Cameron Christensen

The ViSUS Visualization Framework

19.

Fast Multiresolution Reads of Massive Simulation Datasets

Todays massively parallel simulation codes can produce output ranging up to many terabytes of data. Utilizing this data to support scientific inquiry requires analysis and visualization, yet the sheer size of the data makes it cumbersome or impossible to read without computational resources similar to the original simulation. We identify two broad classes of problems for reading data and present effective solutions for both. The first class of data reads depends on user requirements and available resources. Tasks such as visualization and user-guided analysis may be accomplished using only a subset of variables with a restricted spatial extent at a reduced resolution. The other class of reads requires full resolution multivariate data to be loaded, for example to restart a simulation. We show that utilizing the hierarchical multiresolution IDX data format enables scalable and efficient serial and parallel read access on a variety of hardware from supercomputers down to portable devices. We demonstrate interactive view-dependent visualization and analysis of massive scientific datasets using low-power commodity hardware, and we compare read performance with other parallel file formats for both full and partial resolution data.

Efficient I/O and storage of adaptive-resolution data

We present an efficient, flexible, adaptive-resolution I/O framework that is suitable for both uniform and Adaptive Mesh Refinement (AMR) simulations. In an AMR setting, current solutions typically represent each resolution level as an independent grid which often results in inefficient storage and performance. Our technique coalesces domain data into a unified, multiresolution representation with fast, spatially aggregated I/O. Furthermore, our framework easily extends to importance-driven storage of uniform grids, for example, by storing regions of interest at full resolution and nonessential regions at lower resolution for visualization or analysis. Our framework, which is an extension of the PIDX framework, achieves state of the art disk usage and I/O performance regardless of resolution of the data, regions of interest, and the number of processes that generated the data. We demonstrate the scalability and efficiency of our framework using the Uintah and S3D large-scale combustion codes on the Mira and Edison supercomputers.

TOD-Tree: Task-Overlapped Direct Send Tree Image Compositing for Hybrid MPI Parallelism and GPUs

Modern supercomputers have thousands of nodes, each with CPUs and/or GPUs capable of several teraflops. However, the network connecting these nodes is relatively slow, on the order of gigabits per second. For time-critical workloads such as interactive visualization, the bottleneck is no longer computation but communication. In this paper, we present an image compositing algorithm that works on both CPU-only and GPU-accelerated supercomputers and focuses on communication avoidance and overlapping communication with computation at the expense of evenly balancing the workload. The algorithm has three stages: a parallel direct send stage, followed by a tree compositing stage and a gather stage. We compare our algorithm with radix-k and binary-swap from the IceT library in a hybrid OpenMP/MPI setting on the Stampede and Edison supercomputers, show strong scaling results and explain how we generally achieve better performance than these two algorithms. We developed a GPU-based image compositing algorithm where we use CUDA kernels for computation and GPU Direct RDMA for inter-node GPU communication. We tested the algorithm on the Piz Daint GPU-accelerated supercomputer and show that we achieve performance on par with CPUs. Last, we introduce a workflow in which both rendering and compositing are done on the GPU.

Scalable visualization and interactive analysis using massive data streams

Historically, data creation and storage has always outpaced the infrastructure for its movement and utilization. This trend is increasing now more than ever, with the ever growing size of scientific simulations, increased resolution of sensors, and large mosaic images. Effective exploration of massive scientific models demands the combination of data management, analysis, and visualization techniques, working together in an interactive setting. The ViSUS application framework has been designed as an environment that allows the interactive exploration and analysis of massive scientific models in a cache-oblivious, hardware-agnostic manner, enabling processing and visualization of possibly geographically distributed data using many kinds of devices and platforms. For general purpose feature segmentation and exploration we discuss a new paradigm based on topological analysis. This approach enables the extraction of summaries of features present in the data through abstract models that are orders of magnitude smaller than the raw data, providing enough information to support general queries and perform a wide range of analyses without access to the original data.

