GPU Domain Specialization via Composable On-Package Architecture

As GPUs scale their low-precision matrix math throughput to boost deep learning (DL) performance, they upset the balance between math throughput and memory system capabilities. We demonstrate that a converged GPU design trying to address diverging architectural requirements between FP32 (or larger)-based HPC and FP16 (or smaller)-based DL workloads results in sub-optimal configurations for either of the application domains. We argue that a C omposable O n- PA ckage GPU (COPA-GPU) architecture to provide domain-specialized GPU products is the most practical solution to these diverging requirements. A COPA-GPU leverages multi-chip-module disaggregation to support maximal design reuse, along with memory system specialization per application domain. We show how a COPA-GPU enables DL-specialized products by modular augmentation of the baseline GPU architecture with up to 4× higher off-die bandwidth, 32× larger on-package cache, and 2.3× higher DRAM bandwidth and capacity, while conveniently supporting scaled-down HPC-oriented designs. This work explores the microarchitectural design necessary to enable composable GPUs and evaluates the benefits composability can provide to HPC, DL training, and DL inference. We show that when compared to a converged GPU design, a DL-optimized COPA-GPU featuring a combination of 16× larger cache capacity and 1.6× higher DRAM bandwidth scales per-GPU training and inference performance by 31% and 35%, respectively, and reduces the number of GPU instances by 50% in scale-out training scenarios.

Analysis of the folded normal distribution of a random variable

Анализ поведения случайных величин после различных преобразований можно применять при решении многих нетривиальных задач. В частности, решения, которые невозможно выразить аналитически, с точки зрения практической применимости способны давать результаты с точностью, достаточной для вычислений, вынося невыразимую невязку аналитического решения далеко за рамки требуемой погрешности. В настоящей работе исследовано поведение модуля нормально распределенной случайной величины и выяснено, при каких условиях можно пренебречь операцией взятия абсолютного значения и аппроксимировать модуль случайной величины {\it похожим} распределением вероятностей. The analysis of the behavior of random variables after various transformations can be used in the practical solution of many non-trivial problems. In particular, solutions that cannot be expressed purely analytically, from the point of view of practical applicability, are able to give results with accuracy sufficient for real calculations, taking the inexpressible discrepancy of the analytical solution far beyond the actual error. In this paper, the behavior of the modulus of a normally distributed random variable is investigated and it is found out under what conditions it is possible to neglect the operation of taking an absolute value and approximate the modulus of a random variable with a {\it similar} probability distribution.

Evaluating Reliability of Google Scholar, Scopus, and Web of Science: A Study of Faculty in U.S. Advertising and Public Relations Programs

Reliability of Google Scholar (GS), Scopus, and Web of Science (WoS) is examined using publications and citations of 186 scholars in 14 U.S. advertising and public relations (ADPR) programs. Career duration is controlled, and an integrated impact (II) index is proposed as a practical solution. Results suggest there are trade-offs between the uncertainty of GS’s search parameters for more inclusive coverage and the curated collections of Scopus and WoS that might undercount some influential authors or works. To further demonstrate the discipline’s value, we must pay more attention to rigor and accuracy of methods that will lead to improved outcomes.

Path Planning for Bulldozers with Curvature Constraints

There is presently no solution to the problem of an autonomous bulldozer pushing mounds of material to desired goal locations in the presence of obstacles whilst obeying the kinematic constraints of the bulldozer. Past work has solved some aspects of this problem, but not all. This research presents the first complete, practical solution to the problem. It works by creating a fixed RRT in advance, and then during operation connecting pushing poses into this RRT using Bezier curves. The RRT algorithm leverages a novel data structure for performing nearest neighbour comparisons for Ackermann-steering vehicles; termed the Distmetree. The resulting pushing states are searched using greedy heuristic search to find a solution and the final path is smoothed with cubic Bezier curves. The mode of operation chosen for best performance also constructs bidirectional RRTs to reach difficult to access pushing poses. The final mode of the algorithm was tested in simulation and proven to be able to solve a wide variety of maps in a few minutes while obeying bulldozer kinematic constraints. The algorithm, whilst not optimal, is complete which is the more desirable property in industry, and the solutions it produces are both feasible and reasonable.

