Skip to content
The Power of VPUs Compared to CPUs

The Power of VPUs Compared to CPUs

Video streaming has evolved dramatically over the past few decades, and with it, technologies that have enabled high-quality, real-time video encoding and delivery. One of the most significant shifts in recent years is the adoption of Application-Specific Integrated Circuits (ASICs) for high-volume transcoding. The question is, WHY? How have ASICs and the VPUs they power become the preferred architecture for high-volume video streaming and broadcasting over traditional CPU and GPU-based encoding solutions?

The Evolution of Video Encoding: From CPUs to ASICs

Brief History of Video Encoding Technologies

Historically, video encoding has gone through several phases. In the early days of digital video, MPEG-2 was the standard for satellite and cable TV, and ASICs were the primary technology used for encoding. This was because CPUs at the time lacked the power to handle MPEG-2 encoding in real time.

From 2012 to 2020, the industry, seduced by the public cloud OPEX business model, shifted towards cloud-based video processing and encoding. CPUs had become powerful enough to manage real-time encoding and transcoding, significantly decreasing the need for dedicated ASICs. During this period, companies like Elemental, later acquired by AWS, pioneered software-defined video processing, further pushing the trend toward CPU-based solutions.

However, as video codecs became more complex and the demand for high-volume, real-time interactive video services grew, CPU and GPU-based encoding limitations became apparent. This led to a resurgence in the use of ASICs for video encoding, marking the beginning of a new phase in the evolution of video encoding technologies.

The Modern Era: ASICs for High-Volume Video Streaming

In recent years, ASICs have become increasingly prevalent in high-volume interactive video applications such as cloud gaming, live streaming, and user-generated content platforms. Companies like Google and Meta have embraced ASICs for their encoding workflows, signaling a broader industry shift towards this technology.

ASICs offer significant advantages over CPUs and GPUs regarding power consumption, processing density, and cost efficiency. As codecs evolve from lower complexity to higher, H.264 to HEVC to AV1, and beyond, the need for specialized hardware to handle the increased complexity and processing demands is critical.

Leveraging ASICs for Video Encoding

Maximizing Efficiency with ASICs

ASICs are designed for specific tasks, making them highly efficient compared to general-purpose CPUs and GPUs. This efficiency translates to lower power consumption and reduced heat generation, critical factors in data center operations. Google’s Argos ASICs have demonstrated a 20-33x improvement in compute efficiency compared to traditional software on servers, and analysts suggest it’s possible that YouTube eliminated as many as 10 million CPUs using ASICs. Many believe that this massive reduction in server infrastructure cost drives YouTube’s dominance over Twitch and other UGC video-sharing platforms.

Tip: When planning your encoding infrastructure, consider your long-term operational costs and whether they can be sustained at 1 million daily active users, 10 million, or beyond. ASICs can significantly save power and cooling expenses, making them the only cost-reduction solution for high-volume video streaming.

Enhancing Density and Scalability

One key benefit of ASICs is their ability to handle a high number of streams per server, significantly increasing processing density. NETINT’s Quadra Video Server can encode 320 simultaneous 1080p30 live video streams, providing unparalleled scalability for streaming and interactive video services. Resolutions beyond 1080p to 8K are possible using NETINT VPUs and AV1, HEVC, or H.264.

Tip: Evaluate your current and future streaming needs. ASICs can help you scale efficiently without requiring extensive infrastructure investments, allowing you to meet growing demand seamlessly.

Achieving Superior Quality and Performance

ASICs are optimized for video encoding tasks, ensuring high-quality output with minimal latency, just 8ms on Quadra series VPUs. Quadra delivers near-parity with software-based transcoding quality while offering added speed and efficiency benefits. 

Tip: Prioritize quality and performance in your encoding workflows. ASICs can help you achieve the best possible video quality while maintaining low latency, which is crucial for live and interactive video applications.

Ensuring Reliability and Future-Proofing

ASICs provide a reliable and future-proof solution for video encoding. Unlike CPUs and GPUs, which are designed for a wide range of tasks, ASICs are built specifically for encoding, making them more robust and less prone to performance issues. Additionally, ASICs can be customized and optimized for specific use cases, offering greater flexibility and adaptability.

Tip: Consider the long-term reliability of your encoding infrastructure. ASICs offer a stable and future-proof solution that can adapt to evolving video standards and requirements, ensuring your workflows remain efficient and effective.

The New Standard in Video Streaming

Adopting ASICs for high-volume video transcoding is unavoidable in order to remain competitive in the video streaming market. The question is no longer IF but WHEN. ASICs set a new standard for video encoding technologies with superior efficiency, scalability, quality, and reliability. Over time, the loss caused by not integrating ASICs will translate into billions of dollars lost in market share regained by the competitors who implement ASIC encoders.

For more insights and to explore how NETINT’s ASIC-based transcoders can benefit your organization, consider scheduling a meeting with NETINT experts. They can provide detailed guidance and support to help you leverage the full potential of ASICs in your video streaming workflows.

Picture of Mark Donnigan

Mark Donnigan

is a veteran of the video ecosystem, working with disruptive innovation companies like NETINT to increase video codec standards and streaming video technology adoption. In addition to working at the forefront of building one of the world's first T-VOD services and driving early HEVC and AV1 adoption, Mark contributed actively to the development and growth of the digital locker initiative, Ultraviolet, breaking device-based content walled gardens, allowing consumers to enjoy video on any device, any time, and in any location. As a technologist and ecosystem developer, Mark's work building cloud-deployed and hyper-scale WebRTC, live, metaverse, and cloud gaming applications gives him a unique view of the OTT and video streaming landscape.


ACCESS NOW:  ASIC-Based Transcoding
for High-volume Use Cases
Including social media, broadcast, interactive platforms, and service providers