1.1. Introduction
High-performance computing relies on ever finer threading. Advances in processor technology include ever greater numbers of cores, hyperthreading, accelerators with integrated blocks of cores, and special vectorized instructions, all of which require more software parallelism to achieve peak performance. Traditional visualization solutions cannot support this extreme level of concurrency. Extreme scale systems require a new programming model and a fundamental change in how we design algorithms. To address these issues we created Viskores: the visualization toolkit for multi-/many-core architectures.
Viskores supports a number of algorithms and the ability to design further algorithms through a top-down design with an emphasis on extreme parallelism. Viskores also provides support for finding and building links across topologies, making it possible to perform operations that determine manifold surfaces, interpolate generated values, and find adjacencies. Although Viskores provides a simplified high-level interface for programming, its template-based code removes the overhead of abstraction.
CUDA SDK |
Viskores |
431 LOC |
265 LOC |
|
|
Viskores simplifies the development of parallel scientific visualization algorithms by providing a framework of supporting functionality that allows developers to focus on visualization operations. Consider the listings in Table 1.1 that compares the size of the implementation for the Marching Cubes algorithm in Viskores with the equivalent reference implementation in the CUDA software development kit. Because Viskores internally manages the parallel distribution of work and data, the Viskores implementation is shorter and easier to maintain. Additionally, Viskores provides data abstractions not provided by other libraries that make code written in Viskores more versatile.
1.1.1. How to Use This Guide
This user’s guide is organized into 5 parts to help guide novice to advanced users and to provide a convenient reference. Part 1 (Getting Started) provides a brief overview of using Viskores. This part provides instructions on building Viskores and some simple examples of using Viskores. Users new to Viskores are well served to read through Part Part 1 (Getting Started) first to become acquainted with the basic concepts.
The remaining parts, which provide detailed documentation of increasing complexity, have chapters that do not need to be read in detail. Readers will likely find it useful to skip to specific topics of interest.
Part 2 (Using Viskores) dives deeper into the Viskores library. It provides much more detail on the concepts introduced in Part 1 (Getting Started) and introduces new topics helpful to people who use Viskores’s existing algorithms.
Part 3 (Developing Algorithms) documents how to use Viskores’s framework to develop new or custom visualization algorithms. In this part we dive into the inner workings of filters and introduce the concept of a worklet, which is the base unit used to write a device-portable algorithm in Viskores. Part 3 (Developing Algorithms) also documents many supporting functions that are helpful in implementing visualization algorithms.
Part 4 (Advanced Development) explores in more detail how Viskores manages memory and devices. This information describes how to adapt Viskores to custom data structures and new devices.
Part 5 (Core Development) exposes the inner workings of Viskores. These concepts allow you to design new algorithmic structures not already available in Viskores.
Did You Know?
In this guide we periodically use these Did you know? boxes to provide additional information related to the topic at hand.
Common Errors
Common Errors blocks are used to highlight some of the common problems or complications you might encounter when dealing with the topic of discussion.