Step-by-Step Tutorials

The Programmable Vision Accelerator (PVA) is a power efficient VLIW processor that is specialized for image processing and computer vision algorithms.

Key places where PVA’s capabilities are a good match are algorithmic domains where we need to have predictable processing, at low power and low latency. PVA shines on semi-dense or dense regular computation, even on small data sets, which need predictable run-times.

The PVA SDK presents a unified programming model for PVA. The cuPVA runtime is the set of runtime APIs distributed with the PVA SDK.

The tutorials are arranged into series, each one targeting different aspects of PVA programming. Users should work through each tutorial within a series in order. The series themselves are also presented in an order which gradually introduces new concepts. It is recommended that new users first work through the PVA Basics series before moving onto other tutorials.

• PVA Basics

• Processing Tiles in Parallel

• Developing, Optimizing and Profiling a Complete PVA Program

• Using Developer Tools for Debugging PVA Applications

• Advanced PVA Optimization Techniques

• Interacting with Other Engines

• PVA Accelerated Primitives Library (APL)