Video encoding has long been a cornerstone of streaming technology. For years, engineers have relied on CPUs, GPUs, and ASICs to power video workflows, each presenting unique advantages and trade-offs. However, as the demand for ultra-efficient, scalable, and cost-effective encoding grows, a new category of hardware is emerging: Video Processing Units (VPUs).
VPUs Are Changing the Game for Video Engineers
Easy Tools, a company specializing in modular video processing, utilizes NETINT’s VPU technology to enhance the efficiency and reduce the cost of video delivery for broadcasters. This article examines the practical implications of this shift, highlighting cost-effective encoding strategies, real-world deployment considerations, and the increasing role of software-defined solutions in unlocking the full potential of VPUs. If you are a video engineer seeking to optimize your streaming infrastructure, this guide offers essential insights into how VPUs can reduce costs, enhance efficiency, and improve video quality at scale.
“NETINT made a very clever choice: the NVMe form factor means on a regular one-rack-unit server you can put eight or ten modules,so that’s a game-changer.
, Christophe Massiot, CEO, Easy Tools
The Evolution of Video Encoding: From CPUs to VPUs
Video encoding has evolved significantly over the years, with different hardware solutions catering to various use cases. Historically, software-based CPU encoding has been widely used due to its flexibility, but it struggles with efficiency at scale. High power consumption and processing limitations make it difficult to sustain large-scale encoding operations.
Meanwhile, GPUs have offered better parallel processing capabilities, but they remain expensive and power-hungry, making them a less viable option for large-scale streaming applications. On the other end of the spectrum, ASICs (Application-Specific Integrated Circuits) provide high efficiency, but their rigidity makes them difficult to update and adapt to changing video standards.
This is where VPUs come in. Unlike general-purpose CPUs or power-hungry GPUs, VPUs are designed specifically for video compression, offering a unique balance between performance and efficiency. They significantly reduce hardware costs and energy consumption while providing the scalability that modern video applications demand.
From VLC to Broadcast Innovation: Christophe Massiot on Easy Tools & VPUs
– Voices of Video with Christophe Massiot from Easy Tools
Why Easy Tools Chose NETINT VPUs for Broadcast Encoding
Easy Tools, a leader in modular video processing solutions, faced a major challenge: scaling live broadcast encoding efficiently while keeping infrastructure costs low. Their encoding needs were broad, including the simultaneous processing of over 100 channels, each requiring multiple renditions for OTT streaming. Meeting strict compliance standards for transport streams while minimizing power consumption and hardware costs was a top priority.
After evaluating various encoding solutions, NETINT’s VPUs emerged as the optimal choice. With VPUs, Easy Tools replaced 10 CPU-based servers with a single VPU-powered server, the Quadra Video Server, resulting in a drastic reduction in power consumption and infrastructure expenses. The NVMe-based form factor of Quadra VPUs allows 10 to be fitted in a 1RU server, something that would be impossible with GPUs due to space and power constraints.
A significant advantage of NETINT VPUs is their ability to receive firmware updates, enabling adaptability over time. Unlike traditional hardware-based encoders, which are typically fixed-function and unmodifiable, VPUs offer a degree of flexibility that ensures future-proofing for emerging codecs and evolving video workflows.
Overcoming the Biggest Hurdle: Broadcast-Grade Compliance
For broadcasters, encoding quality and efficiency are only part of the equation. The real challenge lies in ensuring full compliance with transport stream (TS) specifications while maintaining a constant bitrate (CBR). Many open-source encoding tools, such as FFmpeg and VLC, while highly versatile, struggle with strict MPEG-TS compliance. This often leads to unpredictable behavior when integrating with broadcast infrastructures.
Easy Tools solved this issue by integrating NETINT VPUs with UPipe, their modular, open-source framework. This integration allowed them to achieve:
- Perfect compliance with MPEG-TS standards, ensuring smooth transmission across broadcast networks.
- Seamless integration with DVB, satellite, and SRT workflows, providing full compatibility with professional broadcasting pipelines.
- Automatic detection and management of new audio and subtitle components, allowing dynamic updates to live streams without disruption.
“You will be able to expand into use cases you’ve never dreamed of – for instance, we can use your NETINT module to build a multi-viewer, offloading hardware decoding and making a bigger multi-viewer thanks to that.”
, Christophe Massiot, CEO, Easy Tools
Multi-View Powered by VPUs
VPUs are not just for encoding. One of the most surprising discoveries from Easy Tools’ experience with NETINT VPUs is their ability to optimize multi-viewer applications. Multi-viewer systems that allow broadcasters to monitor multiple live video feeds are generally built using expensive, dedicated hardware. Easy Tools found that VPUs could be repurposed for real-time multi-viewing by offloading video decoding tasks to VPUs enable:
- Reduced CPU usage, freeing up system resources for other critical tasks.
- An increase in the number of live video feeds that can be processed simultaneously, enabling broader monitoring capabilities.
- Lowered costs by eliminating the need for additional decoding hardware, making multi-viewer setups more affordable.
“NETINT made a very clever choice: the NVMe form factor means on a regular one-rack-unit server you can put eight or ten modules,so that’s a game-changer.”
, Christophe Massiot, CEO, Easy Tools
Should You Ditch CPUs and GPUs for VPUs?
With all the advantages that VPUs bring, should video engineers completely replace CPUs and GPUs in their encoding workflows? The answer depends on the specific requirements of each deployment.
VPUs are particularly well-suited for high-volume encoding, such as live streaming services, OTT platforms, and large-scale broadcasters. They significantly reduce power consumption while maintaining high-performance encoding at scale. Additionally, for next-generation codecs like AV1 and HEVC, VPUs offer a far more efficient encoding approach compared to software-based alternatives.
However, for applications that involve only a small number of video streams or require extreme flexibility, CPU encoding may still be sufficient. The key takeaway is that VPUs do not need to replace CPUs and GPUs entirely; instead, they should be considered the best option for large-scale, power-efficient video processing.
As the video industry moves toward more scalable, efficient, and cost-effective solutions, VPUs are poised to play a leading role. They offer an unparalleled combination of efficiency, flexibility, and compliance, making them a valuable tool for engineers seeking to optimize video workflows.
“The GPUs have a limited number of streams they can do concurrently, and they just suck up a lot of electricity – your regular GPU is 300 watts or 500 watts or something like that.”
, Christophe Massiot, CEO, Easy Tools
Schedule a meeting to learn how NETINT VPUs can enhance live streaming with energy-efficient, scalable solutions.
