Other affiliations: University of Michigan, Eindhoven University of Technology, University of Texas at Austin ...read more
Bio: Onur Mutlu is an academic researcher from ETH Zurich. The author has contributed to research in topics: Dram & Computer science. The author has an hindex of 103, co-authored 543 publications receiving 34806 citations. Previous affiliations of Onur Mutlu include University of Michigan & Eindhoven University of Technology.
Papers published on a yearly basis
••20 Jun 2009
TL;DR: This work proposes, crafted from a fundamental understanding of PCM technology parameters, area-neutral architectural enhancements that address these limitations and make PCM competitive with DRAM.
Abstract: Memory scaling is in jeopardy as charge storage and sensing mechanisms become less reliable for prevalent memory technologies, such as DRAM. In contrast, phase change memory (PCM) storage relies on scalable current and thermal mechanisms. To exploit PCM's scalability as a DRAM alternative, PCM must be architected to address relatively long latencies, high energy writes, and finite endurance.We propose, crafted from a fundamental understanding of PCM technology parameters, area-neutral architectural enhancements that address these limitations and make PCM competitive with DRAM. A baseline PCM system is 1.6x slower and requires 2.2x more energy than a DRAM system. Buffer reorganizations reduce this delay and energy gap to 1.2x and 1.0x, using narrow rows to mitigate write energy and multiple rows to improve locality and write coalescing. Partial writes enhance memory endurance, providing 5.6 years of lifetime. Process scaling will further reduce PCM energy costs and improve endurance.
14 Jun 2014
TL;DR: This paper exposes the vulnerability of commodity DRAM chips to disturbance errors, and shows that it is possible to corrupt data in nearby addresses by reading from the same address in DRAM by activating the same row inDRAM.
Abstract: Memory isolation is a key property of a reliable and secure computing system--an access to one memory address should not have unintended side effects on data stored in other addresses. However, as DRAM process technology scales down to smaller dimensions, it becomes more difficult to prevent DRAM cells from electrically interacting with each other. In this paper, we expose the vulnerability of commodity DRAM chips to disturbance errors. By reading from the same address in DRAM, we show that it is possible to corrupt data in nearby addresses. More specifically, activating the same row in DRAM corrupts data in nearby rows. We demonstrate this phenomenon on Intel and AMD systems using a malicious program that generates many DRAM accesses. We induce errors in most DRAM modules (110 out of 129) from three major DRAM manufacturers. From this we conclude that many deployed systems are likely to be at risk. We identify the root cause of disturbance errors as the repeated toggling of a DRAM row's wordline, which stresses inter-cell coupling effects that accelerate charge leakage from nearby rows. We provide an extensive characterization study of disturbance errors and their behavior using an FPGA-based testing platform. Among our key findings, we show that (i) it takes as few as 139K accesses to induce an error and (ii) up to one in every 1.7K cells is susceptible to errors. After examining various potential ways of addressing the problem, we propose a low-overhead solution to prevent the errors
TL;DR: An algorithm (mrFAST) is presented to comprehensively map next-generation sequence reads, which allows for the prediction of absolute copy-number variation of duplicated segments and genes, and can distinguish between different copies of highly identical genes.
Abstract: Despite their importance in gene innovation and phenotypic variation, duplicated regions have remained largely intractable owing to difficulties in accurately resolving their structure, copy number and sequence content. We present an algorithm (mrFAST) to comprehensively map next-generation sequence reads, which allows for the prediction of absolute copy-number variation of duplicated segments and genes. We examine three human genomes and experimentally validate genome-wide copy number differences. We estimate that, on average, 73-87 genes vary in copy number between any two individuals and find that these genic differences overwhelmingly correspond to segmental duplications (odds ratio = 135; P < 2.2 x 10(-16)). Our method can distinguish between different copies of highly identical genes, providing a more accurate assessment of gene content and insight into functional constraint without the limitations of array-based technology.
••13 Jun 2015
TL;DR: This work argues that the conventional concept of processing-in-memory (PIM) can be a viable solution to achieve memory-capacity-proportional performance and designs a programmable PIM accelerator for large-scale graph processing called Tesseract.
