Search

We investigate the OpenMP parallelization and optimizationof two novel data classification algorithms. The new algorithms arebased on graph and PDE solution techniques and provide significant accuracy and performance advantages over traditional data classificationalgorithms in serial mode. The methods leverage the Nystrom extensionto calculate eigenvalue/eigenvectors of the graph Laplacian and thisis a self-contained module that can be used in conjunction with othergraph-Laplacian based methods such as spectral clustering. We use performancetools to collect the hotspots and memory access of the serialcodes and use OpenMP as the parallelization language to parallelize themost time-consuming parts. Where possible, we also use library routines.We then optimize the OpenMP implementations and detail the performanceon traditional supercomputer nodes (in our case a Cray XC30),and test the optimization steps on emerging testbed systems based on Intel’sKnights Corner and Landing processors. We show both performanceimprovement and strong scaling behavior. A large number of optimizationtechniques and analyses are necessary before the algorithm reachesalmost ideal scaling.

The Cori system at NERSC has two computepartitions with different CPU architectures: a 2,004 nodeHaswell partition and a 9,688 node KNL partition, whichranked as the 5th most powerful and fastest supercomputeron the November 2016 Top 500 list. The compute partitionsshare a common storage configuration, and understanding theIO performance gap between them is important, impactingnot only to NERSC/LBNL users and other national labs, butalso to the relevant hardware vendors and software developers.In this paper, we have analyzed performance of single coreand single node IO comprehensively on the Haswell and KNLpartitions, and have discovered the major bottlenecks, whichinclude CPU frequencies and memory copy performance. Wehave also extended our performance tests to multi-node IOand revealed the IO cost difference caused by network latency,buffer size, and communication cost. Overall, we have developeda strong understanding of the IO gap between Haswell and KNLnodes and the lessons learned from this exploration will guideus in designing optimal IO solutions in many-core era.

We investigate the OpenMP parallelization and optimizationof two novel data classification algorithms. The new algorithms arebased on graph and PDE solution techniques and provide significant accuracy and performance advantages over traditional data classificationalgorithms in serial mode. The methods leverage the Nystrom extensionto calculate eigenvalue/eigenvectors of the graph Laplacian and thisis a self-contained module that can be used in conjunction with othergraph-Laplacian based methods such as spectral clustering. We use performancetools to collect the hotspots and memory access of the serialcodes and use OpenMP as the parallelization language to parallelize themost time-consuming parts. Where possible, we also use library routines.We then optimize the OpenMP implementations and detail the performanceon traditional supercomputer nodes (in our case a Cray XC30),and test the optimization steps on emerging testbed systems based on Intel’sKnights Corner and Landing processors. We show both performanceimprovement and strong scaling behavior. A large number of optimizationtechniques and analyses are necessary before the algorithm reachesalmost ideal scaling.

There are many potential issues associated with deploying the Intel Xeon PhiTM (code named Knights Landing [KNL]) manycore processor in a large-scale supercomputer. One in particular is the ability to fully utilize the high-speed communications network, given that the serial performance of a Xeon PhiTM core is a fraction of a Xeon®core. In this paper, we take a look at the trade-offs associated with allocating enough cores to fully utilize the Aries high-speed network versus cores dedicated to computation, eg, the trade-off between MPI and OpenMP. In addition, we evaluate new features of Cray MPI in support of KNL, such as internode optimizations. We also evaluate one-sided programming models such as Unified Parallel C. We quantify the impact of the above trade-offs and features using a suite of National Energy Research Scientific Computing Center applications.

The newest NERSC supercomputer Cori is a Cray XC40 system consisting of 2,388 Intel Xeon Haswell nodes and 9,688 Intel Xeon-Phi “Knights Landing” (KNL) nodes. Compared to the Xeon-based clusters NERSC users are familiar with, optimal performance on Cori requires consideration of KNL mode settings; process, thread, and memory affinity; fine-grain parallelization; vectorization; and use of the high-bandwidth MCDRAM memory. This paper describes our efforts preparing NERSC users for KNL through the NERSC Exascale Science Application Program, Web documentation, and user training. We discuss how we configured the Cori system for usability and productivity, addressing programming concerns, batch system configurations, and default KNL cluster and memory modes. System usage data, job completion analysis, programming and running jobs issues, and a few successful user stories on KNL are presented.

De novo genome assembly is one of the most important and challenging computational problems in modern genomics; further, it shares algorithms and communication patterns important to other graph analytic and irregular applications. Unlike simulations, it has no floating point arithmetic and is dominated by small memory transactions within and between computing nodes. In this work, we introduce MerBench, a compact set of PGAS benchmarks that capture the communication patterns of the parallel algorithms throughout HipMer, a parallel genome assembler pipeline that has been shown to scale to massive concurrencies. We also present results of these microbenchmarks on the Edison supercomputer and illustrate how these empirical results can be used to assess the scaling behavior of the pipeline.

