224 research outputs found
Managing death
No full textDrawing on provocative case studies, personal interviews, and detailed research, James Hoefler examines the medical, legal, ethical, and clinical aspects of right-to-die issues. Beginning with the legal struggle of a woman whose son existed in a persistent vegetative state (PVS) for seventeen years, the author moves into a broader look at consensus among professional organizations, from the AMA to the President's Commission to the National Center for State Courts; beliefs of mainstream religious groups; public opinion; issues surrounding end-stage Alzheimer's and other organic brain disorders that can slowly lead to PVS; and the role of artificial nutrition and hydration in these casesHoefler concludes with recommendations on how to improve the quality of right-to-die decisionmaking. An absorbing read with a minimum of technical jargon, this book is a valuable guide to care givers, public policy students, medical ethicists, family members, and anyone facing questions about an individual's right to di
Scaling betweenness centrality using communication-efficient sparse matrix multiplication
No full textBetweenness centrality (BC) is a crucial graph problem that measures the significance of a vertex by the number of shortest paths
leading through it. We propose Maximal Frontier Betweenness Centrality (MFBC): a succinct BC algorithm based on novel sparse matrix multiplication routines that performs a factor of p 1/3 less communication on p processors than the best known alternatives, for
graphs withn vertices and average degree k = n/p 2/3. We formulate, implement, and prove the correctness of MFBC for weighted graphs by leveraging monoids instead of semirings, which enables a surprisingly succinct formulation. MFBC scales well for both extremely sparse and relatively dense graphs. It automatically searches a space of distributed data decompositions and sparse matrix multiplication algorithms for the most advantageous configuration. The MFBC implementation outperforms the well-known CombBLAS library by up to 8x and shows more robust performance. Our design methodology is readily extensible to other graph problems
A RDMA Interface for Ultra-Fast Ultrasound Data-Streaming over an Optical Link
No full textDigital ultrasound (US) probes integrate the analog-to-digital conversion directly on the probe and can be conveniently connected to commodity devices. Existing digital probes are however limited to a relatively small number of channels, do not guarantee access to the raw US data, or cannot operate at very high frame rates (e.g., due to exhaustion of computing and storage units on the receiving device). In this work, we present an open, compact, power-efficient, 192-channels digital US data acquisition system capable of streaming US data at transfer rates greater than 80 Gbps towards a host PC for ultra-high frame rate imaging (in the multi-kHz range). Our US probe is equipped with two power-efficient Field Programmable Gate Arrays (FPGAs) and is interfaced to the host PC with two optical-link 100G Ethernet connections. The high-speed performance is enabled by implementing a Remote Direct Memory Access (RDMA) communication protocol between the probe and the controlling PC, that utilizes a high-performance Non-Volatile Memory Express (NVMe) interface to store the streamed data. To the best of our knowledge, thanks to the achieved datarates, this is the first high-channel-count compact digital US platform capable of raw data streaming at frame rates of 20 kHz (for imaging at 3.5 cm depths), without the need for sparse sampling, consuming less than 40 W
Contribution of melanocortin 1 receptor gene variants to cutaneous sporadic melanoma risk in an Italian population
No full textStiffness Enhancement of Rotationally Moulded Products
No full textThe demand for high-performance solutions for rotational moulding (RM) of plastics is steadily increasing. However, the materials currently available are significantly limiting the feasibility and design of products that can be manufactured. There are substantial challenges for overcoming these limitations and finding new materials or composites that can enhance the performance of products and subsequently expand the field of application for rotational moulding. This research project focused on closing gaps in the existing knowledge of reinforced plastics in the RM industry, and on the development of a new composite that provides enhanced stiffness and impact strength, without forfeiting other properties. The characteristics of RM, typically using low-cost raw materials and equipment, often impede performance increases as they usually are accompanied by significantly increased material costs or machine adaptations. Therefore, investigations