Path Planning for Bulldozers with Curvature Constraints

There is presently no solution to the problem of an autonomous bulldozer pushing mounds of material to desired goal locations in the presence of obstacles whilst obeying the kinematic constraints of the bulldozer. Past work has solved some aspects of this problem, but not all. This research presents the first complete, practical solution to the problem. It works by creating a fixed RRT in advance, and then during operation connecting pushing poses into this RRT using Bezier curves. The RRT algorithm leverages a novel data structure for performing nearest neighbour comparisons for Ackermann-steering vehicles; termed the Distmetree. The resulting pushing states are searched using greedy heuristic search to find a solution and the final path is smoothed with cubic Bezier curves. The mode of operation chosen for best performance also constructs bidirectional RRTs to reach difficult to access pushing poses. The final mode of the algorithm was tested in simulation and proven to be able to solve a wide variety of maps in a few minutes while obeying bulldozer kinematic constraints. The algorithm, whilst not optimal, is complete which is the more desirable property in industry, and the solutions it produces are both feasible and reasonable.

DENSITY CHALLENGES OF HIGH-RISE RESIDENTIAL DEVELOPMENT IN MALAYSIA

Although the term ‘urbanisation’ was first coined in the 19th century, the phenomenon experienced a significant impact and received a lot of attention in the 21st century. One of the major results is density, causing effects such as excessive demand for residential buildings. To cope with the increasing urban population and limited land availability, cities can no longer opt for horizontal development strategies. Going vertical seems a practical solution, but it can lead to convoluted problems if it is not done with proper planning and mitigation measures at the preliminary stages of planning. This article describes the challenges of residential planning density for high-rise development in Malaysia using a systematic literature review on three identified real cases which separated by pre-development, post-development, and development control. The findings show the major challenges in pre-development and post development related to dissatisfaction with the increasing numbers of high-rise residences due to the increase in population and residential density. As a strategic development control, there must be uniformity in the act or law to control the development of this highrise residential. Thus, this article led to a better understanding of density related to high-rise residential development in Malaysia.

How to Retrieve Music using Mood Tags in a Folksonomy

A folksonomy is a classification system in which volunteers collaboratively create and manage tags to annotate and categorize content. The folksonomy has several problems in retrieving music using tags, including problems related to synonyms, different tagging levels, and neologisms. To solve the problem posed by synonyms, we introduced a mood vector with 12 possible moods, each represented by a numeric value, as an internal tag. This allows moods in music pieces and mood tags to be represented internally by numeric values, which can be used to retrieve music pieces. To determine the mood vector of a music piece, 12 regressors predicting the possibility of each mood based on acoustic features were built using Support Vector Regression. To map a tag to its mood vector, the relationship between moods in a piece of music and mood tags was investigated based on tagging data retrieved from Last.fm, a website that allows users to search for and stream music. To evaluate retrieval performance, music pieces on Last.fm annotated with at least one mood tag were used as a test set. When calculating precision and recall, music pieces annotated with synonyms of a given query tag were treated as relevant. These experiments on a real-world data set illustrate the utility of the internal tagging of music. Our approach offers a practical solution to the problem caused by synonyms.

GoZone: A Numerical Model for Travelling Fires Based on Cellular Automata Concept

Fires in large compartments tend to burn locally and to move across the floor over a period of time; this particular behaviour has been discovered to challenge the assumption of uniform gas temperature in the fire compartment. Recent studies on fires in large compartments have led to the now widely known concept of “travelling fires”. Several models have been proposed to describe the evolution in time of travelling fires. Although these models represented an innovative step in the field of travelling fires, the major drawbacks of these models can be found in the simplification of fire dynamics (constant spread rate, 1D imposed fire path) and limited field of application (rectangular based geometries). The purpose of this paper is to present a numerical model of travelling fire. The model was based on an improved zone model combined with a cellular automata model. The software GoZone, in which the model was implemented, is intended to be a practical solution to analyse fires in large compartments of potentially any shape. GoZone is aimed to describe the complex dynamics of the fire from ignition to a phase of growing localised fire that may eventually travel in the compartment, possibly followed by a flashover. The main sub models comprising GoZone are presented. A comparison is given with the results of under ventilated fire test 2 of the BST/FSR 1993 test series and with respect to the Veselì travelling fire test is shown. GoZone shows a promising capacity to represent fires in a large compartment in both air and fuel controlled fire conditions.