Search for: All records

Octo-Tiger is a code for modeling three-dimensional self-gravitating astrophysical fluids. It was particularly designed for the study of dynamical mass transfer between interacting binary stars. Octo-Tiger is parallelized for distributed systems using the asynchronous many-task runtime system, the C++ standard library for parallelism and concurrency (HPX) and utilizes CUDA for its gravity solver. Recently, we have remodeled Octo-Tiger’s hydro solver to use a three-dimensional reconstruction scheme. In addition, we have ported the hydro solver to GPU using CUDA kernels. We present scaling results for the new hydro kernels on ORNL’s Summit machine using a Sedov-Taylor blast wave problem. We also compare Octo-Tiger’s new hydro scheme with its old hydro scheme, using a rotating star as a test problem. 

We describe JetLag, a Python-based environment that provides access to a distributed, interactive, asynchronous many-task (AMT) computing framework called Phylanx. This environment encompasses the entire computing process, from a Jupyter front-end for managing code and results to the collection and visualization of performance data.We use a Python decorator to access the abstract syntax tree of Python functions and transpile them into a set of C++ data structures which are then executed by the HPX runtime. The environment includes services for sending functions and their arguments to run as jobs on remote resources.A set of Docker and Singularity containers are used to simplify the setup of the JetLag environment. The JetLag system is suitable for a variety of array computational tasks, including machine learning and exploratory data analysis. 

We describe an interactive computing environment called JetLag. JetLag implements the following features of Phylanx project: (1) Phylanx, a Python-based asynchronous array computing toolkit; (2) the APEX performance measurement library; (3) a performance visualization framework called Traveler; (4) the Tapis/Agave Science as a Service middleware; and (6) a container infrastructure that includes Docker-based Jupyter notebook for the client and a singularity image for the server. The running system starts with a user performing array computations on their workstation or laptop. If, at some point, the calculation the user is performing becomes sufficiently intensive or numerous, it can be packaged and sent to another machine where it will run (through the batch queue system if there is one), produce a result, and have that result sent back to the user’s local interface. Whether the calculation is local or remote, the user will be able to use APEX and Traveler to diagnose and fix performance related problems. The JetLag system is suitable for a variety of array computational tasks, including machine learning and exploratory data analysis. 

As the era of high frequency, single core processors have come to a close, the new paradigm of many core processors has come to dominate. In response to these systems, asynchronous multitasking runtime systems have been developed as a promising solution to efficiently utilize these newly available hardware. Asynchronous multitasking runtime systems work by dividing a problem into a large number of fine grained tasks. However, as the number of tasks created increase, the overheads associated with task creation and management cannot be ignored. Task inlining, a method where the parent thread consumes a child thread, enables the runtime system to achieve the balance between parallelism and its overhead. As largely impacted by different processor architectures, the decision of task inlining is dynamic in nature. In this research, we present adaptive techniques for deciding, at runtime, whether a particular task should be inlined or not. We present two policies, a baseline policy that makes inlining decision based on a fixed threshold and an adaptive policy which decides the threshold dynamically at runtime. We also evaluate and justify the performance of these policies on different processor architectures. To the best of our knowledge, this is the first study of the impacts of adaptive policy at runtime for task inlining in an asynchronous multitasking runtime system on different processor architectures. From experimentation, we find that the baseline policy improves the execution time from 7.61% to 54.09%. Furthermore, the adaptive policy improves over the baseline policy by up to 74%. 

Despite advancements in the areas of parallel and distributed computing, the complexity of programming on High Performance Computing (HPC) resources has deterred many domain experts, especially in the areas of machine learning and artificial intelligence (AI), from utilizing performance benefits of such systems. Researchers and scientists favor high-productivity languages to avoid the inconvenience of programming in low-level languages and costs of acquiring the necessary skills required for programming at this level. In recent years, Python, with the support of linear algebra libraries like NumPy, has gained popularity despite facing limitations which prevent this code from distributed runs. Here we present a solution which maintains both high level programming abstractions as well as parallel and distributed efficiency. Phylanx, is an asynchronous array processing toolkit which transforms Python and NumPy operations into code which can be executed in parallel on HPC resources by mapping Python and NumPy functions and variables into a dependency tree executed by HPX, a general purpose, parallel, task-based runtime system written in C++. Phylanx additionally provides introspection and visualization capabilities for debugging and performance analysis. We have tested the foundations of our approach by comparing our implementation of widely used machine learning algorithms to accepted NumPy standards. 

Adaptations to an organism's environment often involve sensory system modifications. In this study, we address how evolutionary divergence in sensory perception relates to the physiological coding of stimuli. Mormyrid fishes that can detect subtle variations in electric communication signals encode signal waveform into spike-timing differences between sensory receptors. In contrast, the receptors of species insensitive to waveform variation produce spontaneously oscillating potentials. We found that oscillating receptors respond to electric pulses by resetting their phase, resulting in transient synchrony among receptors that encodes signal timing and location, but not waveform. These receptors were most sensitive to frequencies found only in the collective signals of groups of conspecifics, and this was correlated with increased behavioral responses to these frequencies. Thus, different perceptual capabilities correspond to different receptor physiologies. We hypothesize that these divergent mechanisms represent adaptations for different social environments. Our findings provide the first evidence for sensory coding through oscillatory synchrony.