfiguration) to a small subset of nodes. In contrast, this work presents a way to scalably deploy tightly-coupled distributed systems that require significant coordination among a large number of nodes in the wide area. Our design relies upon a new reliable group membership abstraction to ensure that either group members are capable of communicating or that new groups form. In particular, we deploy a distributed rate limiting (DRL) service within a global testbed infrastructure. Unlike most distributed services, DRL can safely operate in separate partitions simultaneously, but requires timely snapshots of global state within each. Our DRL implementation leverages our proposed group membership abstraction and a robust gossip-based communication protocol, conjoining the fates of view maintenance and... https://escholarship.org/uc/item/4gp4r437 Tue, 14 Mar 2023 00:00:00 +0000 Webb, Kevin C Vattikonda, Bhanu C Yocum, Kenneth Snoeren, Alex C https://escholarship.org/uc/item/4f90w1zq We present a protocol for general state machine replication -- a method that provides strong consistency -- that has high performance in a wide-area network. In particular, our protocol Mencius has high throughput under high client load and low latency under low client load even under changing wide-area network environment and client load. We develop our protocol as a derivation from the well-known protocol Paxos. Such a development can be changed or further refined to take advantage of specific network or application requirements.<p>Pre-2018 CSE ID: CS2008-0930</p> https://escholarship.org/uc/item/4f90w1zq Tue, 14 Mar 2023 00:00:00 +0000 Mao, Yanhua Junqueira, Flavio Marzullo, Keith https://escholarship.org/uc/item/4bf9f938 GPU and FPGA-based clouds have already demonstrated the promise of accelerating computing-intensive workloads with greatly improved power and performance. In this paper, we examine the design of ASIC Clouds, which are purpose-built datacenters comprised of large arrays of ASIC accelerators, whose purpose is to optimize the total cost of ownership (TCO) of large, high-volume chronic computations, which are becoming increasingly common as more and more services are built around the Cloud model. On the surface, the creation of ASIC clouds may seem highly improbable due to high NREs and the inflexibility of ASICs. Surprisingly, however, large-scale ASIC Clouds have already been deployed by a large number of commercial entities, to implement the distributed Bitcoin cryptocurrency system. We begin with a case study of Bitcoin mining ASIC Clouds, which are perhaps the largest ASIC Clouds to date. From there, we design three more ASIC Clouds, including a YouTube-style video transcoding... https://escholarship.org/uc/item/4bf9f938 Tue, 14 Mar 2023 00:00:00 +0000 Magaki, Ikuo Khazraee, Moein Gutierrez, Luis Vega Taylor, Michael Bedford https://escholarship.org/uc/item/49d79511 HVAC systems constitute 40% of energy in commercial buildings, and faults in HVAC account for 5% to 20% of its energy consumption. Typically, HVAC in modern buildings are managed using Building Managing Systems (BMS), and fault detection is one of the essential services provided by BMS to keep HVAC operational. We study the fault management practices followed in real commercial buildings, and find that current techniques used for fault detection fail to detect large number of efficiency related faults. Based on our findings, we developed BuildingSherlock (BDSherlock), which is a web service based fault management framework that exposes building information to automatically detect faults using useful algorithms. We deploy BDSherlock in a 145000 sqft building at UC San Diego, successfully detected 87 faults using data driven analysis.<p>Pre-2018 CSE ID: CS2014-1007</p> https://escholarship.org/uc/item/49d79511 Tue, 14 Mar 2023 00:00:00 +0000 Teraoka, Hidetoshi Balaji, Bharathan Zhang, Rizhen Nwokafor, Anthony Narayanaswamy, Balakrishnan Agarwal, Yuvraj https://escholarship.org/uc/item/49297740 Recent years have witnessed significant gains in the adoption of flash technology due to increases in chip bit density, enabling higher capacities and lower prices. Unfortunately, these improvements come at a significant cost to performance with trends pointing toward worst-case flash program latencies on par with disk writes. We extend a conventional flash translation layer to schedule flash program operations to flash pages based on the operations' performance needs and the pages' performance characteristics. We then develop policies to improve in two scenarios: First, we improve peak performance for latency-critical operations of short bursts of intensive activity by 36%. Second, we realize steady-state bandwidth improvements of up to 95% by rate-matching garbage collection performance and external access performance.