Abstract: The explosion of digital data and the ever-growing need for fast data analysis have made in-memory big-data processing in computer systems increasingly important. In particular, large-scale graph processing is gaining attention due to its broad applicability from social science to machine learning. However, scalable hardware design that can efficiently process large graphs in main memory is still an open problem. Ideally, cost-effective and scalable graph processing systems can be realized by building a system whose performance increases proportionally with the sizes of graphs that can be stored in the system, which is extremely challenging in conventional systems due to severe memory bandwidth limitations. In this work, we argue that the conventional concept of processing-in-memory (PIM) can be a viable solution to achieve such an objective. The key modern enabler for PIM is the recent advancement of the 3D integration technology that facilitates stacking logic and memory dies in a single package, which was not available when the PIM concept was originally examined. In order to take advantage of such a new technology to enable memory-capacity-proportional performance, we design a programmable PIM accelerator for large-scale graph processing called Tesseract. Tesseract is composed of (1) a new hardware architecture that fully utilizes the available memory bandwidth, (2) an efficient method of communication between different memory partitions, and (3) a programming interface that reflects and exploits the unique hardware design. It also includes two hardware prefetchers specialized for memory access patterns of graph processing, which operate based on the hints provided by our programming model. Our comprehensive evaluations using five state-of-the-art graph processing workloads with large real-world graphs show that the proposed architecture improves average system performance by a factor of ten and achieves 87% average energy reduction over conventional systems.
••01 Dec 2007
TL;DR: This paper proposes a new memory access scheduler, called the Stall-Time Fair Memory scheduler (STFM), that provides quality of service to different threads sharing the DRAM memory system and shows that STFM significantly reduces the unfairness in theDRAM system while also improving system throughput on a wide variety of workloads and systems.
Abstract: DRAM memory is a major resource shared among cores in a chip multiprocessor (CMP) system. Memory requests from different threads can interfere with each other. Existing memory access scheduling techniques try to optimize the overall data throughput obtained from the DRAM and thus do not take into account inter-thread interference. Therefore, different threads running together on the same chip can ex- perience extremely different memory system performance: one thread can experience a severe slowdown or starvation while another is un- fairly prioritized by the memory scheduler. This paper proposes a new memory access scheduler, called the Stall-Time Fair Memory scheduler (STFM), that provides quality of service to different threads sharing the DRAM memory system. The goal of the proposed scheduler is to "equalize" the DRAM-related slowdown experienced by each thread due to interference from other threads, without hurting overall system performance. As such, STFM takes into account inherent memory characteristics of each thread and does not unfairly penalize threads that use the DRAM system without interfering with other threads. We show that STFM significantly reduces the unfairness in the DRAM system while also improving system throughput (i.e., weighted speedup of threads) on a wide variety of workloads and systems. For example, averaged over 32 different workloads running on an 8-core CMP, the ratio between the highest DRAM-related slowdown and the lowest DRAM-related slowdown reduces from 5.26X to 1.4X, while the average system throughput improves by 7.6%. We qualitatively and quantitatively compare STFM to one new and three previously- proposed memory access scheduling algorithms, including network fair queueing. Our results show that STFM provides the best fairness, system throughput, and scalability.
TL;DR: This book by a teacher of statistics (as well as a consultant for "experimenters") is a comprehensive study of the philosophical background for the statistical design of experiment.
Abstract: THE DESIGN AND ANALYSIS OF EXPERIMENTS. By Oscar Kempthorne. New York, John Wiley and Sons, Inc., 1952. 631 pp. $8.50. This book by a teacher of statistics (as well as a consultant for \"experimenters\") is a comprehensive study of the philosophical background for the statistical design of experiment. It is necessary to have some facility with algebraic notation and manipulation to be able to use the volume intelligently. The problems are presented from the theoretical point of view, without such practical examples as would be helpful for those not acquainted with mathematics. The mathematical justification for the techniques is given. As a somewhat advanced treatment of the design and analysis of experiments, this volume will be interesting and helpful for many who approach statistics theoretically as well as practically. With emphasis on the \"why,\" and with description given broadly, the author relates the subject matter to the general theory of statistics and to the general problem of experimental inference. MARGARET J. ROBERTSON
TL;DR: Machine learning addresses many of the same research questions as the fields of statistics, data mining, and psychology, but with differences of emphasis.