were focused on composites which are easy to manufacture and can be used on the majority of existing RM machines in the industry. Additional processing methods that introduce shear and pressure to the composite during initial blending were studied for their ability to enhance filler-matrix bonding compared to dry-blended composites. The biggest obstacle in this additional processing is achieving the correct post-preparation of the material. This is required to produce particles with proper melt and flow behaviour that is essential to produce an excellent end product. Natural nano-sized clay, halloysite nanotubes (HNT), was incorporated into three different polyethylenes (PE) to develop a composite based on an abundant and renewable resource. The addition of HNT led to an increase in Young’s modulus, flexural modulus and flexural strength, but a loss in impact strength of PE. Another approach was the utilisation of micron-sized and nano-sized synthetic reinforcements, such as ultra-thin glass fibres (UTGF) and carbon nanotubes (CNT), which have previously not been widely considered in the field of RM. The influence of aspect ratios of various reinforcement materials and the benefits of their higher values were investigated. The use of a compounded UTGF/HDPE masterbatch resulted in significantly increased impact strength and Young’s modulus, which were only accompanied by a small decrease in flexural modulus. Low weight fractions of CNT were able to enhance the overall performance of HDPE (improved tensile, flexural and impact properties), making it a desirable candidate for further, industrial-scale manufacturing. A low-cost alternative which requires less processing, is glass fibre powder with an aspect ratio of 12:1; offering 53% increased Young’s modulus compared to the pure matrix material. However, this glass fibre powder has reduced flexural and impact properties. Theoretical models were evaluated in terms of their accuracy and ability to predict Young’s modulus and tensile strength of the composites investigated in this research. Limitations of these observed models include assumptions of perfect fibre-matrix bonding, lack of defects, and the high influence of aspect ratio on the predicted values. The latter is the most substantial with very high aspect ratios as seen in UTGF and CNT, where models led to vastly overpredicted and impossible values. Considering the advances in performance of certain composites used in this research - especially with synthetic reinforcements - their cost-effectiveness and possible implementation in most existing industrial RM machines may be beneficial. These composites have the potential to expand the applications for RM products and give manufacturers a range of new material options without having to upgrade or adapt current RM machinery and technology. Additionally, such composites may allow for the manufacture of larger products with RM, without major design changes or the introduction of stiffening elements
Network-accelerated non-contiguous memory transfers
No full textApplications often communicate data that is non-contiguous in the send- or the receive-buffer, e.g., when exchanging a column of a matrix stored in row-major order. While non-contiguous transfers are well supported in HPC (e.g., MPI derived datatypes), they can still be up to 5x slower than contiguous transfers of the same size. As we enter the era of network acceleration, we need to investigate which tasks to offload to the NIC: In this work we argue that non-contiguous memory transfers can be transparently network-accelerated, truly achieving zero-copy communications. We implement and extend sPIN, a packet streaming processor, within a Portals 4 NIC SST model, and evaluate strategies for NIC-offloaded processing of MPI datatypes, ranging from datatype-specific handlers to general solutions for any MPI datatype. We demonstrate up to 8x speedup in the unpack throughput of real applications, demonstrating that non-contiguous memory transfers are a first-class candidate for network acceleration
The role of the pathologist in tissue banking: European Consensus Expert Group Report
No full textHuman tissue biobanking encompasses a wide range of activities and study designs and is critical for application of a wide range of new technologies (-"omics") to the discovery of molecular patterns of disease and for implementation of novel biomarkers into clinical trials. Pathology is the cornerstone of hospital-based tissue biobanking. Pathologists not only provide essential information identifying the specimen but also make decisions on what should be biobanked, making sure that the timing of all operations is consistent with both the requirements of clinical diagnosis and the optimal preservation of biological products. This document summarizes the conclusions of a Pathology Expert Group Meeting within the European Biological and Biomolecular Research Infrastructure (BBMRI) Program. These recommendations are aimed at providing guidance for pathologists as well as for institutions hosting biobanks on how to better integrate and support pathological activities within the framework of biobanks that fulfill international standards