<p>Pre-2018 CSE ID: CS2013-0996</p> https://escholarship.org/uc/item/49297740 Tue, 14 Mar 2023 00:00:00 +0000 Grupp, Laura Davis, John D Swanson, Steven https://escholarship.org/uc/item/47h846gc We observe that Monte Carlo (SPICE) simulation provides the most accurate and trustable statistical timing analysis, while the existing SSTA method has completely ignored the effect of input statistics on chip timing performance, and provides either accurate estimate nor pessimistic bound of the actual chip timing performance statistics. We propose signal probability (i.e., the logic one occurrence probability for a signal) based statistical timing analysis for improved accuracy and reduced pessimism over the existing SSTA methods, and improved efficiency over Monte Carlo (SPICE) simulation. Our experimental results show that our proposed SPSTA computes mean (standard deviation) of signal arrival times within 6.2% (18.6%), while SSTA computes mean (standard deviation) of signal arrival times within 13.40% (64.3%) of Monte Carlo simulation results; SPSTA also provides signal probaiblity estimation within 14.28% of Monte Carlo simulation results for the ISCAS'89 benchmark circuits. <p>Pre-2018... https://escholarship.org/uc/item/47h846gc Tue, 14 Mar 2023 00:00:00 +0000 Liu, Bao https://escholarship.org/uc/item/46d93544 A basic problem in distributed computing has to do with assigning unique labels --- that is, names or addresses --- to network elements. Some approaches to solving this problem include using static assignment (e.g., MAC addresses), or using a centralized authority (e.g., DHCP). In this paper, we present an approach that is suitable for dynamic environments: where the rules constraining the label choices depend on the network topology, which in turn can change. This problem arose in the context of automatic address assignment in large-scale data center networks, and so we consider issues such as the scalability of message load and convergence time. We give a new algorithm, called the Decider/Chooser Protocol, and show its use in the assignment of labels in data center networks. We evaluate the correctness of the Decider/Chooser Protocol through proofs and model checking, and explore its performance via mathematical analysis and simulation. Through this evaluation, we find that... https://escholarship.org/uc/item/46d93544 Tue, 14 Mar 2023 00:00:00 +0000 Walraed-Sullivan, Meg Niranjan Mysore, Radhika Marzullo, Keith Vahdat, Amin https://escholarship.org/uc/item/4272w41n Availability of slide scanners and electronic medical record systems have led to increased digitization of pathology images. With digitization, Computer-Aided Diagnosis (CAD) tools can be built to reduce pathologist's fatigue and improve diagnosis workflow. As a first step towards building such tools, we developed a cancer detector for prostate needle core biopsy images. The detector is trained using boosting. We use color and texture features. We also built new structural features that express visual cues not captured by color and texture features. These structural features inform the classifier whether cancer has deformed the glands, the main functional unit of prostate. On the test images we scored, we got an Area Under the ROC Curve (AUC) of 0.94.<p>Pre-2018 CSE ID: CS2012-0987</p> https://escholarship.org/uc/item/4272w41n Tue, 14 Mar 2023 00:00:00 +0000 Kabra, Mayank Baird, Stephen Mahooti, Sepi Kim, Jihoon Ohno-Machado, Lucila Freund, Yoav https://escholarship.org/uc/item/4210s00j Resource allocation continues to be one of the primary challenges in federated computing systems. Currently, users are forced to either suffer the delays inherent to batch scheduling used by the supercomputing community or the tragedy of the commons that befalls propotional share allocation as employed on PlanetLab. Market-based schemes have repeatedly been proposed as a possible solution but none have yet to see wide deployment. We present our initial experience with two operational auction-based schedulers---one for PlanetLab, the other for a large SensorNet testbed---and propose two key mechanisms to combat the challenges faced by real-world use of economic schedulers. Our experience shows that some users are unwilling or unable to accept the uncertainty of an auction; hence, we develop a buy-it-now mechanism that allows risk-averse users to instantly acquire resources at a price premium. Further, we describe how intelligent monetary policy, in particular the judicious use... https://escholarship.org/uc/item/4210s00j Tue, 14 Mar 2023 00:00:00 +0000 AuYoung, Alvin Chun, Brent Ng, Chaki Parkes, David Vahdat, Amin Snoeren, Alex C