Beyond the GPU: Offloading Vision Workloads to PVA on NVIDIA Jetson
24
4:00 PM - 5:00 PM
Join RidgeRun experts for a free webinar on how the Programmable Vision Accelerator (PVA) can help free GPU resources on NVIDIA Jetson. We’ll explore how suitable vision and sensor-processing workloads can be moved from the GPU to the PVA, creating more headroom for AI inference and helping increase throughput without immediately moving to a higher-end Jetson platform.
What you'll learn:
- Understanding the architecture: Memory organization, processing model, and how it fits into NVIDIA Jetson
- Development and tooling: Using the SDK, profiling, debugging, and performance analysis
- When to use it: Advantages, limitations, trade-offs, and how it compares with other compute options
- Interoperability and resource sharing: Integrating workloads alongside the GPU, CPU, and other accelerators
- Real-world applications: Camera, radar, and sensor-processing use cases
- Performance: Practical results and what you can achieve with the hardware accelerator
- Final Remarks & Q&A
Speaker
Marco Herrera
Embedded SW Team Lead at RidgeRun
Marco Herrera Valverde is an Embedded Software Team Lead at RidgeRun with over five years of experience developing solutions for NVIDIA Jetson platforms. He has successfully led real-world projects across a range of Jetson platforms, with expertise in embedded Linux, computer vision, performance optimization, and multimedia systems. He currently leads RidgeRun’s internal efforts around NVIDIA’s PVA. His work focuses on optimizing PVA workloads and enabling high-performance computer vision applications on embedded platforms.
Replays
See all
24
4:00 PM - 5:00 PM