The Cori system at NERSC has two computepartitions with different CPU architectures: a 2,004 nodeHaswell partition and a 9,688 node KNL partition, whichranked as the 5th most powerful and fastest supercomputeron the November 2016 Top 500 list. The compute partitionsshare a common storage configuration, and understanding theIO performance gap between them is important, impactingnot only to NERSC/LBNL users and other national labs, butalso to the relevant hardware vendors and software developers.In this paper, we have analyzed performance of single coreand single node IO comprehensively on the Haswell and KNLpartitions, and have discovered the major bottlenecks, whichinclude CPU frequencies and memory copy performance. Wehave also extended our performance tests to multi-node IOand revealed the IO cost difference caused by network latency,buffer size, and communication cost. Overall, we have developeda strong understanding of the IO gap between Haswell and KNLnodes and the lessons learned from this exploration will guideus in designing optimal IO solutions in many-core era.

NERSC has partnered with 20 representative application teams to evaluate performance on the Xeon-Phi Knights Landing architecture and develop an application-optimization strategy for the greater NERSC workload on the recently installed Cori system. In this article, we present early case studies and summarized results from a subset of the 20 applications highlighting the impact of important architecture differences between the Xeon-Phi and traditional Xeon processors. We summarize the status of the applications and describe the greater optimization strategy that has formed.

De novo genome assembly is one of the most important and challenging computational problems in modern genomics; further, it shares algorithms and communication patterns important to other graph analytic and irregular applications. Unlike simulations, it has no floating point arithmetic and is dominated by small memory transactions within and between computing nodes. In this work, we focus on the highly scalable HipMer assembler and identify the dominant algorithms and communication patterns, also using microbenchmarks to capture the workload. We evaluate HipMer on a variety of platforms from the latest HPC systems to ethernet clusters. HipMer performs well on all single node systems, including the Xeon Phi manycore architecture. Given large enough problems, it also demonstrates excellent scaling across nodes in an HPC system, but requires a high speed network with low overhead and high injection rates. Our results shed light on the architectural features that are most important for achieving good parallel efficiency on this and related problems.

Mechanical sensitivity is commonly affected in chronic pain and other neurological disorders. To discover mechanisms of individual differences in punctate mechanosensation, we performed quantitative trait locus (QTL) mapping of the response to von Frey monofilament stimulation in BXD recombinant inbred (BXD) mice. Significant loci were detected on mouse chromosome (Chr) 5 and 15, indicating the location of underlying polymorphisms that cause heritable variation in von Frey response. Convergent evidence from public gene expression data implicates candidate genes within the loci: von Frey thresholds were strongly correlated with baseline expression of Cacna2d1, Ift27 and Csnk1e in multiple brain regions of BXD strains. Systemic gabapentin and PF-670462, which target the protein products of Cacna2d1 and Csnk1e, respectively, significantly increased von Frey thresholds in a genotype-dependent manner in progenitors and BXD strains. Real-time polymerase chain reaction confirmed differential expression of Cacna2d1 and Csnk1e in multiple brain regions in progenitors and showed differential expression of Cacna2d1 and Csnk1e in the dorsal root ganglia of the progenitors and BXD strains grouped by QTL genotype. Thus, linkage mapping, transcript covariance and pharmacological testing suggest that genetic variation affecting Cacna2d1 and Csnk1e may contribute to individual differences in von Frey filament response. This study implicates Cacna2d1 and Ift27 in basal mechanosensation in line with their previously suspected role in mechanical hypersensitivity. Csnk1e is implicated for von Frey response for the first time. Further investigation is warranted to identify the specific polymorphisms involved and assess the relevance of these findings to clinical conditions of disturbed mechanosensation.

We investigate the OpenMP parallelization and optimization of two novel data classification algorithms. The new algorithms are based on graph and PDE solution techniques and provide significant accuracy and performance advantages over traditional data classification algorithms in serial mode. The methods leverage the Nystrom extension to calculate eigenvalue/eigenvectors of the graph Laplacian and this is a self-contained module that can be used in conjunction with other graph- Laplacian based methods such as spectral clustering.We use performance tools to collect the hotspots and memory access of the serial codes and use OpenMP as the parallelization language to parallelize the most timeconsuming parts. Where possible, we also use library routines. We then optimize the OpenMP implementations and detail the performance on traditional supercomputer nodes (in our case a Cray XC30), and test the optimization steps on emerging testbed systems based on Intel’s Knights Corner and Landing processors. We show both performance improvement and strong scaling behavior. A large number of optimization techniques and analyses are necessary before the algorithm reaches almost ideal scaling.

The Roofline Performance Model is a visually intuitive method used to bound the sustained peak floating-point performance of any given arithmetic kernel on any given processor architecture. In the Roofline, performance is nominally measured in floating-point operations per second as a function of arithmetic intensity (operations per byte of data). In this study we determine the Roofline for the Intel Knights Landing (KNL) processor, determining the sustained peak memory bandwidth and floating-point performance for all levels of the memory hierarchy, in all the different KNL cluster modes.We then determine arithmetic intensity and performance for a suite of application kernels being targeted for the KNL based supercomputer Cori, and make comparisons to current Intel Xeon processors. Cori is the National Energy Research Scientific Computing Center’s (NERSC) next generation supercomputer. Scheduled for deployment mid-2016, it will be one of the earliest and largest KNL deployments in the world.