A comparison of pathomolecular markers of fibrosis and morphology in kidney from autopsies of African Americans and whites
No full textAfrican Americans have an increased incidence of chronic kidney disease (CKD) due to hypertension and arteriosclerosis and increased death due to coronary artery disease, compared with whites. The pathogenesis of CKD involves the increased presence and activation of myofibroblasts and macrophages, promotion of tubulointerstitial fibrosis, and effects of tubulointerstitial cell mitosis and apoptosis. We hypothesized that increased risk of hypertensive vascular disease may be identified by renal pathomolecular markers that are associated with progressive CKD. Renal sections were available from 50 autopsies of 33 African Americans (55% males) and 17 whites (76% males) undergoing forensic autopsy for unexpected death. Sclerotic glomeruli, severity of cortical fibrosis, and renal arterioloselerosis, total glomerular number (N-glom), average glomerular volume (V-glom), birth weights, and blood pressure were known. Presence and locality of markers for myofibroblasts (alpha-SMA), macrophages (CD68), collagen, pro-fibrotic transforming growth factor-beta1 were scored in renal autopsies, and tubulointerstitial apoptosis was recorded. The results demonstrated a strong positive correlation between age, cortical fibrosis and alpha-SMA (p < 0.05), and between CD68 and hypertension and coronary artery disease (p < 0.05). The findings confirm the role of myofibroblasts and macrophages in pathogenesis of human CKD. However, the markers showed no significant relationships to V-glom, N-glom, birth weight, or race
Describing typeforms: a designer's response
No full textThe paper sets out an overview of a pragmatic research investigation initiated within a doctoral enquiry, and which continues to inform design practice and pedagogy. Located within the fields of typography and information design, and very much concerned with design history, enquiry emphasized exploration of alternative design research methodologies in the production of a design outcome loaded with pedagogical ambition.
The issue being addressed within the investigation was the limited scope of existing typeface classificatory systems to adequately describe the diversity of forms represented within current type design practice and thus, recent acquisitions to an established teaching collection in London.
Addressing this issue unexpectedly came to utilize the researcher’s own design practice as a methodology for managing emergent enquiry, and for organizing and generating new knowledge through the employment of visual information management methods.
A primary outcome of the enquiry was a new framework for the description of typeforms. This new framework will be described in terms of its operation, divergence from existing models and potential for application
Near-Optimal Sparse Allreduce for Distributed Deep Learning
No full textCommunication overhead is one of the major obstacles to train large deep
learning models at scale. Gradient sparsification is a promising technique to
reduce the communication volume. However, it is very challenging to obtain real
performance improvement because of (1) the difficulty of achieving an scalable
and efficient sparse allreduce algorithm and (2) the sparsification overhead.
This paper proposes O-Top, a scheme for distributed training with sparse
gradients. O-Top integrates a novel sparse allreduce algorithm (less than
6 communication volume which is asymptotically optimal) with the
decentralized parallel Stochastic Gradient Descent (SGD) optimizer, and its
convergence is proved. To reduce the sparsification overhead, O-Top
efficiently selects the top- gradient values according to an estimated
threshold. Evaluations are conducted on the Piz Daint supercomputer with neural
network models from different deep learning domains. Empirical results show
that O-Top achieves similar model accuracy to dense allreduce. Compared
with the optimized dense and the state-of-the-art sparse allreduces,
O-Top is more scalable and significantly improves training throughput
(e.g., 3.29x-12.95x improvement for BERT on 256 GPUs).Comment: Published in Proceedings of the 27th ACM SIGPLAN Symposium on
Principles and Practice of Parallel Programming (PPoPP'22), April 2-6, 2022,
Pages 135-149, https://doi.org/10.1145/3503221.350839
- …