Abstract: Machine Learning is the study of methods for programming computers to learn. Computers are applied to a wide range of tasks, and for most of these it is relatively easy for programmers to design and implement the necessary software. However, there are many tasks for which this is difficult or impossible. These can be divided into four general categories. First, there are problems for which there exist no human experts. For example, in modern automated manufacturing facilities, there is a need to predict machine failures before they occur by analyzing sensor readings. Because the machines are new, there are no human experts who can be interviewed by a programmer to provide the knowledge necessary to build a computer system. A machine learning system can study recorded data and subsequent machine failures and learn prediction rules. Second, there are problems where human experts exist, but where they are unable to explain their expertise. This is the case in many perceptual tasks, such as speech recognition, hand-writing recognition, and natural language understanding. Virtually all humans exhibit expert-level abilities on these tasks, but none of them can describe the detailed steps that they follow as they perform them. Fortunately, humans can provide machines with examples of the inputs and correct outputs for these tasks, so machine learning algorithms can learn to map the inputs to the outputs. Third, there are problems where phenomena are changing rapidly. In finance, for example, people would like to predict the future behavior of the stock market, of consumer purchases, or of exchange rates. These behaviors change frequently, so that even if a programmer could construct a good predictive computer program, it would need to be rewritten frequently. A learning program can relieve the programmer of this burden by constantly modifying and tuning a set of learned prediction rules. Fourth, there are applications that need to be customized for each computer user separately. Consider, for example, a program to filter unwanted electronic mail messages. Different users will need different filters. It is unreasonable to expect each user to program his or her own rules, and it is infeasible to provide every user with a software engineer to keep the rules up-to-date. A machine learning system can learn which mail messages the user rejects and maintain the filtering rules automatically. Machine learning addresses many of the same research questions as the fields of statistics, data mining, and psychology, but with differences of emphasis. Statistics focuses on understanding the phenomena that have generated the data, often with the goal of testing different hypotheses about those phenomena. Data mining seeks to find patterns in the data that are understandable by people. Psychological studies of human learning aspire to understand the mechanisms underlying the various learning behaviors exhibited by people (concept learning, skill acquisition, strategy change, etc.).
01 Aug 2000
TL;DR: Assessment of medical technology in the context of commercialization with Bioentrepreneur course, which addresses many issues unique to biomedical products.
Abstract: BIOE 402. Medical Technology Assessment. 2 or 3 hours. Bioentrepreneur course. Assessment of medical technology in the context of commercialization. Objectives, competition, market share, funding, pricing, manufacturing, growth, and intellectual property; many issues unique to biomedical products. Course Information: 2 undergraduate hours. 3 graduate hours. Prerequisite(s): Junior standing or above and consent of the instructor.
TL;DR: An analysis tool for the detection of somatic mutations and copy number alterations in exome data from tumor-normal pairs is presented and new light is shed on the landscape of genetic alterations in ovarian cancer.