Serial section registration of axonal confocal microscopy datasets for long-range neural circuit reconstruction

In the context of long-range digital neural circuit reconstruction, this paper investigates an approach for registering axons across histological serial sections. Tracing distinctly labeled axons over large distances allows neuroscientists to study very explicit relationships between the brains complex interconnects and, for example, diseases or aberrant development. Large scale histological analysis requires, however, that the tissue be cut into sections. In immunohistochemical studies thin sections are easily distorted due to the cutting, preparation, and slide mounting processes. In this work we target the registration of thin serial sections containing axons. Sections are first traced to extract axon centerlines, and these traces are used to define registration landmarks where they intersect section boundaries. The trace data also provides distinguishing information regarding an axons size and orientation within a section. We propose the use of these features when pairing axons across sections in addition to utilizing the spatial relationships among the landmarks. The global rotation and translation of an unregistered section are accounted for using a random sample consensus (RANSAC) based technique. An iterative nonrigid refinement process using B-spline warping is then used to reconnect axons and produce the sought after connectivity information.

Trace driven registration of neuron confocal microscopy stacks

Active research in the area of 3-D neurite tracing has predominantly focused on single sections. Ultimately, however, neurobiologists desire to study the long range connectivity of the brain, which requires tracing axons across multiple serially-cut sections. Registration of axonal sections is challenging due to several factors, such as sparseness of the axons and complications of the sectioning process, including tissue deformation and loss. This paper investigates a method for registering sections using centerline traces which provide the locations of axons at section boundaries and the angles at which the axons approach the boundaries. This information is used to determine correspondences between two serial sections. Both global and local differences are accounted for using rigid and non-rigid transforms. Results show that utilizing information from traced axons allows axon continuity across sections to be restored.

Topology-aware image compositing using NVLink

Compositing is a significant factor in distributed visualization performance at scale on high-performance computing (HPC) systems. For applications such as Para VieworVisIt, the common approach is “sort-last” rendering. For this approach, data are split up to be rendered such that each MPI rank has one or more portions of the over-all domain. After rendering its individual piece(s), each rank has one or more partial images that must be composited with the others to form the final image. The common approach for this step is to use a tree-like communication pattern to reduce the rendered images down to a single image to be displayed to the user. A variety of algorithms have been explored to perform this step efficiently in order to achieve interactive rendering on massive systems [7, 3, 8, 4].

Embedded domain-specific language and runtime system for progressive spatiotemporal data analysis and visualization

As our ability to generate large and complex datasets grows, accessing and processing these massive data collections is increasingly the primary bottleneck in scientific analysis. Challenges include retrieving, converting, resampling, and combining remote and often disparately located data ensembles with only limited support from existing tools. In particular, existing solutions rely predominantly on extensive data transfers or large-scale remote computing resources, both of which are inherently offline processes with long delays and substantial repercussions for any mistakes. Such workflows severely limit the flexible exploration and rapid evaluation of new hypotheses that are crucial to the scientific process and thereby impede scientific discovery. Here we present an embedded domain-specific language (EDSL) specifically designed for the interactive exploration of large-scale, remote data. Our EDSL allows users to express a wide range of data analysis operations in a simple and abstract manner. The underlying runtime system transparently resolves issues such as remote data access and resampling while at the same time maintaining interactivity through progressive and interruptible computation. This system enables, for the first time, interactive remote exploration of massive datasets such as the 7km NASA GEOS-5 Nature Run simulation, which previously have been analyzed only offline or at reduced resolution.

TOD-tree: task-overlapped direct send tree image compositing for hybrid MPI parallelism

Modern supercomputers have very powerful multi-core CPUs. The programming model on these supercomputer is switching from pure MPI to MPI for inter-node communication, and shared memory and threads for intra-node communication. Consequently the bottleneck in most systems is no longer computation but communication between nodes. In this paper, we present a new compositing algorithm for hybrid MPI parallelism that focuses on communication avoidance and overlapping communication with computation at the expense of evenly balancing the workload. The algorithm has three stages: a direct send stage where nodes are arranged in groups and exchange regions of an image, followed by a tree compositing stage and a gather stage. We compare our algorithm with radix-k and binary-swap from the IceT library in a hybrid OpenMP/MPI setting, show strong scaling results and explain how we generally achieve better performance than these two algorithms.