Odalys Torres-Luquis,1,2 Krystal Madden,3 N’sanh MR N’dri,1 Richard Berg,4 Olufunmilayo F Olopade,5 Wilfred Ngwa,6,7 Dafalla Abuidris,8 Suresh Mittal,1,2 Beverly Lyn-Cook,9 Sulma I Mohammed1,2 1Department of Comparative Pathobiology, Purdue University, West Lafayette, IN, USA; 2Purdue University Center for Cancer Research, Purdue University, West Lafayette, IN, USA; 3Department of Learning Sciences, University of Illinois at Chicago, Chicago, IL, USA; 4Department of Surgery, Indiana University Health, Lafayette, IN, USA; 5Center for Clinical Cancer Genetics and Global Health, Department of Medicine, University of Chicago, Chicago, IL, USA; 6Department of Radiation Oncology, Brigham and Women’s Hospital, Boston, MA, USA; 7Department of Radiation Oncology, Harvard Medical School, Boston, MA, USA; 8Department of Oncology, National Cancer Institute, University of Gezira, Gezira, Sudan; 9Division of Biochemical Toxicology, National Center for Toxicological Research, Jefferson, AR, USA Background: Triple-negative breast cancer (TNBC) is more prevalent in African and African American (AA) women compared to European American (EA) women. African and AA women diagnosed with TNBC experience high frequencies of metastases and less favorable outcomes. Emerging evidence indicates that this disparity may in fact be the result of the uniquely aggressive biology of African and AA disease. Purpose: To understand the reasons for TNBC in AA aggressive biology, we designed the present study to examine the proteomic profiles of TNBC and luminal A (LA) breast cancer within and across patients’ racial demographic groups in order to identify proteins or molecular pathways altered in TNBC that offer some explanation for its aggressiveness and potential targets for treatment. Materials and methods: Proteomic profiles of TNBC, LA tumors, and their adjacent normal tissues from AA and EA women were obtained using 2-dimensional gel electrophoresis and bioinformatics, and differentially expressed proteins were validated by Western blot and immunohistochemistry. Our data showed that a number of proteins have significantly altered in expression in LA tumors compared to TNBC, both within and across patients’ racial demographic groups. The differentially overexpressed proteins in TNBC (compared to LA) of AA samples were distinct from those in TNBC (compared to LA) of EA women samples. Among the signaling pathways altered in AA TNBC compared to EA TNBC are innate immune signaling, calpain protease, and pyrimidine de novo synthesis pathways. Furthermore, liver LXR/RXR signaling pathway was altered between LA and TNBC in AA women and may be due to the deficiency of the CYP7B1 enzyme responsible for cholesterol degradation. Conclusion: These findings suggest that TNBC in AA women enriched in signaling pathways that are different from TNBC in EA women. Our study draws a link between LXR/RXR expression, cholesterol, obesity, and the TNBC in AA women.Keywords: breast, triple-negative breast cancer, African American, European American, luminal breast cancer, luminal

Terrestrial ecosystems sequester 2.1 Pg of atmospheric carbon annually. A large amount of the terrestrial sink is realized by forests. However, considerable uncertainties remain regarding the fate of this carbon over both short and long timescales. Relevant data to address these uncertainties are being collected at many sites around the world, but syntheses of these data are still sparse. To facilitate future synthesis activities, we have assembled a comprehensive global database for forest ecosystems, which includes carbon budget variables (fluxes and stocks), ecosystem traits (e.g. leaf area index, age), as well as ancillary site information such as management regime, climate, and soil characteristics. This publicly available database can be used to quantify global, regional or biome-specific carbon budgets; to re-examine established relationships; to test emerging hypotheses about ecosystem functioning [e.g. a constant net ecosystem production (NEP) to gross primary production (GPP) ratio]; and as benchmarks for model evaluations. In this paper, we present the first analysis of this database. We discuss the climatic influences on GPP, net primary production (NPP) and NEP and present the CO2 balances for boreal, temperate, and tropical forest biomes based on micrometeorological, ecophysiological, and biometric flux and inventory estimates. Globally, GPP of forests benefited from higher temperatures and precipitation whereas NPP saturated above either a threshold of 1500 mm precipitation or a mean annual temperature of 10 °C. The global pattern in NEP was insensitive to climate and is hypothesized to be mainly determined by nonclimatic conditions such as successional stage, management, site history, and site disturbance. In all biomes, closing the CO2 balance required the introduction of substantial biome-specific closure terms. Nonclosure was taken as an indication that respiratory processes, advection, and non-CO2 carbon fluxes are not presently being adequately accounted for. © 2007 Blackwell Publishing Ltd.