Abstract: Exome sequencing of tumor samples and matched normal controls has the potential to rapidly identify protein-altering mutations across hundreds of patients, potentially enabling the discovery of recurrent events driving tumor development and growth (International Cancer Genome Consortium 2010; Stratton 2011). Yet the analysis of such data presents significant challenges. Sequencing coverage is nonuniform across targeted regions and from one sample to the next (Ng et al. 2009; Bainbridge et al. 2010; Teer et al. 2010). Many regions achieve high read depth (more than 100×), which can confound variant callers and depth-based filters if not properly addressed (Ku et al. 2011). Repetitive and paralogous sequences can give rise to numerous false positives. The detection of somatic mutations in tumor genomes is even more challenging. The genomes of primary tumors are genetically heterogeneous (Ding et al. 2010), with frequent rearrangements (Campbell et al. 2008) and copy number alterations (CNAs) (Beroukhim et al. 2010). Further, somatic mutations are relatively rare compared with germline variation, often representing <0.1% of variants in a tumor genome (Ley et al. 2008; Mardis et al. 2009). Simply subtracting variants in the matched normal from variants in the tumor (Wei et al. 2011) is poorly suited for the analysis of exome sequence data, because it fails to account for regions that were undersampled in the normal. Accurate mutation detection requires a direct, simultaneous comparison of tumor–normal pairs at every position in the exome, but few algorithms to do so have been described. Numerous algorithms have been developed to assess genome-wide copy number using whole-genome sequencing (WGS) data. Most of these approaches (Campbell et al. 2008; Alkan et al. 2009; Chiang et al. 2009; Yoon et al. 2009; Abyzov et al. 2011) would be confounded by exome data sets, because of the biases introduced by hybridization and the sparse and uneven coverages throughout the genome. However, when both DNA samples in a tumor–normal pair were captured and sequenced under identical hybridization conditions, we reasoned that it might be possible to detect somatic CNAs (SCNAs) as deviations from the log-ratio of sequence coverage depth within a tumor–normal pair, and then quantify the deviations statistically. Such an approach would provide a gene-centric view of copy number in a tumor sample, though it would be limited to the ∼1% of the genome captured by current exome platforms. Previously, we published VarScan (Koboldt et al. 2009), an algorithm for variant detection in next-generation sequencing data. We have since released a new tool, VarScan 2 (http://varscan.sourceforge.net), with several improvements, including the ability to identify somatic mutation, loss of heterozygosity (LOH), and CNA events in tumor–normal pairs. VarScan 2 analyzes sequence data from a tumor sample and its corresponding normal sample simultaneously, applying heuristic methods and a statistical test to detect variants—single nucleotide variants (SNVs) and insertions/deletions (indels)—and classify them by somatic status. By direct comparison of normalized sequence depth, our method also detects SCNAs in the tumor genome. Here, we utilize VarScan 2 for the analysis of exome sequence data from 151 patients with high-grade serous ovarian adenocarcinoma (HGS-OVCa) that were initially characterized within the Cancer Genome Atlas (TCGA) project (Cancer Genome Atlas Research Network 2011). We present a robust pipeline for the detection of both germline (inherited) and somatic (acquired) mutations by exome sequencing and describe filtering approaches for detecting variants with high sensitivity and specificity. To evaluate the performance of our SCNA detection algorithm, we compare our results to copy number data from high-density SNP array and WGS approaches. Our results demonstrate the accuracy of VarScan 2 for somatic mutation and CNA detection and enable a new survey of the genetic landscape in ovarian carcinoma.
Max Planck Society1, Broad Institute2, University of California, Berkeley3, European Bioinformatics Institute4, National Institutes of Health5, University of Massachusetts Medical School6, University of Washington7, Spanish National Research Council8, University of Montana9, Croatian Academy of Sciences and Arts10, University of Oviedo11, University of Bonn12, Emory University13, University College Cork14, Harvard University15
TL;DR: The genomic data suggest that Neandertals mixed with modern human ancestors some 120,000 years ago, leaving traces of Ne andertal DNA in contemporary humans, suggesting that gene flow from Neand Bertals into the ancestors of non-Africans occurred before the divergence of Eurasian groups from each other.
Abstract: Neandertals, the closest evolutionary relatives of present-day humans, lived in large parts of Europe and western Asia before disappearing 30,000 years ago. We present a draft sequence of the Neandertal genome composed of more than 4 billion nucleotides from three individuals. Comparisons of the Neandertal genome to the genomes of five present-day humans from different parts of the world identify a number of genomic regions that may have been affected by positive selection in ancestral modern humans, including genes involved in metabolism and in cognitive and skeletal development. We show that Neandertals shared more genetic variants with present-day humans in Eurasia than with present-day humans in sub-Saharan Africa, suggesting that gene flow from Neandertals into the ancestors of non-Africans occurred before the divergence of Eurasian groups from each other.